of pupils at the lowest - ed
TRANSCRIPT
DOCUMENT RESUME
24 AA 000 318
Survey and Analyses of Results from Title I Funding for Compensatory Education. Final Report.
General Electric Co., Washington, DC. TEMPO .Spons Agency -Office of Education (DHEW), Washington, DC . Bureau of Resear ch.
Report No -FR -67TMP -115Bureau No -BR -7 -0979Pub Date 1 Mar 68Contract -OEC -05 -67 -55Note-227p.EDRS Price MF -$100 I-IC -$1145Descriptors -Academic Performance, *Achievement Rating, Achievement Tests, Community Characteristics,
*Comparative Analysis, *Compensatory Education Programs, Data Analysis, Economically Disadvantaged,
Educational Finance, Educational:), Disadvantaged, Elementary Grades, Environmental Influences, *Federal Aid,
*Program Evaluation, Secondary Grades, Student Characteristics, SurveysIdentifiers -*ESEA Title I Programs
The aim of the study was to provide the Department of Health, Education, andWelfare with evidence as to the productivity of compensatory education (CE) programs
for disadvantaged children, particularly the effects of Title I of the Elementary and
Secondary Act of 1965 during its first year and a half. Data were coHected on pupilperformance and exposure to CE in 11 school districts (132 schools): in addition toachievement test scores for 1965-66 and 1966-67, information was gathered on thecharacteristics of the pupils, their schools, and their communities. Results indicated:(1)
a slight decline in average pupil achievement level in the sample schools. (2) a slightimprovement in achievement of pupils at the lowest achievement levels in their
respective grades; and (3) considerable variation in changes in achievement among
school districts. Preliminary results suggested that the amount of improvement was
related to level of Title I expenditures. The overall study provided evidence that morespecific studies were needed to properly evaluate the effects of Title I. Appendicescontain technical discussions as well as supporting material for the main text.(JAM)
-
U.S. DEPARTMENT OF HEALTH, EDUCATION 8 WELFARE
OFFICE OF EDUCATION
THIS DOCUMENT HAS BEEN REPRODUCED EXACTLY AS RECEIVED FROM THE
PERSON tlt ORGANIZATION ORIGINATING IT. POINTS OF VIEW OR OPINIONS
STATED DO NOT NECESSARILY REPRESENT OFFICIAL OFFICE OF EDUCATION
POSITION OR POLICY.
SURVEY AND ANALYSES OF RESULTS FROMTITLE I FUNDING FOR COMPENSATORY EDUCATION
FINAL REPORT
67TMP-115
1 March 1968
Submitted ToProject Officers
Department Health, Education and WelfareWashington, D.C. 20211
TEMPOGENERAL ELECTRIC COMPANY
WASHINGTON OPERATION
Copy
iii
ININIPPA Al.r'rK crm...c
This report summarizes the work done to date by General Electric,TEMPO on Contract No. HEW-05-67-55, A Survey and PreliminaryCost-Benefit Analysis in Elementary-Secondary Education, Phase I.The study, which began 21 April 1967, was conducted as a joint effortwith the Department of Health, Education and Welfare (DHEW) and
the Office of Education (OE). Although this report is submitted as afinal report on the above contract, it should be viewed as a progressreport on the overall study, and should be added to the body of infor-mation which was reported earlier.
Two interim progress reports preceded the present report:
67TMP-67 (submitted 27 June, revised 21 July 1967)
67TMP-89 (submitted 15 September 1967)
The June report describes data collection formats and discussesproblems in data collection. The proposed analytical approach ispresented in detail.
The September report summarizes the fourteen field trips con-ducted in the study and provides descriptive information on those com-pensatory education (CE) programs. The trip reports provide descrip-tions of the information and impressions obtained from sits to four-teen school districts.
The trip reports are reprinted as a separate appendix to this report.
SYNOPSIS OF WORK
The first several weeks of the study were devoted to designing ana-lytical approaches and planning a survey of CE programs in selectedschool districts. Models for testing the statistical significance of
observed changes in selected pupil performance measures were de-veloped and presented in 67TMP-67, the June progress report.
:
iv67TMP-115
Plans ,`.c obtain data on pupil performance and exposure to CE were
expanded to include information on the characteristics of the pupils,
their schools and their communities. The large variation in these
later variables was jiidged as important for explaining observed
changes in pupil performance between 1965-66 and 1966-67.
Following guidelines of the Office of Education (as contained in
their financial and pupil accounting handbooks), special forms weredesigned to obtain data in a standardized fashion to conserve the time
of school personnel. (The forms are displayed in 67TMP-67.)
The fourteen school districts included in the study were selected
by DHEW but were not intended to be representative of all school dis-
tricts pursuing compensatory education programs. Rather, for most
of the districts selected, there was reason to believe that successful
programs were in progress in at least some of the schools. One ob-
jective of the study wa i. to investigate the characteristics of programs
which held promise of favorable impact on the performance of de-
prived children.
The field trips were made by joint tearis of DHEW, OE and TEMPO
personnel and entailed 3-day to 2-week visits to each district. Un-
doubtedly more time would have been helpful in every case but the
need to visit many locations in order to expand the sample sizes of
schools and programs was considered more important. One negative
effect of the intense travel schedule was the inability to summarize
and analyze adequately the information and impressions after each
trip.
Problems encountered in collecting information provide insight
into the complexity of such a study and are, therefore, useful. back-
ground information. Briefiy stated, the more severe problems were:
amount of detailed information required for making per-
formance measures compatible is very great and the number
of special conversion factors that are required prevents
extensive use of automatic data processing;
achievement scores came from different tests and came
from tests which were administered at different points during
the school year among and within the sample districts;
much of the readily available data on CE were not in usable
form for study because they did not give Information for spe-
cific grades;
PREFACE
it was difficult to distinguish between CE and regular schoolprograms and between CE programs funded by federal andnon-federal agencies;
the large transfer of pupils into a sample school makes itdifficult to identify the amount of CE to which pupils havebeen exposed.
Achievement data received from the school were reviewed for ap-plicability in comparing 1965-66 with 1966-67 performance levels.Because of incompatibility and incompleteness of data, it was neces-sary to delete three of the fourteen school districts in the statisticalanalysis. The trip reports include information of fourteen school dis-tricts but the rest of the report is based on information from elevenschool districts.
SYNOPSIS OF THIS REPORT
This final report consists of four sections, and seven attached ap-pendices. An appendix of trip reports, prepared for use within DHEWand OE, has been bound separately to maintain the school-districtanonymity requested for this report.
Section 1 describes the original study objectives and the. technicalapproach used in analyzing sample data. It gives a brief summaryof the final selection of observations for statistical analysis.
Section 2 presents the results of an analysis which changes inachievement test scores and the association of ch,-_nges with the spe-cific state varirbles are examined. The variation in achievementtest scores among school districts is described and a summary ofthe procedures used in processing achievement data is presented.
Section 3 analyzes the correlation between changes in achievementand Title I expenditures measured at the district level. It also il-lustrates, for two sample school districts, procedures used and prob-lems inherent in determining the type and level of CE at the gradelevel in specific schools. It provides insight into variations in thetype, purpose, duration and intensity of CE programs, and in thetype of students involved in them, and the relationship of these pro-grams to regular education programs.
Section 4 presents observations, conclusions and recommendationsbased on the Phase I effort.
vi 67TMP-115
The attached appendices contain technical discussions as well as
supporting material for the discussions in the main text.
vii
SUMMARY
This report analyzes data from a sample of 132 schools whichreceived funds from Title I for compensatory education to aid educa-tionally disadvantagee. pupils. Most of the eleven school districts fromwhich the schools were drawn were selected because there was rea-son to believe tint successful compensatory education programs werein progress in at least some of the district schools. Conclusions arebased on a comparison of achievement scores in 1966-67, after pupilswere exposed to compensatory education from Title I funds, withachievement scores in 1965-66.
There appears to have been a slight decline in average pupil achieve-ment level in the sample schools. For the entire sample the averagegrade equivalence score in 1966-67 was approximately one-half monthlower than the corresponding grade equivalence score in 1965-66.The percentage of the pupils in the 1966-67 test results who also par-ticipated in programs funded by Title I is not known but is believedto be less than 50 percent.
On the other hand, there appez...._ a to have been a slight improve-ment in achievement of pupils who are at the lowest achievement levelsin their respective grades. The average grade equivalence score of
pupils at the lowest decile in the 1966-67 tests was -Tproximately one-fourth month higher than the average grade equivalence score of cor-responding pupils in the 1965-66 tests. Although the one-fourth monthchange is very small, it is statistically different from the observednegative changes in both the mean score and the score at the upperquartile.
There is considerable variation in changes in achievement amongschool districts. One district shows a statistically significant increasein the average score while two show significant declines. With re-spect to achievement at the lowest decile, none of the school districtsshows significant decrease, but two districts show significant increases.
viii 67TMP-115
The very preliminary results suggest that amount of improvementis related to level of Title I expenditures. The districts which showedthe largest improvement at the lowest decile are the districts whichhad the higher average Title 1 expenditures per pupil.
The two variables most closely related to changes in achievementare initial achievement level and percent Negro. Lower initialachievement levels in 1965-66 are associated with iarger gains be-tween 1965-66 and 1966-67. This suggests that the availability ofTitle I funds is probably helping pupils at the lowest achievementlevels the most. Schools which had 40 to 60 percent Negro pupilsshowed the poorest response to compensatory education programs.Schools with 0 to 20 percent Negro pupils showed the best response.
Examinations of schools in two districts reveal extremely widevariationo among schools and among grades within a school in ex-penditures for both regular school programs and compensatory edu-cation programs. in one of the districts there was a positive cor-relation between changes in achievement and total expenditures butin the other district no significant relationship could be detected.
The overall study provides considerable evidence that more spe-cific studies are needed to properly evaluate the effects from TitleI. In addition, more emphasis should be placed on getting participat-ing schools to keep systematic records on pupil, school, and programcharacteristics. The records from many schools are not adequatefor the types of analysis required for proper evaluation of compensa-tory education.
It is always possible that the positive changes which have been at-tributed to CE are due to sampling variation. However, a must berecognized that to judge statistical results as insignificant also in-volves risk. There can be a loss to society in failing to support aprogram that is actually successful but available data do not clearlyindicate the success. TEMPO relates the above conclusions and thedetailed discussions in the remainder of the report as an objectiveevaluation in light of available data. It must be kept in mind that TitleI funded programs were still relatively new at the time of 1966-67tests and it is not reasonable to expect large gains in achievement sosoon.
-
ix
TABLE OF CONTENTS
PREFACE iii
Synopsis of Work iii
Synopsis of _This Report v
SUMMARY
LIST OF ILLUSTRATIONS xi
LIST OF TABLES xii
SECTION
1 I NTRODUCTI ON 1
Study Objectives 1
Summary of Data Collected 2
Summary of Technical Approach 6
2 CHANGES IN ACHIEVEMENT TEST SCORESBETWEEN 1 965-66 AND 1 966-67 12
The Measurement of Achievement andInfluence of Title I 12
Processing Achievement Data 18
Results of Statistical Analysis 25
3 PRELIMINARY ANALYSIS IN IDENTIFICATION OFDISTINGUISHING FEATURES OF SUCCESSFULCE PROGRAMS 60
District Level Title I Expenditure perPupil-11 Districts 60
Purpose of Two Case Studies in Allocationof Program Resources to the Grade Level 62
Information Required 64
x 67TMP-115
Analysis of District 10 CE Programs
Analysis of District 13 CE Programs 93
Conclusions from the Case Studies 120
4 OBSERVATiONS, CONCLUSIONS, AND
RECOMMENDATIONS 122
General Observations 123
Conclus3ons 127
Recommendations 131
65
APPENDIX
A PRINTOUT OF THE DATA USED IN THE SURVEY 136
B DESCRIPTION OF VARIABLES 156
C STATISTICAL MODELS FOR ANALYSIS OF
ACHIEVEMENT DATA 160
Confidence Intervals for Observation by Grade 160 .
Rationale and Models for Multivariate Analyses 162
Results from Analysis of Variance andAnalysis of Covariance 168
Correction for Effect of Errors of Measurementon Observed Correlation Between Initial TestScores and Changes in Test Scores 169
Variance of Weighted Averages 172
D SPECIFICATIONS OF ACHIEVEMENT TEST DATA 174
E CORRECTIONS FOR DIFFERENCES IN PRE-POST
TESTING DATES 180
F USE OF STANDARD T-SCORES 187
G DETAILED INFORMATION FOR DISTRICT 10 190
209REFERENCES
xi
LIST OF ILLUSTRATIONS
FIGURE NO. TITLE PAGE
1 Definitions of types of data and types of conclusiom. 8
2 Measures of educationally disadvantaged and effect
of CE. 9
3 Trend in achievement level. 15
4 Trend in achievement level. 16
5 Trend in achievement level. 17
6 Distribution of achievement scores in a typical Title Ischool compared to that for the entire notion. 19
7 Example of data processing and summary ofstatistics on each grade for 1965-66 and 1966-67. 24
8 Changes in test scores in District 13 grouped by school. 37
9 Racial distribution, mobility and trends of attendance(percent) in sample schools. 69
0 Observed change of reading achievement for test periodSeptember 1965 to September 1966District 10. 88
11 Exposure to CE--selected schools.
xii 67TMP-115
LIST OF TABLES
TABLE NO. TITLE PAGE
1 Description of sample school districts includedin statistical analysis. 3
2 Characteristics of sample elementary schools inDistrict 4. 4
3 School and pupil samples sizes for eleven sampleschool districts. 5
4 Grade levels tested (1965-66 and 1966-67) withvarious achievement tests and subtests in elevensample school districts. 21
5 Sample r,f schools analyzed in Phase I . 22
6 Average change in reading achievement test scoresfor the total sample. 28
7 Frequency of changes in achievement test scoresin the mean and at the first decile. 30
8 Contrasts between changes in the mean and at thefirst decile by district. 31
9 Cha%-:s in mean and lowest deci le achievemen l. testscores by school district. 33
10 Results from analysis of covariance. 36
11 Summary of selected state and allocation variables bydistrict. 40
12 .::orrelations between selected variables, computed fromcombined data on all districts. 41
13 Correlations between selected variables, computed fromdata for District 13. 42
14 Estimated regression coefficients relating change inmean achievement to seleclild variables. 44
..........1110116.111M11,110111MMOM.........................----....-...........1.00.1.1011.11~11
TABLES
15 Estimated regression coefficients relaHng change inachievement at the first decile to selected variables. 46
1 6 Correlations between grade level and changes in
achievement scores. 47
17 Summary of achievement level 1 965-66 and changes inachievement between 1965-66 and.1966-67 by grade. 48
18 Average of changes in achievement scores, by percent
Negro in school. 51
1 9 Correlations between percent Negro in a school and
achievement scores. 52
20 Significance levels of differences in mean change
among groups based on percent Negro. 53
21 Correlations between change in percent Negro in aschool and achievement scores. 55
22 Average change in achievement test scores, by type
of change in racial composition. 56
23 Correlations between change in attendance rate and
changes in achievement scores. 58
24 Average Title I funds per pupil and changes in
achievement. 62
25 Sample selection criteria and compensatory education
programs District 10. 67
26 Attendance, mobility and racial distribution of sampleschools (percentages) in District 10. 68
27 Teacher aide assignments. 72
28 Expenditures for regular and compensatory educationprogramsSchool 1. 82
29 Summary of program expenditures of District 10. 83
3 0 Program expenditures by grades of School 1. 87
31 Expenditures Grade 2 (1 965-66) and achievement data
for Grade 3 (SeptemlAer 1966) of District 10. 90
32 Comparisons between grades with CE projects and grades
for which achievement data could be analyzed. 91
33 Title I project budgets in District 13. 94
xiv 67TMP-115
34 Scmp'e school characteristics, 1965-66. 96
35 Estimated CE resources for primary remedial readingSchool 3.
1 AA, 0"7,
36 CE activities serving sixth grade cohorts and levels of
detail of available information. 107
37 Summary of CE expenditures for sixth grade cohorts. 108
38 Demographic characteristics of the sixth grades of
Schools 5 and 7. 109
39 Regular classroom teacher expenditure for sixth grade
cohorts. 113
40 Expenditures,per pupil for sixth grade cohorts. 115
41 CE expenditures for sixth grade cohorts at Schools 5
and 7. 116
42 Regular classroom teacher expenditure for the fifthgrade cohorts at Schools 5 and 7. 119
43 Expenditures per pupil for fifth grade cohorts. 119
44 Analysis of covarianceDistrict 13 schools (usingchange in lowest quartile as a measure of change in
achievement). 164
45 Analysis of covarianceseven school districts (usingchange in mean reading score as a measure of change
in achievement). 165
46 Analysis of covariance seven school districts (usingchange in lowest decile as a measure of change in
achievement). 166
47 Ariclysis of covarianceseven school districts (usingchange in lowest quartile score as a measure of change in
achievement). 167
48 Analysis of variance among school districts (IA. 169
49 Relation between assumed reliability values ofestimates of mean test scores and "true" correlationbetween initial level and gain (when measured
correlation is -0.43). 171
50 Preparation of District 1 achievement test data. 176
TABLES
51 Preparation of District 3 achievement test data.
Preparation of District 4 achievement test data.
53 Preparation of District 13 achievement test data.
54 Bases for adjustment factors required by variationsin dates between administration of pre and post tesh. 182
55 Change from beginning of year to middle of year norms:
SAT paragraph meaning. 183
56 Change from middle of year to end of year norms:SAT paragraph meaning. 184
57 Change from beginning of year norms to middle ofyear: MAT Reading. 185
58 Change from eleventh grade to twelfth grade norms:
MAT Reading. 185
59 Adjusted average changes in districts where pre and
post test dates differed by three or more weeks. 186
60 Conversion of raw score on Advanced Battery, Form W,of SAT to grade score and national percentile (for testsgiven in sixth grades during May and June of academic
year). 188
61 Converting percentiles to Standard T scores. 189
XV
177
178
179
62 Expenditures for regular
63 Expenditures for regularprograms: School 3.
64 Expenditures for regularprograms: School 4.
65 Expenditures for regularprcgrams: School 5.
66 Expenditures for regularprograms: School 6.
67 Expenditures for regularprograms: School 7.
68 Expenditures for regularprograms: School 8.
69 Expenditures for regularprograms: School 9.
and compensatory education
and compensatory education
and compensatory education
and compensatory education
and compensatory education
and compensatory education
and compensatory education
and compensatory education
192
193
194
195
196
197
198
xvi
.1111111.1..1111.0.01,311111110.14
70 Expenditures for regular and compensatory education
67TMP-115
programs: School 10. 199
71 Program expenditures by grades: School 2. 200
72 Program expenditures by grades: School 3. 201
73 Program expenditures by grades: School 4. 202
74 Program expenditures by grades: School 5. 203
75 Program expenditures by grades: School 6. 204
76 Program expenditures by grades: School 7. 205
77 Program expenditures by grades: School 8. 206
78 Program expenditures by grades: School 9. 207
79 Program expenditures by grades: School 10. 208
STUDY OBJECTIVES
SECTION 1
sk rreeNni serIENKIIry MN., Ilia No I g
1
The study is aimed at providing the Department of Health, Educa-tion, and Welfare with evidence as to the productivity of compensa-tory education ICE) programs for disadvantaged children, The studyhas focused particularly on the effects of Title I of the Elementaryand Secondary Education Act of 1965 during its first year and a halfof operation. The specific objectives of the study were to answerthe following questions:
1. Has statistically significant enhancement of pupil perform-ance resulted to date from CE programs ?
2. What school, pupil, and environmental characteristics areassociated with enhanced pupil performance?
3. What are the distinguishing features of successful CE pro-grams ?
Initial study plans included the following measures of enhance-ment: achievement test scores, attendance rate, drop-out rate, and
frequency of disciplinary actions. The effect of CE was to be mea-sured by comparing results on each measure before and after expo-sure to CE. However, because of time constraints, it became neces-sary to restrict analysis to achievement test scores and attendance.
qi
Within a school district each grade level received the same testthroughout the district. For each grade, one sub-portion of the testwas selected as a measure of achievement. In most cases this wasthe reading sub-portion, but in some cases a suitable but related sub-portion was used. .
The objectives and measures of student performance used in thisstudy are short-range ones; they do not deal with duration of changes
nor was an attempt made to forecast the students' future performance
2 67TMP-115
and earning power or benefits to society. Although these longer--range objectives and associated measures would be more appropriateand revealing as to the productivity of CE programs, such an approachwould require a much longer and different kind of study. However,it is believed that the present approach, which deals with the immedi-ate and observable impacts of CE programs, is an essential firststep in any examination of longer-range impacts.
SUMMARY OF DATA COLLECTED
The Office of the Secretary, DHEW, and the Office of Educationselected school districts (see Table 1) from which to obtain datafor this study. Most of the districts selected were those for whichthere was reason to believe that successful CE programs were inprogress in at least some of the schools. In each of these districtsa sample of schools was selected so as to include high, medium, andlow rankings on each of tke following characteristics:
Student Attributes:
1, Educational deprivation
2. Economic deprivation
3. Mobility
4. Grade level.School Attributes:
1. Size of enrollment
2. Racial composition
3. Attendance rates.
CE Program Attributes:1. History of prior CE programs
2. Intensity of current CE programs.,
In each school district, schools eligible for ESEA Title I supportwere identified, pertinent characte%istics recorded, and a samplechosen so as to include the desired wide range of values for each ofthe above characteristics. For example, in a given district, severalschools were chosen with predominemtly Negro pupil populations butwith different degrees of economic and educational deprivation.
SECTION 1
Table 1. Description of sample school districts included instatistical analysis.°
3
School DistrictI TN./1)e of Local1 '
Education AgencyC74 eographic Region
1 City East North Central
2 County South Atlantic3 City West North Central4 City East North Central5 City East North Central6 County South Atlantic8 City East South Central
10 County West South Central
12 County South Atlantic13
_City Pacifi c
14 City Pacific
NOTE:a By agreement with MEW and the participating school districts, the school
districts will not be identified in the results of the study.
Within a given set of characteristics, selection was made withoutprior knowledge of the success of compensatory programs at any ofthe schools within a school district. Table 2 illustrates the spec-,trum of characteristics presented in one of the sample districts, Asimilar diversity of characteristics exists in other districts.
Field trips were made to fourteen school districts, but sufficientdata for analysis were available from only the eleven districts shownin Table 1. The number of sample schools in each district thatwere included in the statistical analyses are shown in Table 3.
The principal typos of data obtained for the fourteen school dis-tricts were: (1) achievement test scores, attendance records, andother available measures of pupil performance, (2) descriptions ofcompensatory education programs, (3) financial data describing ex-penditures for regular and CE programs, and (4) information onschool and pupil characteristics.
The additional data, over ar-c.; above those analyzed in this report,included some achievement te :;,,t. data on the 3 school districts not in-cluded in the statistical analyses and data for several prior years forseveral of the other school districts. Data already received include
4 67TMP-115
Table 2. Characteristics of sample elementary schools in District 4.
School
Stui4entPopulation
(Oct. 1966 ADM)
PercentNegro
(1966-67)
Sixth GradeMean
Reading
Achievement(percent; loc
Percent ofStudents
EconomicallyDisadvantagedd
Years of CEa._ ............
IoyJune 1967
4 1010 99 2 70 1-1/2
6c 1617 98 10 60 3+ib 1193 95 8 75 3+
9 411 38 1 85 1-1/2
2 561 15 15 75 7
10 758 94 29 35 1
5 295 24 26 75 1
11 882 19 36 50 1-1/2
7 1330 97 41 35 1
3 541 38 45 45 1
8 661 4 52 35 1
NOTES:c Intensive Teacher Aide Pmgram in addition to Title I Teacher Aides.b Special Remediation Program in addition to Title I Remediation Pragrams.
Percentile rank within District 4, 1964-66.lcd Family annual income s $2,000 (1960 census data).
over 2600 grade-years and more than 250,000 pupil test scores. Sub-
stantial but varying amounts of information have been accumulated on
other measures of pupil performance such as attendance and drop-
out records.
A considerable amount of information has been obtained on up to
37 different CE activities in each of these districts. Additional infor-
mation describes characteristics of specific schools and pupils par-.
ticipating in CE projects. Financial data have been received from
ten of the fourteen districts in varying degrees of detail. In the other
four districts, total Title I appropriations are available from OHEW
reports. Average expenditures of Title I funds per pupil in these dis-
tricts up to the time of the 1966-67 achievement test varied from $21
to $140 (see Table 11 in Section 2, column entitled "Effective Title I
Dollars per Pupil").*
*These averages are total district Title I expenditures divided by total
pupil population in all schools receiving Title I funds.
_
SECTION 1
Table 3. School and pupif samples sizes for eleven sample school districts.
5
School DistrictNumber of Schools in Sample
Total AverageDaily Membershipin Sample Schools
1966-1967Elementary Secondary
1 19 5 19,100
2 14 5- ,... 9,3003 15 7 15,600
4 11 8 24,3005 14 9,2006 2 0 700
8 9 7 14,600
10 10 0 4,60012 4 0 1,800
13 7 4 10,700
14 6 4 7,700 .
TOTAL 111 40 117,600
In addition to the basic data on test scores considerable informa-tion is available on the relationship among the following types of testrecords: standard T, national percentile, standard scores for varioustypes of tests, grade equival:nce and stanine scores. Information onthe frequency of each type of test and the frequency with which testsare changes from year to year is also available. Since achievementtests were administered at different times during the school year, adiscussion of the significance of the differing testing times in evaluat-ing year-to-year changes has been presented in Appendix E, alongwith a summary of the testing schedules.
Frequent differences of definitions were found among school records.For example, some districts considered an excused absence as "pre-sent" for attendance records, others did not. Some computed pupilmobility for the school year, others for the calendar year. Recordsof pupil "drop-outs" were based on different definitions. The oc-currence of ungraded schools required that an equivalent grade levelof pupils be determined for purpose of comparisons. Sometimes asa result of these differences in definition, approximations, based onsuch data as were available, were required in order to create a setof comparable data (e. g. , enrollment data had to be used in lieu ofADM*).
*Average daily membership.
----"migmerdeleXelbtic.-1
6 67TMP-115
Some CE projects were defined in terms of activities or objectivessuch as remedial I ,ading and English instruction for pupils from fami-
lies where English is not spoken; other CE programs were defined in
terms of resource inputs such as teacher assistants and dollars forschool equipment. The latter definition permits simple accounting,but does not permit easy identification of all the resources involved
in specific program objectives.
SUMMARY OF TECHNICAL APPROACH
In TEMPO's first interim report (67TMP- 7, 27 June 67) a detailed
discussion of the proposed technical approach was provided. In this
report, discussion will be limited to a summary of the interim re-port and a short discussion of subsequent changes and additions.
Data from th Q. eleven school districts were analyzed to provide
an estim ,tE ,.)4 the effectiveness of CE programs by comparing achieve-
ment scc- , 1966-67 (after schools had implemented CE programsfunded irom Zit le I) with achievement scores in 1965-66. In most
cases, there was little or no CE funded from Title I up to the time of
the 1965-66 tests.
An attempt was made to identify the student, school, and programcharacteristics that were most highly correlated with such changes
in effectiveness. This approach allowed for the possibility that stu-
dents from different socio-economic groups respond differently to the
various CE programs. For example, a certain program or activity
may be highly successful in one group but not in another group with
differesit socio-economic characteristics.
The "fixed guide" approach was used because of the availability ofachievement data from sample schools. In this approach distri-bution of pupil test scores for a spe;:lfic grade in a school in 1966-67
was compared to the distribution of test scores for correspondingpupils in the same gra("1 and school in 1965-66. The statistical ana-lysis included 314 specific grades (see Table 5, Section 2) representingapproximately 35,000 pupils in each year. This means that there were
314 basic observations, however, there is still a choice as to whichparameter(s) of the distribution should be used for measuring charse
in achievement. The analysis in this phase of work was limited to
the following four parameters: change in mean test score; change in
the test score at the first decile; change in the test score at the firstquartile; and change in the test score at the third quartile.
SECTION 1 7
In many schools there were several classes in each grade. Thescores from each class within a grade were combined so that all pu-pils in a grade constituted one observation. This seemed advisablehpr=uae thpre w=f: pn rptinble way to select clamses in each of aca-demic years 1965-66 and 1966-67 so that they would be comparableexcept for CE programs. This procedure does, however, lead toanother question in interpreting results.
The number of pupils in each of the 314 basic observations in eachof the years 1965-66 and 1966-67 varied from 16 to 598. This raisesthe question whether each of the 314 observations should be giventhe same weight in judging the magnitude and statistical significanceof average change in test scores between the two years. If longitudi-nal data (i.e., test scores for the same pupils in both years) had beenused, the obvious choice would be to weight each pupil equally. How-ever, in the "fixed grade" approach there is no unequivocal way tomatch individual pupils in 1965-66 with those in 1966-67. The choicewas made to present two types of statistics, namely,
1. Unweighted averages and standard deviations of the 314basic observations2. Weighted averages of the 314 observations based on num-ber of pupils in each observation.
The difference between the two averages can be easily discussed withthe aid of Figure 1.
The data available for statistical analysis in this study were in theform indicated in the lower right box in Figure 1. However, wheneach observation is weighted by the sample size, it is equivalent todata appropriate for the lower left box. The distinction between leftand right boxes is clear with respect to the average test score but itis not so clear with respect tc test scores at the first decile, firstquartile, and third quartile. The latter scores may represent thescore of only one pupil even though there are 598 pupils in the grade.In the case of decile and quartile scores, it is best to think of theweighted average as representing a simple average of scores of allpupils who ranked within an in:-.erval, say, 8 to 10 percentile in theirgrade in their school. The interval concept removes the problem ofa einglc score representing the decile or quartile score regardlessof grade size.
8
LONGITUDINAL
(e.g., comparisonof pupi 1 (s) score in4th grade in postyear with score ofsome pupil(s) in3rd grade in preyear)
FIXED GRADE
(e.g., comparisonof pupi 1(s) score in4th grade in postyear with score ofpupil(s, in 4thgrade in pre-year)
67TMP-115
PUPIL GRADE
I
314 314
;ElNI + N23 Ix2
X11 El [X2 X11
aveAX aveilX -17.Tato! pupits post and pre 314
PRE YEAR: N1 'X
1 '- POST YEAR: N2 ' X2
1
Figure 1. Definitions of types of data and types of conclusions.
The weighting process indicated by the equation in the lower leftbox of Figure 1 is not a simple average of pupils in an interval but
represents an average of single scores adjusted for the greater sta-bility (or confidence) in the measure (decile and quartile scores) for
grades with a larger number of pupils.
None of the conclusions presented in this report are based on lon-
gitudinal data as represented by the upper two boxes in Figure I.It is possible to use data from the lower boxes to make statementsappropriate for the upper boxes but it requires some added data or
assumptions.
Since data as to which pupils in sample schools received CE were
not available, all pupils for which test scores were available were in-cluded in the sample. Measures of average achievement indicated in
the upper part of Figure 2 are available but measures of potentialachievement level indicated in the lower part are not.
lf, for a given grade level, the gap between the two lines in the
upper part of Figure 2 is increasing each year, the simple compari-
son of pre and post years is a biased measure of the effect of CE;
that is, the average score for the pre year is nat a good estimate of
SECTION 1 9
AVERAGE PUPILENTIRE NATION
NI0-0.4°
CE IN ....4.0
<1th GRADE -40"
40,0e.
EDUCATIONALLY DISADVANTAGED
,
POTENTIAL LEVEL OFEDUCATIONALLYDISADVANTAGED
6 10 12
o)CE IN
StIl GRADE -- 4.do-,°14EDUCATI-ONALLY DISADVANTAGED
2 4 6 8 10 12
YEARS OF SCHOOLING
Figure 2. Measures of educationally disadvantaged and effect of CE,
10 67TMP-115
the broken line in the upper part of Figure 2. In an attempt tocompensate for this possible bias the analysis of pre and post testscores was extended to include analysis of the contrast betweenchanges in the lower decile and changes in the mean and quartiles.The later analysis is based on the assumption that a high percentageof pupils in the lower decile of their class received CE but a smallerpercentage of total pupils received significant amounts of CE. Thespecific statistical procedure for analyzing the contrast is given inSection 2.
The objective of the statistical analysis was to estimate the dif-ference between pupil scores when they had CE and what their scoreswould have been in the absence of CE (broken lines in Figure 2).However, data on scores on the same pupils without CE are not avail-able so most analyses were done by a straight comparison of 1965-66and 1966-67 test scores.
The technical approach called for analyzing each of the 314 obser-vations separately if statistically reliable results could be developed.This would avoid the problems inherent on averaging data from dif-ferent tests and averaging data across different grade levels. It soonbecame apparent that this approach was not appropriate and datawould have to be pooled. First, there is considerable variation intest scores between years that cannot be estimated from availabledata on the variation among pupils within a year. For example, thetest could be given under very good conditions one year and poor con-ditions the other year. Second, the sampling variation in test scoreswas so large that few of the observed differences in pre and postscores for individual observations were statistically significant.Therefore, the scores were all converted to Standard T-Scores so thatobservations could be pooled. The assumptions underlying the eval-uation of T-Scores for different grades and schools are given in Sec-tion 2. .
In summary, the basic approach in Phase I was, first, to deter-mine on a grade-by-grade basis whether significant enhancement hadoccurred and second, to identify whether, in the aggregate, the en-tire sample reflected significant enhancement. Regression analysisand analysis of covariance described in Section 2 were used to helpidentify student and school characteristics that are correlated withchanges in student performance.
SECTION 111
Statistical analysis is not sufficient for a complete evaluation ofthe compensatory education programs. Inconclusive results of sta-tistical tests on selected measures of effectiveness do not prove lackof progress. First, they may indicate that the clract.:.r4s14.-a anrnpl Ad
by the selected measures were not changed by the programs; changesmay have occurred in other characteristics that were not measured.Second, even if there may have been no short-term benefits, theremay be latent long-range benefits that are difficult to forecast oridentify after only one year of intensive nationwide compensatory ac-tivities. Third, the sampling variation in the relatively small sam-ples might be too great to detect actual changes in achievement.Even when statistically significant results are obtained, these mustbe evaluated in light of other relevant irlormation.
Limited attempt was made during the Phase I effort to collect sev-eral diverse types of information not amenable to formal statisticalanalysis but which might be useful in the overall evaluation of Title Iprograms. Some of this information is presented in trip reports inthe separate appendix and some is presented in the detailed analysisof two school districts in Section 3.
Nir
[
1
12
SECTION 2
CHANGES IN ACHIEVEMENT TEST SCORESRFTWFFN 1965-66 AND 1966-67
THE MEASUREMENT OF ACHIEVEMENTAND INFLUENCE OF TITLE I
67TMP-115
The primary measure of achievement used in this study is thescore on the reading sub-portion of various standardized achievementtests*. Obviously, this measure does not cover the entire range ofobjectives of compensatory education. However, reading is an es-pecially appropriate element in the evaluation of compensatory educa-tion for several reasons: there is some evidence that verbal behaviormay be more affected than other skills by cultural disadvantages(References 1 and 2), reading is a fundamental academic skill, andreading is part of all the major achievement tests used throughoutthe United States.
The study employs a "fixed-grade" approach in measuring theeffects of compensatory education programs on achievement. Thedistribution of achievement scores for children in a particular gradeand school for the year prior to compensatory education is comparedwith the distribution of achievement scores for children in that samegrade and school in the following year, when compensatory educationis implemented. An alternative method, the "longitudinal" approach,is to observe changes in scores of individual pupils in successivegrades (i. e., the change between an individualts score in his gradeprior to CE and his score in the succeeding grade after exposure toCE). The fixed-grade approach is based on results for two differentsets of individuals and its usefulness and validity are influenced byour ability to identify and cope with the additional factors introducedby comparing results for two different sets of pupils.
*There is one exception: in one district a composite score (i.e., anaverage score of several subtests of a test such as reading, vocabu-lary, spelling, language, arithmetic, etc.) was used because CEprograms were known to be oriented toward other academic skills.
SECTION 1 13
The longitudinal approach is attractive because it uses the sameset of individuals but it requires comparing results in two (ora-a-v.& a./ Att.J. V J. GAL I. 6 .11. CM& C it) .. t. A-1...
-I-) U 1., 1.11C comparison of achievement beforeand after compensatory education can involve large and possiblysystematic errors because pre and post scores for each pupil wouldbe based on different tests and evaluated against different norms.
In this specific study the choice of the fixed-grade approach wasdictated by greater availability of achievement data and becauseschool records do not link individual test scores with amount of ex-posure of the pupil to CE. Longitudinal data are very limited due tomobility of pupils, lack of records on individual pupils, and the gene-ral policy of not testing every grade every year.
Most of the analyses on Phase I were based on test data for twoyears, the 1965-66 and 1966-67 school years. Ideally, the "post"year (1966-67) would represent performance after exposure to TitleI programs and the "pre" year (1965-66) would represent performanceprior to exposure. However, some of the pre-year (1965-66) testswere conducted in the Spring of 1966 after Title I funding started. Itis assumed that pupils tested in the Spring of 1966 showed no enhance-ment attributable to Title I. This assumption seems justified becauseTitle I activities usually did not start until February 1966.and theactivities often took months to get under way. Also, a large portionof the total first year outlay was for equipment and construction (Ref-erence 3), which would take relatively long to have effect, comparedto reading programs, for example, which might give fairly rapidenhancement.
Another assumption in a simple comparison of pre and post yearsis that the major positive changes in achievement test scores aredue to benefits derived from Title I. The general descriptions re-ceived at the school districts support this assumption. Certainly formost schools sampled, the amount of special CE expenditures wasgreatly enhanced by Title I. Also, this CE was designed as an addi-tion to the on-going school program, not a substitute for it. As willbe seen later in this report, there is wide variation among gradeunits in both CE and regular program expenditures per pupil. Thisis a possible cause of large sampling variation in achievement testresults.
The most important uncertainty with respect to judging the effectsof CE on the basis of observed differences in scores in successive
14 67TMP-115
years is the possibility of trends in achievement which are independent
of CE. There are several reasons for expecting a downward trendin achievement of pupils at inner-city schools (which constitute mostof the sample) relative to the entire nation. First, there is an exitof middle and upper income whites to the suburbs (Reference 4).Second, there is an inflow of families from rural areas where theeducational level c.f. pupils is probably quite low. Third, riots andother forms of demands for racial and economic equality are disrup-tive forces that can hinder the educational process when these forcesare being exerted.
A second source of systematic year-to-year change in achievementlevels springs from the nature of achievement measures. In this
case the expected trend is in the direction of improvement in averageachievement level in successive years. Several test publishers*report that successive classes throughout the nation obtain higherscores on tests. It is not clear why this result occurs; it may bedue to increased spread of knowledge, or to teachers incorporatingtest content into their teaching.t In any event, this trend makestest norms progressively obsolete. When this occurs, and at what
rate, is not clear. However, from the publisher's SAT equivalencetablest this rate was estimated as approximating 0.5 Standard T-score units, per year. This is approximately a 0,05 increase ingrade equivalence score per year (1. e. , the upper line in the tophalf of Figure 2 is shifting upward). In analyses to date we havenot taken potential trend influences into account. However, someachievement data for earlier years were processed to gain some in-sight on how trend information might be utilized. All of the datawere from one district (8); they include five grade levels (4, 6, 7, 8and 11) and four different years. A representative sample of thesedata was plotted and appears below as Figures 3, 4, and 5. Thesefigures illustrate the difficulty in giving a precise interpretadon tosuch information. There are some cases (for example, the graphof the 1st decile in Figure 3) where the achievement level movesrather erratically, and the existence of a trend is not obvious. In
other cases (the graph of the 1st decile in Figure 4, for example)there is an apparent absence of any trend. A third situation is shownin Figure 5, where a trend is apparent in the lower decile and lowerquartile, and one is tempted to predict what the 1967 level would
have been in the absence of any CE by simple extrapolation.
*See, for example, References 5, p 53, and 6, pp 28-29.t See, for example, Reference 7, pp 152-155.
SECTION 2
DISTRICT 8SCHOOL 14GRADE 11
s............./......
50 0
.0.000,00,,4C-41
+ 1st DECILEo 1st QUARTILE
MEDIANO 3rd QUARTILEs 9th DECILEA MEAN
1964 1965 1966 1967 YEARS
Figure 3. Trend in achievement level.
15
16
701 DISTRICT 8SCHOOL 15GRADE 11
s'..,I-2 *D 50I ....T....42<0Z a
6-a
.''..°°°°°°"-----------z.....
2CIk..) +.(.)
20
10
0
67TMP-115
1st DECILE1st QUARTILEMEDIAN3rd QUARTILE9th DECILEMEAN
1964 1965 1966 1967
.......--sv,Figure 4. Trend in achievement level.
YEARS
DIS
TR
ICT 8
SC
HO
OL 9
GR
AD
E 4
+ ht DE
CIL
E
0 1st
QU
AR
TIL
E
ME
DIA
N
El
3rd
QU
AR
TIL
E
* 9th
DE
CIL
E
A ME
AN
1966
1967
Fig
ure 5. T
rend in ac
hiev
emen
t
leve
l.
YE
AR
S
17
18 67TMP-115
There are difficulties in interpreting changes in test scores insuccessive years in the fixed-grade approach when there is possibilityof a trend in achievement level in the sample schools. It may be nec-essary to utilize the kngitudinal method along with the fixed-grademethod in order to develop more precise estimates of tha effertive-ness of Title I programs.
PROCESSING ACHIEVEMENT DATA
There are two main aspects in obtaining comparable units for ob-served differences in test scores. One is comparability over a rangeof achievement levels. The other is comparability among differenttests and different test dates within an academic year.
The comparability over different achievement levels within a gradewas obtained by converting to the Standard T-score. The scale forthe Standard T- score has been constructed so that a change from 30to 35, for example, is comparable to a change from 60 to 65. Com-parable in this case means that the amount of effort in CE requiredto raise the achievement level 5 points is approximately the samein both cases. As a result, T-scores can be averaged and subjectedto statistical analysis, whereas percentile scores, for example, can-not (Reference 8, p 64). In some of the analyses, results for differentgrades were combined and in these cases it was assumed that dif-ferences in T-scorc:s are also comparable among grades. That is,the amount of effort in CE required to increase an achievement levelfrom say 30 to 35 in one grade is approximated the same as for asimilar movement in -tiler grades. The T- score is computed directlyfrom a percentile score (see Table 61 ). The second aspect of com-parability among different tests and test dates enters through the useof the publishers' norm tables for computing percentile scores (seeTable 60 as an example of this conversion). The assumptions usedin computing percentile scores were:
I. The reading, paragraph meaning, and composite subtestsmeasure the same pupil attribute
2. The pu- thers' norm populations for each test is similarwith respec. to the distribution of achievement levels
3. The publishers' norms provide proper adjustment for thetin. e of year at which the test is given.
The first assumption was necessary because each of the three sub-tests was used in one or more school districts, Since our primary
-
SECTION 2 19
interest was in analyzing differences between pre and post test scores,it was not necessary to assume that norm populations were equal inabsolute level, but only in the range and distribution over the range.
Each of these assumptions will be treated in more detail in the sub-sections below. One feature of the data deserves special notice.Since pupils in eligible Title I schools were selected for being cultur-ally and educationally deprived, their achievement test scores tendto be lower than those of the norm populations. Figure 6 illustratesthe distribution of achievement scores of a typical fourth grade in asample school in 1965-66. Eighty percent of the pupils in this gradereceived test scores below the mean score of the pupils in the normpopulation. Table 11, discussed later, shows the mean achievementlevel (based on 1965-66 tests) for the entire sample and for sampleobservations in each district.
The major characteristics of the data received were as follow.
Type of Test
Data analysis was limited to standardized achievement tests ad-ministered to entire grades of pupils in a school district for each of
-ram=20
PUPILS IN A 4th GRADE IN A&"*" TYPICAL SAMPLE SCHOOL
NATION
30 40 50 60
STANDARD T SCORE
Figure 6. Distribution of achievement scores in a typical Title I schoolcompared to that for the entire nation.
70 80
20 67TMP-115
2 years, 1965-66 and 1966-67. In the 11 school districts, 5 differenttests were used: Stanford Achievement Test (SAT), MetropolitanAchievement Tests (MAT), Iowa Tests of Basic Skills (ITBS), IowaTests. of Educational Development (ITED), and Sequ-ential r.I:ests ofEducational Progress (STEP). The variety of tests and subtests indifferent grades in the 11 school districts is shown in Table 4.
For any given test there was often variation in the forms used inthe pre and post years. Usually this variation was between two formswhich were constructed so as to provide equivalent measures ofachievement. However, in cases where different editions of a testwere used, raw scores were converted to a common scale, usingpublishers norms.
Time of Testing
There was considerable variation in the time of the year in whichtests were given among and even within school districts. This re-sulted in differences in amounts of CE to which pupils were exposedprior to the 1966-67 test. Such differences in exposure to CE cannoteasily be taken into account in a simple comparison of pre and posttest results but can be incorporated in the regression analysis. Asshown in Appendix B, a new variable was constructed to show whatis called "effective Title I dollars" per pupil. This was computedby adding the Title I funds spent in 1965-66 and the fraction of TitleI funds for 1966-67 spent up to the time of the 1966-67 achievementtests and dividing the sum by total number of pupils in Title I schoolsin each district. This new variable was included as one of the de-termining variables in regrecsion and analysis of covariance.
Differences in test dates between 1965-66 and 1966-67 requireduse of different norm tables for pre and post tests or adjustmentswithin a particular norm table. The occasions on which test datesin the twoyears differed by three or more weeks are shown in Appen-dix E, Table 54. The corrections required for these differencesare shown in Appendix E, Table 59. The data presented in AppendixA have been corrected for differences in test dates.
Specific Grades/Students Exposed to Title 1 Programs
Relatively few school systems keep records on the amount andtype of CE programs at specific schools. This means that it wasoften not possible to identify precisely which grades in a school
Tab
le 4
. Gra
de le
vels
test
ed (
1965
-66
and
1966
-67)
with
var
ious
ach
ieve
men
t tes
tsan
d su
btes
ts in
ele
ven
sam
ple
scho
ol d
istr
icts
.
Sch
ool
Dis
tric
t
Sta
nfor
dA
chie
vem
ent
Tes
ts:
.. rara
grap
hM
eani
ng
tM
erop
olita
nA
chie
vem
ent
Tes
ts:
Rea
ding
Iow
a T
ests
of B
asic
Ski
llsIo
wa
Tes
ts o
f Edu
c. D
evel
.S
eque
ntia
l
Rea
ding
Com
posi
teA
bilit
y to
Inte
rpre
tlit
erar
ym
ater
ials
Com
posi
teT
ests
of
Edu
c.N
ogre
ss:
Rea
ding
14,
5, 6
, 7, 8
22,
3, 4
, 5, 6
34,
6, 8
10, 1
2
44,
6, 8
10, 1
2
56
4
63,
6
82,
3, 4
, 5,
6, 7
, 8, 1
110
3, 4
, 5
123
131,
2, d
ocs,
811
5
143,
4, 6
, 8...
......
......
......
.....
NO
TE
:a
Inst
ance
3 in
whi
ch d
iffer
ent e
ditio
ns o
f the
test
wer
e us
ed in
yea
rs 1
965-
66 a
nd 1
966-
67.
(A m n 1 0 Z " n)
Tab
le 5
. Sam
ple
of s
choo
ls a
naly
zed
in P
hase
l .a
Sch
ool D
istr
ict
Num
ber
of S
choo
ls fo
r G
rade
Tot
al G
rade
sT
otal
Sch
ools
.) ..,A -t
eI
I
ajoi
.,8
I9
1
10,
11i _ .2
110
174
44
3921
210
1212
1210
5612
315
155
11
3821
411
115
2716
514
1428
14
62
24
2
89
99
99
34
355
16
1092
213
9
44
4
137
77
72
232
11
146
66
186
TO
TA
L7
2642
7947
787
200
11
314
132
.
NO
TE
:a
Thi
s do
es n
ot in
clud
e da
ta u
n ai
l sch
ools
liste
d in
Tab
le 3
.
SECTION 2 23
received Title I programs. Furthermore, it was not possible toidentify the specific students in a gr-,.de who received CE from TitleI programs. Consequently, it was decided to use test scores for anentire grade and to use all grades in Title I designated schools forwhich test data were available for both 1965-66 and 1966-67.
It was also decided to limit the sample to grades 1-12 in publicschools. The sample of schools by grade and district which wereanalyzed is shown in Table 5.
Form of Data
Data received from school districts were in the following forms:class listings of individual pupils, punched cards of individual pupils,computer tab runs by individuals, and computer printouts of fre-quency distributions. Test scores included one or more of the follow-irg: raw scores, standard scores (for specific type of test), grade(or grade equivalent) scores, percentile scores, and stanines.
All scores not in national percentile were converted to nationalpercentiles by using the appropriate conversion table provided byeach test publisher for each separate form of test. Percentile scoreswere then converted to Standard T- scores. The distribution ofStandard T-scores in the population is assumed to be normal with amean of 50 and a standard deviation of 10. This assumption was util-ized in converting national percent-Iles to Standard T- scores.* De-tails on the form of data obtained from certain school districts aregiven in Appendix D and an example of the detailed procedures forconverting scores from one particular test is given in Appendix F.
Test results for each grade in each school for 1965-66 and 1966-67have been summarized in terms of the nineteen statistics (see lowerpart of example of computer output in Figure 7). The central partof Figure 7, illustrates the conversion from raw scores to StandardT- scores. The summary statistics are all in terms of T-scores.The examples show 84 pupils in a grade; in the overall sample, thenumber of pupils in a grade ranged from 16 to 598.
*This is consistent with reasoning by the Office of Education presentedin Reference 8.
24 67TMP-115
GRADE 6, t'OST YEAR scHaat. XXXDATA IN RAW SCORES TO HUN THRU UNE- SHARI NG COMPUTER STAII STICAL PROGRAM
41.46.608105s 41* 46. 6(8* 88. 41, 46, 59* 8 5, 8 5, 59, 4 6. 4 1, 4 6. 41. 57. 8 5
39* 46* 57.83.80* 56* 46* 39, 38, 4 4. 7 6. 5 4. 7 1. 5 4. 4 4, 36, 33. 52* 4467*32* 33. 44, 51.. 66. 64, 49. 42. 30. 39. 33, 42, 42.49.47. 49* 47.64. 62.6251* 51.36.39* 47* 47* 46* 41.33.30* 42* 30.62.62.54r 30* 32* 36* 30.38.49* 52* 30, 57
RAW SCORES RANKED IN ASCENDING 3HDER:
39 30 30 30 30 30 30 32 32 33 3333 33 36 36 36 38 38 39 39 39 4141 41 41 41 41 42 42 42 42 44 4444 44 46 46 46 46 46 46 46 46 4747 47 47 49 49 49 49 51 51 51 5252 54 54 54 56 57 57 57 59 59 6060 62 62 62 62 64 64 66 67 71 7680 83 85 85 85 88 105
RAM SCORES CONVERTrD TO STANDARD 1-SCORES AND KANKEU:
26.7 26.7 26.7 26.7 26.7 26.7 26.7 29. 529.5 31.2 31.2 31.2 31.2 32.5 32.5 32.534.5 34.5 34.5 34.5 34.5 36.1 36.1 36.136.1 36.1 36.1 36.6 36.6 36.6 36.6 37.737.7 37.7 37.7 38.3 3863 38.3 38.3 38.338.3 38 Z. 38.3 39.2 39.2 39.2 39.2 411., 8
40.8 408 40.8 41.9 41.9 41.9 42 3 42 343.6 43.6 43.6 44.2 44.7 44.7 44.7 45.445.4 46.4 46.4 47.5 47.5 47.5 47.5 48.148.1 50.1 59. 5 52. 1 54. 4 56.4 57. 7 58.858.8 588 59. 5 64.8
SUMMARY STATISTICS IN STANDARD T-SCORES:
SMALLEST VARXATEm 26.7 NUMBER OF VARIATES* 84LOWER DECILE* 29.5 ARITHMETIC MEAN' 40.381
FIRST QUARTILE" 34.9 STANDARD DEVIATION. 8.52709MEDIAN* 38.3 VARIANCE. 72.7113
THIRD QUARTILE* 45.225 COEFF OF VAR [PC11. 21.117UPPER DSCILEm 53.25 STANDARD SKEWNESS' .623
LARGEST VARIATEm 64.8 STANDARD EXCESS' .181
TOTAL RANGES 38. 1
DECILE RANGEm 23. 75
SEMI-QUART RANGE* 5.1625BOILEYS SKEWNESS= .341PEARSON SKEWESSm 732
Figure 7. Example of data processing and summary of statisticson each grade for 1965-66 and 1966-67.
160.10.N.411.111......
SECTION 2 25
Volume of Data
It was desired that the study include a wide variety of student andenvirorwnental charnctevistics. This required that a wide variety ofschool and program conditions had to be included, and that the over-all sample had to be large so that there would be a sufficient numberof observations within each type of condition to draw firm conclusions.Also, .since school personnel suggested that only a small amount ofchange in achievement could be expected so soon after the initiationof Title I programs, our sample would have to be correspondinglylarge to detect aty differences.
The amount of achievement test data used in the statistical analy-sis is shown in Table 5. The statistics shown in Figure 7 are avail-able for each of the 314 grades shown in Table 5 for both pre andpost years.
There is great variation in the amount oz: data collected from thevarious school districts. This variation resulted from differencesin size of school district, variety in CE programs among schoolsand grade levels and difficulty in obtaining information.
There are fewer secondary than elementary schools. In largepart, this is a natural consequence of the larger enrollment of secon-dary schools. It means, of course, that more students are representedby sample units at the secondary level.
The more relevant summary statistics along with pertinent dis-trict, school and grade characteristics for each of 314 observationsin the sample have been retained in punched cards (see complete list-ing of data in Appendix A). Ps. list and description of variables analy-zed in Phase I are given pendix B.
RESULTS OF STATISTICAL .A1YSIS
The summary tables and discussions will be presented in sectionscorresponding to the objectives stated on page 1. To focus attentionon the study objectives most of the technical discussion on statisticalanalysis has been put in appendices. The most relevant mathematicsand assumptions underlying the various statistical analyses are pre-sented in the main text but more complete details are in appendices.Also, the basic data on each of the 314 observations appear in Ap-pendix A.
26 67TMP-115
General observations that have been developed from the overallproject but not necessarily from the statistical analysis are presentedin the first part of Section. 4.
Results from many of the statistical analyses are not reported herebecause they were not statistically significant. Because of the greatvolume of data it seemed best to describe each analysis that wascarried out, but include in tables only those results which proved tobe significant. The attempt to judge whether the observed change ineach of the 314 grades was significant produced inconclusive results.First, the only measure of standard error was the observed varia-tion among scores within the pre year and within the post year btathis is known to be a biased underestimate of the variance of the ob-served change between pre and post years.* Second, even with thebiased underestimate of the variance of the change in achievementthe percentage of the 314 observations that would be judged as sig-nificant was approximately what would be expected by chance. There-fore, the tables and discussion in the remainder of Section 2 arebased on average changes or frequency of positive changes in variousgroups (a group could be total sample, a specific school district, aspecific grade, etc.).
The results using change in attendance as a performance measureshow no significant improvement between 1965-66 and 1966-67. Themean change for the entire sample was + 0.12 percent but the standarderror of this estimate is 1. 60 which means that the estimate is notsignificant at the 50 percent level. Some of the individual districtsshowed a larger change but there are other explanations for the change.For example, the severe weather in New Orleans reduced the atten-dance in 1965-66 and it was natural to expect a positi-ye change aslarge as the 2.8 percent that was observed. Because of the statisti-cally insignificant results for change in attendance, most of the dis-cussion of change in performance in the remainder of this section isin terms of change in academic achievement.
*Analyses of sample data indicates that the variance of the differencebetween pre and post means is at least twice as large as the sum ofthe estimated variances of the pre and post means. This means thatthe information for one grade (i.e., one pre year and one post year)is 1 It sufficient for judging statistical significance.
-
SECTION 2
_
27
lias Statistically Significant Enhancement ofPupil Performance Resulted to Date from CE Nograms?
This question will be addressed in terms of the overall sample and
in terms of the sample of schools within each of eleven school districts.The term "enhancement," as used here, is the difference betweenachievement level after exposure to CE programs and the achieve-ment level which would have been expected in the absence of suchprograms. Enhancement cannot be measured directly from avail-able data. It is possible to measure the achievement level of pupilswho received CE but the achievement level of the same pupils in theabsence of CE can only be estimate.A. Therefore, conclusions on de-.gree of enhancement of specific pupils must be developed by inference.
The results of most statistical tests presented below were basedon observed differences in achievement scores between 1965-66 and1966-67 rather than estimated differences between achievement withand without exposure to CE. That is, no adjustment was made for apossible negative trend; this means that observed differences betweenthe two years understate differences between achievement with and
without CE.
OVERALL SAMPLE. There is no indication of general improve-ment in the entire student population in the 314 grade observations.There is, however, indication that the achievement of students in thelower part of the distribution in their respective grades was slightlyenhanced between 1965-66 and 1966-67. Table 6 shows both theweighted and unweighted average change at the first decile as signifi-cant at the 20 percent level. As described in the Section on technicalapproach (see page 8), the weighted average is essentially anaverage of all pupils whereas the unweighted average is an averageof one statistic for each of the 314 grades.
Although data on the application of CE to various levels in theachievement distribution are not available it is reasonable to expectthat CE and, especially, remedial programs were usually orientedtowards those students whom are the most seriously disadvantagedand, therefore, have the lowest achievement scores. This meansthat we should expect more favorable results at the first decile.The change at the first decile is 0.25 Standard T-Score units. Thisis not easily interpreted in terms of grade equivalence because itrepresents the average change in many different grade levels. We
know however that it is in the neighborhood of one-fourth month andis, therefore, quite small in importance. If, however, this small
Tab
le 6
. Ave
rage
cha
nge
in r
eadi
ng a
chie
vem
ent t
est s
core
sfo
r th
e to
ta;
sam
ple.
°
Ach
ieve
men
t Tes
t Sta
tistic
Unw
eigh
ted
Obs
erva
tions
bE
ach
Obs
erva
tion
bW
eigh
ted
by S
ampl
e S
ize
Ave
rage
Cha
nge
Sta
ndar
dE
rror
Sig
. Lev
el°
Ave
rage
Cha
nge
Sta
ndar
dE
rror
Sig
. Lev
el
Mea
n (3
)
Low
est
lieci
le (
D1)
Low
est Q
uart
ile (
Q1)
Upp
er Q
uart
ile (
Q3)
- 0.
29
+ 0
.25
- 0.
30
-0.
48
0.13
0.18
0.16
0.17
0.05
0:20
0.10
0.01
- 0.
40+
0.2
1
- 0.
42
- 0.
69
0.11
0.16
0.14
0.15
0.00
1
0.20
0.01
0.00
1
Not
es:
a In
uni
ts o
f Sta
ndar
d T
-sco
res
and
base
d on
obs
erve
d ch
ange
s in
314
sch
ool-g
rade
s in
11 d
istr
icts
.31
431
4b
Tne
unw
eigh
ted
aver
age
is c
ompu
ted
as31
4 X
i and
the
wei
ghte
d av
erag
eis
co
mpu
ted
asm
. X.
Ei=
1i=
1
314 E m
. whe
re m
. is
the
aver
age
num
ber
of p
upils
who
took
the
pre
and
post
test
and
X. i
s th
e ob
serv
ed
i=1
chan
ge in
the
spec
ified
sta
tistic
. App
endi
x C
dis
cuss
esth
e co
mpu
tatio
nal p
roce
dure
for
the
estim
ated
stan
dard
err
or.
c T
his
is th
e pr
obab
ility
that
the
obse
rved
sam
ple
resu
lt co
uld
have
happ
ened
by
chan
ce if
the
true
cha
nge
betw
een
1965
-66
and
1966
-67
was
inde
ed z
ero.
CO
711''
SECTION 2 29
positive value indicates a reversal of a negative trend it is of consi-derable importance. In judging the relevance of the g.0 percent sig-nificance level for the unweighted decile it is important to keep inmind that although it is not highly significant it surely does not sup-port the hypothesis that there was no change.
Table 6 also indicates a decrease in achievement in all but thelowest decile between 1965-66 and 1966-67. These results supporta second important conclusion namely that even with CE programspresent there is a negative trend in achievement level in schools to-wards which CE is oriented, reflecting the sociological and economicchanges that are taking place in the student population.
The positive change in the decile, while not very large, becomesmore significant when contrasted with the negative change in AX,AQ 1, and AQ3. The four statistics used (AX, 4D1, AQ1, AQ 3)* arenot independent as indicated by the correlation coefficients shownlater in Table 12. Also, except for the effects of CE one expectsall four statistics to move in the same direction since they all comefrom the same distribution. One can then ask the question, how un-usual is it to obtain the deviations actually observed if in fact therewere no differences between the true means of AD
1and AX? The
answer to this question is that the observed differences between AD].and AX would be highly unlikely by chance alone (less than 0.001probability).1
Thus, although the absolute change in the first decile (+0.25) wasnot highly significant, the difference between the positive change inthe decile and the negative change in the mean was significant at the0.001 level. Hence, this is an indication of a positive effect of CEprograms if one assumes that CE programs are usually concentratedon the lowest achievers in each grade.
The contrast between changes in the mean and changes at. the firstdecile can be studied further by analyzing the frequency dam in Tables7 and 8. For example, there are more positive than negative changes
*A indicates change in the statistic from 1965-66 to 1966-67 (postyear minus pre year); X, Di, Q1, Q3, refer respectively to achieve-ment at the mean, first decile, first quartile, and third quartile.thetermined by testing for differences between correlated meansusing the correlations found in Table 12.
Tab
le 7
. Fre
quen
cy o
f cha
nges
in a
chie
vem
ent t
est s
core
s in
the
mea
n an
d at
the
first
dec
i le.
Dis
tric
t
All
Dis
tric
ts
1 2 3 4 5 6 8 10 12 13 14
Tot
al G
rade
sin
Sam
ple
Incr
ease
s
314
145
3923
5618
3819
278
2811
43
5525
136
43
3221
188
Mea
n
Dec
reas
esI N
o C
hang
e
Firs
t Dec
ile
163
I6
15I
1
380
190
181
161
10
291
70
1I
o
101
91
Incr
ease
s
126 9
24 20 10 7 3 16 7 3 17 10
Dec
reas
esN
o C
hang
e
103
85
1020
2012
153
134
1011
10
11A
51
1I
0
11
6
4 2
SECT10.4 2 31
Table 8. Contrasts between changes in the mean and at the firstdeci le by distri ct.°
DistrictObservations withChange in Means
more favorable
1 22.5
2 14
3 16
4 11
5 15.5
6 0
8 20
10 7
12 1
13 14
14 8
TOTAL 129.
I
1
1
Observations withChange in Decile 1 TOTAL I Sig. Levelblmore favorable I
16.5 ov). ..4k)
42 56 0.001
22 38 I 0.40
16 27 0.40
125 28 c
t 4 0.005
35 55 0.05
6 13c
3 4 0.40
18 32 0.50
10 18c
185. 314. 0.01
NOTES:a Differences among districts in iratio of "observations favorable to the decile
to total observations" was not significant at the 5 percent level (x2 =17.56 < x2
05(10)= 18.3).
20'
b The x test was used to determine if the function of observations favorable tothe deci le in each district was significantly different from one-half.
c Greater than 0.50 percent.
at the first decile but the reverse is true for the mean. The numberof nbservations for which the change was measured as zero is impor-tant in evaluating the precision of the measuring instruments. Eighty-five or 27 percent of the 314 observations showed no change in thefirst decile while only 2 percent showed no change in the mean. Thetest scores are less precise in measuring low achievers within eachgrade and the publishers' conversion tables for computing nationalpercentile are also less precise at loyv achievement levels within eachgrade. This lower precision at the first decile is possibly one ofthe reasons why the averlge change for the entire sample is not highlysignificant.
1
.a,...:11..1010.111.111W+.:.-u.
32 67TMP-115
The tendency for the decile to increase relative to the other param-eters can be examined by counting the number of observations forwhich the decile showed a more favorable change than the mean.As shown in Table 8 there were considerably more observationsfavoring the decile than the mean. Based on 314 observations, thedifference in proportions is significant at the 1 percent level. Thedistricts which show this contrast between changes in the decile andmean most clearly and strongly are 2, 6, and 8, for each of which thedisproportion between changes in 1.--e mean and decile is statisticallysignificant.
This study did not include analysis of the nature of the CE pro-gram in each of the 11 districts. Therefore, no full description ofthe programs in districts 2, 6, and 8 is avaiiable to help explain theemphasis at the lower end of the achievement distribution. It mightbe noted, however, that the amount of Title * funds per pupil expendedby the time the post test date tends to be relatively low in two of thethree districts (see Table 11).
INDIVIDUAL SCHOOL DISTRICTS AND VARIATION AMONG DIS-TRICTS. The average change in the mean and at first decile for eachdistrict is shown in Table 9. Some of the statistically significantchanges are positive and some are negative.
Due to the large variation in results among observations withina school district only three individual districts show a significantchange in the mean and only one district shows a significant changeat the first decile. The lesser number of observations in each in-dividual district makes the test for statistical significance less power-ful than the eorresponding te, t for the entire sample. The expectedlarge variation within and among districts is one of the reasons whythis study was designed to include a large sample and to include ob-servatio is from several different school districts.
,
School District 13 shows significant positive changes in the meanand at the first decile. This district is analyzed in gr,:ater detailin Section 3. District 6 shows a greater positive change than District13 but there are only 4 observations in District 6 and the results arenot statistically significant. There is no obvious explanadon of thelarge negative changes in Districts 2 and 4.
Comparing the weighted and unweighted averages in Table 9 onecan see that weighting by sample size produced more favorable re-sults in some districtr_ and less favorable results in other districts.
...row., -. r.....00.-1 ,.e........... ....... ....--......--*
Tab
le 9
. Cha
nges
in m
ean
and
low
est d
ecile
ach
ieve
men
t tes
t sco
res
bysc
hool
dis
tric
t.
Num
ber
of G
rade
sO
bser
ved
12
3
3956
42
A. U
nwei
ghte
d O
bser
-va
tions
Ave
rage
Cha
nge
inM
ean
0.25
- 1.
350.
34-
1.03
Sta
ndar
d E
rror
0.31
0.38
0.39
0.27
Sig
. Lev
eP0.
500.
010.
400.
01
Ave
rage
Cha
nge
inLo
wes
t Dec
ile0.
050.
040.
220.
68S
tand
c-ci
Err
or0.
320.
480.
580.
72S
ig. L
evel
ab
bb
0.40
B. O
bser
vatio
ns W
eigh
-te
d by
No.
of
Pup
ils in
Gra
de
Ave
rage
Cha
nge
inM
ean
- 0.
07-
1.19
- 0.
210.
92S
tand
ard
Err
or0.
260.
370.
300.
2AS
ig. L
evel
°0.
010.
500.
01A
vera
ge C
hang
e in
Low
est D
ecile
- 0.
410.
130.
21-
0.19
Sta
ndar
d E
rror
0.32
0 47
0.45
0 58
Sic
. Lev
el°
0.30
6b
6
Sch
ool D
istr
ict
5 28 0.16
0 27
,I, 0.70
0.50
0.20
0.25
0.24
0.30
0.76
0.46
0.20
68
1012
1:3
14
5513
432
18
1.28
- 0.
37-
0.21
0.42
1.16
0.01
0.97
0.27
0.83
0.97
0.46
0.34
0.30
0.20
bb
0.05
4.20
0.26
0.55
1.78
1.32
0.95
1.82
0.37
1.00
2.02
0.69
0.64
0.10
0050
b0.
500.
050.
20
1.58
- 0.
650.
071.
010.
690.
040.
850.
230
067
0.86
0.41
0.36
0.20
0.01
0.30
0.10
5.23
0.36
0.45
3.17
0.97
0.88
1.81
0.30
01.
770.
530.
670.
050.
3068
50.
200.
100.
30
NO
TE
S:
a T
hepr
obab
ility
of o
bser
ving
an
aver
age
chan
ge a
s la
rge
or la
rger
than
the
one
show
n in
the
prec
edin
g ro
ww
hen
the
time
chan
ge is
zer
o is
equ
al to
the
prob
abili
ty s
how
n in
this
row
.b
Gre
ater
than
0.5
0 pe
rcen
t.
CA
)C
A)
34 67TMP-115
The significance level of the positive changes in District 13 is lessfor the weighted than for the unweighted. On the other hand, the sig-nificance level for the positive changes in District 6 was greater forthe weighted than for the unweighted. Except for testing a specifichypothesis TEMPO suggests that the weighted and unweighted aver-ages be viewed as supporting the same general conclusions on ef-fectiveness of CE.
Tests of significance using analysis of variance were performedto determine if there were significant differences in observed changesin achievement test scores among school districts. These analysesshowed that differences in changes in mean achievement were signifi-cant at the 1 percent level (see Table 40 ). However, differencesmeasured in terms of changes in achievement test scores at the firstdecile and first quartile were not statistically significant.
Analysis of covariance was used as a more discriminating testfor detecting significant differenccs among districts. In this tech-nique analysis of variance is performed after an adjustment is madefor differences caused by variation in specific state variables. Thereis considerable variation in state variables among districts (seeTable 11) which could cause changes in achievement. In other words,significant difference regardless of the cause or conditions is testedby analysis of variance, while significant differences from causesother than the specified covariates is tested by analysis of covariance(see discussion in Anpendix C ).
The analysis of covariance was extended to include a test for sig-nificant differences among schools within District 13. The resultshave important hearing on whether the effort in Phase II shculd con-centrate on obtaining information at the school or grade level.
The basic model for analysis of covariance was
Y = T. + a1
($) + a2
(L) + a3
(A) + a4(%N) + c
where
(1)
Y = measure of change in achievement (A 5, A D1, or A 01)
T. = the variation in changes in achievement level that is at-]. tributable to the ith group
SECTION 2 35
$ = average Title I CE program funds per student (for allstudents in Title I schools in the school district) expendedby the post test date,
L = mean achievement level at the beginning of Title I programs,
A = attendance rate in a school (1965-66)
rAN = percent of pupils who were Negro (1965-66)
c = an error component in the postulated relation.
There were seven districts and therefore seven groups in the analy-sis of covariance for testing significant differences amok.s districts.There were eleven schools and therefore eleven groups in the analy-sis for testing significant differences among schools within District13. The variable for Title I $ was dropped from model in the analy-sis of District 13 schools because Title I expenditures were not avail-able at the school level.
The results from analysis of covariance are shown in Table 10.The major conclusions are:
1. The variation in changes in the meat among districts aresignificant at the 1 percent level. As in the analysis of variance,the differences as measured at the first decile are not signifi-cant at the 5 percent level. It must be realized that samplingvariation at the first decile and first quartile are larger thanat the mean and therefore the F test based on these measuresis not as sensitive as the test based on the mean.2. The analysis of District 13 schools shows no significantdifferences among schools. This means that changes in tesi,scores between 1965-66 and 1966-67 are not associated withspecific scnools and, therefore, variation in results cannot beattributed to differences among schools.
The analysis of covariance for testing significant differencesamong schools within a district was limited to a test based on changein the lower quartile for schools in District 13. There seems littleneed to extend these analyses since other data also indicate no signifi-cant differences among schools.
Detailed results for analysis of covariance are presented in Ap-pendix C. The coefficients of the covariates presented in the detailedtables are analogous to regression coefficients. They provide another
Tab
le 1
0. R
esul
ts fr
om a
naly
sis
of c
ovar
ianc
e.
Diff
eren
ces
AM
ong
Sev
en S
choo
lD
istr
icts
°N
o. o
f Gra
deO
bser
vatio
nsF
S:a
tistic
s fo
r N
ull
Hyp
othe
sisb
Cha
nge
in m
ean
achi
evem
ent s
core
s24
0F
= 3
.85
>F
(1%
)=
2.9
0
Cha
nge
in lo
wer
dec
ileac
hiev
emen
tsc
ores
240
F =
1.7
1 <
F (
5%)
= 2
.14
.C
hang
e in
low
er q
uart
ileac
hiev
emen
tsc
ores
240
F =
2.6
0 >
F (
5%)
= 2
.14
Diff
eren
ces
.Am
ong
11 s
choo
ls in
Dis
tric
t13
1. C
hang
e in
low
er q
uart
ile a
chie
vem
ent
scor
es32
F =
1.4
2 <
F (
5%)
= 2
.38
NO
TE
S:
Onl
y sc
hool
dis
tric
ts 1
, 2, 4
, 8,
10, 1
3, a
nd 1
4 fo
r w
hich
data
on
the
stat
e va
riabl
es u
sed
wer
eav
ail-
able
wer
e in
clud
ed in
this
ana
lysi
s. D
ata
used
in a
naly
sis
of c
ovar
ianc
e iid
not
incl
ude
the
adju
st-
men
ts in
dica
ted
in T
able
59.
The
se a
djus
tmen
ts a
re s
mal
l and
wou
ld n
ot li
kely
cha
nge
the
re3u
lts in
this
tabl
e if
they
wer
e us
ed.
bT
he fi
rst v
alue
of F
is c
ompu
ted
from
sam
ple
data
and
the
seco
nd F
val
ue is
from
Sne
deco
r's F
tabl
e
with
deg
rees
of f
reed
om 6
and
230
for
the
first
thre
e te
sts
and
10 a
nd 1
9 fo
rth
e fo
urth
test
.
"1,1
tibt
_
SECTION 2 37
measure of the influence of specified variables on achievement re-sults. The results from analysis of covariance support the observa-tions from regression analysis that the mean achievement level atthe beginning of Title I programs is negatively correlated with changein achievement.
The basic unit of measurement of achievement in Phase I is thegrade within a specific scho91. The only lower level of aggregationthat could be used is the individual student. Results indicate thatmajor differences in changes of achievement are among grades withina school and not among schools. If the several grades within a schoolare grouped together, possible sources of differences in relationsbetween CE and enhanced achievement may be hidden by the averag-ing process.
Further insight into variation in observed changes in achievementcan be obtained from examination of Figure 8. This shows the changein achievement test scores, at the first decile and mean respectively,for the grades and schools that comprise the District 13 sample.For example, in school "1" the mean achievement score in the 6thgrade in 1966-67 was 2 Standard T-scores higher than the correspond-ing 6th grade score in 1965-66. The score at the first decile in 1966-67
1 2 3 4 5 6 7 8 9 10 11SCHOOL
.86
U J ...-6 *6.....
CI .1I
6 21,1II1
2
0Z 1 .2
5
(1.4.44'5
1Z i5e51 .-4u
.-6
.4
s 28
.1111
.5
1 2 3 4 5 6 "7 8 9 10 11SCHOOL
Figure 8. Changes in test scores in District 13 grouped by school.
38 671MP-115
was one standard T- score lower than the corresponding score in1965-66. These results are representative of the variation withinand among schools in most school districts. In almost every schoolthere are both positive and negative changes spread over a substantialrange of year-to-year variation. Similar situations exist for otherdistricts for both the change in mean and change at the first decile.This could be caused by sampling variation or it could be caused bydifferences in type and amount of CE.
What school, pupil, and environmental characteristics are associated
with enhanced pupil performance?
The first part of this section presents a brief summary of the dif-ferent analyses and presents tables of est. nated regression coef-ficients and simple correlation coefficients. The next part discussesthe statistical results relevant to each of several state variables.The last part of the section presents general conclusions concerningthe relationship among state variables and changes in achievement.
The following four types of analyses were carried out in an attemptto identify the relationships between state variables and changes inachievement:
a. Simple correlations between changes in achievement andstate variaLdes were computed from the combined data on alldistricts.b. Simple correlations between changes in achievement andstate variables were computed from data on eacn of the 11 dis-tricts.c. Analysis of variance was computed for A X, D1, and AQ1,using all 314 observations. The total group was partitionedaccording to contrasting levels of state variables to estimatethe effects of these factors as reflected by the three measures.
d. Multiple regression analyses were done using a number ofstate variables for which data were available.
The results from each of the above analyses are addressed in thediscussion on each of the state variables in the sub-sections below.That is, conclusions on the relationships between each state variableand changes in achievement are based on statistical results fromeach of the above analyses.
_ mimemp161ffkAgt,-
SECTION 2 39
Table 11 presents a summary of selected state and allocation var-iables by district. The simple correlation coefficients presented inTables 12 and 13 show the degree of association between two variables.These coefficients were computed under the assumption that any re-lationship between two variables that might exist is linear. Forexample, the coefficient of - 0.10 between percent Negro and changein the mean (see Table 12) reflects only a small degree of associa-tion between these two variables. It is possible that the numericalvalue of the estimated coefficient is low because the relationship be-tween the two variables is non-linear. It is also possible that it islow because other variables are correlated with both percent Negroand change in achievement.
Correlation coefficients computed from the combined data for alldistricts are shown in Table 12. Similar correlation coefficientswere computed from data within each of the 11 school districts. Only
one of these latter sets is presented in tabular form, namely, thosefor District 13 in Table 13. The correlations from the other 10 dis-tricts have been extracted and used to illustrate discussion of spe-cific variables below.
Analysis of variance is helpful in identifying variables that areassociated with changes in achievement in that it can be used for de-tecting significant diffe.eimces among specified sets of results. Itwas used, for example, to determine if the change in achievementfor schools with high percent Negro wer ntly different fromchanges in schools with low percent Negro (see Table 18). Analysisof variance does not require the assumption of a linear relationshipamong the two variables of inte-rest. It is also applicable for testingfor significant differences among groups that cannot be described innumerical terms. For example, it would not be meaningful to com-pute a correlation coefficient between change in achievement anddistrict but it is possible to use analysis of variance for testing dif-ferences among distri-.:ts (see Table 48 ). Thus, analysis of varianceis applicable in situations where simple correlation coefficients arenot applicable.
Multiple regression analysis is helpful for identifying the effect ofeach state variable on changes in achievement and it is helpful in re-ducing the sampling variation when testing for significant effect fromCE. This technique is especially helpful when the variable of interest(change in achievement in this case) has been affected by many differentvariables. In the model described in Equation 2 (see page 43),
Tab
le 1
1. S
umm
ary
of s
elec
ted
stat
e an
d al
loca
tion
varia
bles
by
dist
rict.
Dis
tric
tM
ean
Tes
t Sco
re19
65-6
6c1!
Tot
al T
itle
I Fun
ds__
_($
mill
list
hlT
itle
I $ 1
Tor
NO
1965
-671
5
Effe
ctiv
eT
itle
I $pe
r P
upi l
c
Mob
il ity
(per
cent
a )
Per
cent
Neg
ro19
65-6
60
Per
cent
Spa
nish
1965
-66a
Mea
nG
rade
AD
M°
1965
-66
1966
-67
138
.22.
61.
567
6248
61d
150
244
.43.
53.
332
2151
6412
124
345
.70.
70.
894
51d
dd
115
442
.712
.010
.210
0'6
449
586
51
542
.01.
91.
669
64d
600
80
643
.21.
21.
215
814
016
440
60
844
.01.
21.
256
5342
510
152
1035
.13.
64.
111
360
:792
178
1243
.50.
20.
730
010
519
750
77
1341
.02.
62.
217
213
760
4620
129
1440
.71.
51.
387
5533
560
80
All
Dis
tric
ts42
.431
.028
.11
8275
I47
567
120
NO
TE
S:
a A
vera
ge a
cros
s al
l obs
erva
tions
in e
ach
dist
rict.
Mob
ility
is c
ompu
ted
asnu
mbe
r of
pup
ils m
ovin
g in
to p
lus
num
ber
mov
ing
out o
f a s
choo
l, ex
pres
sed
as a
per
cent
of s
choo
l AD
M.
b D
olla
rspz
r pu
pil c
ompu
ted
by s
umm
ing
tota
l Titl
e I f
unds
for
both
yea
rs a
nd d
ivid
ing
by d
istr
ict A
DM
inT
itle
Isc
hool
s.c
Dol
lars
per
pup
il co
mpu
ted
as th
e su
m o
fthe
ave
rage
1965
-66
Titl
e !d
olla
rs p
er p
upil
plus
the
frac
tion
of th
eav
erag
e 19
66-6
7 T
itle
Ido
llars
per
pup
ilco
rres
pond
ing
to w
hen
the
post
yea
r te
st w
cs a
dmin
iste
red.
° D
ata
not a
vaila
ble.
Tab
le 1
2. C
orre
latio
ns b
etw
een
sele
cted
var
iabl
es, c
ompu
ted
from
com
bine
dda
ta o
n al
ldi
stric
ts.°
cn
rncl 1
Var
iabl
eD
istr
ict
47
89
1011
1213
1 2 3 4 5 6 7 8 9 10 11 12 13
Cha
nge
in M
ean
Cha
nge
in 1
st D
ecile
Cha
nge
in 1
st Q
uart
ileC
hang
e in
3rd
Qua
rtile
Mea
n P
re A
chie
vem
ent
Sco
reIn
dex
of M
obili
tyG
rade
Atte
ndan
ce R
ate
(196
5-66
)C
hang
e in
Atte
ndan
ceR
ate
Per
cent
Neg
ro (
1965
-66)
Cha
nge
in P
erce
nt N
egro
Pre
Tes
t Dat
eP
ost T
est D
ate
Titl
e I $
Per
Stu
dent
1.0
0.64
1.0
0.77
0.59
1.0
0.85
0.39
0.50
1.0
- 0.
43-
0.19
- 0.
32-
0.35
1.0
0.03
0.00
0.01
0.04
-0.1
41.
0
0.05
0.08
0.03
0.07
0.35
0.01
1.0
0.02
0.09
0.03
-0.0
3
- 0.
15-
0.15
- 0.
30
1.0
-0.1
0-
0.07
-0.1
4-0
.10
-0.3
0-
0.47
-0.1
6
0.22
1.0
- 0.
08-0
,05
- 0.
06-
0.08
0.00
- 0.
05
0.06
-0.0
60.
001.
0
0.23
0.03
0.10
0.28
-0.1
70.
03
0.14
-0.2
2-0
.13
0.06
1.0
0.16
0.00
0.09
0.20
-0.0
9-
0 .1
0
0.17
-0.2
0-0
.14
01,0
5
0095
1,0
0.30
0.13
0.19
0.31
-0.2
30.
03
0.25
-0.0
70.
130.
030.
390.
341.
0
NO
TE
:a
With
314
obs
erva
tions
an
abso
lute
valu
e of
0.1
5 fo
r th
e co
rrel
atio
n co
effic
ient
is s
igni
fican
t at t
he 1
per
cent
leve
l
and
0.11
is s
igni
fican
t at t
he 5
per
cent
leve
l. W
ith 2
40 o
bser
vatio
ns th
e co
rres
pond
ing
valu
es a
re 0
.16
and
0.12
.
ICor
rela
tions
with
var
iabl
es 7
, 8, 9
, mei
10
wer
eco
mpu
ted
from
240
obs
erva
tions
in th
e7
dist
ricts
for
whi
ch d
ata
wer
e av
aila
ble
onth
ese
varia
bles
.
Tab
le 1
3. C
orre
latio
ns b
etw
een
sele
cted
var
iabl
es,
com
pute
d fr
om d
ata
for
Dis
tric
t13
.a
Var
iabl
eD
istr
ict
.
12
34
56
78
910
1112
13-
1C
hang
e in
Mea
n1.
0 0.
660.72 0.92 -0.73
0.05
-0.19
0.03
0.09
-0.0
40.
400.
210.
21
2 C
hang
e in
1st
Dec
ile1.
00.
52 0
.46
-0.4
30.
09-0
.16
0.20
0.01
0.02
0.09
-0.0
3-0
.03
3 C
hang
e in
1st
Qua
rtile
1.0
0.53
-0.
61-0
.01
-0.0
20.
06-0
.17
- 0.
0101
5-0
.01
-0.0
14
Cha
nge
in 3
rd Q
uart
ile1.
0-0
.70
0.12
-0.2
5--
0.05
0.21
- 0.
080.
460.
290.29
5A
vera
ge A
chie
vem
ent
Leve
l1.
00.
080.
270.
05-0
.10
0.05
-0.5
7-0
.32
-0.3
26
Inde
x of
Mob
ility
1.0
-0.4
50.
38-n
.20
0.05
0.26
0.20
0.20
7 G
rade
Atte
ndan
ce R
ate
1.0
-0.27
-0.46
0.21
-0.1
8-0
.12
-0.1
28
Cha
nge
in A
ttend
ance
Rat
e1.
0-0
.13
0.19
0.06
0.04
0.04
9 P
erce
nt N
egro
1.0
0.28
-0:0
4-0
04-0
.04
10 C
hang
e in
Per
cent
Neg
ro1.
0-0
.01
0.00
0 00
11P
re T
est D
ate
1.0
0.94
0.94
12
Pos
t Tes
t Dat
e1.
01.
0
13 $
per
Stu
dent
1.0
NO
TE
:a
With
32
obse
rvat
ions
an
abso
lute
val
ue o
f 0.4
4 fo
rthe
corr
elat
ion
coef
ficie
nt iJ
sig
nfic
ant a
t the
1 p
erce
nt le
vel
and
an a
bsol
ute
valu
e of
0.3
4 is
sig
nific
ant a
tth
e 5
perc
ent l
evel
.
SECTION 2 43
there are five different variables that are postulated to be significantin explaining observed change in achievement.
The model for estimating regression equations using data fromseveral school districts was
where
Y
$
Y = a + al ($) + a2
(ivi) + a3
(%I1) + a4
(L) 4 a5
(A) 4- c ,0
= change in achievement level (WC or AD 1)
= average Title I CE program funds per student expended bythe post test date 27.0 2' all students in Title I schools in theschool district
L = achievement level at the beginning of Title I programs
M = mobility in and out of the school
%N= percent Negro in a school (1965-66)
A = attendance rate ia a school (1965-66)
E = an error component in the postulated relation.
Descriptions of the way each of these variables were measured aregiven in Appendix B. The regression equations were estimatedwithout the corrections indicated in Table 59.
If the determining variables in the regression equation are inde-pendent each estimated regression coefficient is an estimate of theeffect of a one unit change in the determining variable on the depen.:dent variable. For example, an estimate of a2 in the above equationwould be an estimate of the effect of a one percent increase in mo-bility rate on change in achievement. A regression coefficient ismore meaningful than a simple correlation because the latter oftenreflects spurious correlation caused by the fact that both variablesare related to a third variable. Although the determining variablesmight not be independent, the regression coefficient for each variableis another measure of the relationship between that variable and thedependent variable.
The estimated regression coefficients using change in the meanas dependent variable are shown in Table 14. The estimated coef-ficients using change in achievement at the first decile are shown in
44 67TMP-115
Table 14. Estimated regression coerneents relating change in mean achievement
to selected variables.°
DistrictConstant
Termhit-bill:-
Rate'FercentNegroc
Mean
Achieve-ment
Scorec
Attend-onceRatec
CE Fundingper Student
(Title I)
2R- I
Pool 0.73 - 0.003 - 0.018 - 0.387 0.175 0.0027 0,37(0.006) (0.004) (0.039) (0.048) (0.0040)
1 38.5 - 0.075 - 0.039 - 0,91 0.027 0.60(0.021) (0.009) (0.136 (0.064)
2 3.21 - 0.003 - 0.034 - 0.628 - 0.343 0,30(0.011) (0.011) (0,141) (0.250)
3 - 39.0 d 0.028 - 0.142 0.482 0.09
i (0.036) (0.132) (0.319)
4 24.0 2 0.004 - 0.306 - 0.136 0.39(0.009) (0.143) (0.118)
5 8.11 d - 0.003 - 0.193 d 0.10(0.007) (0.116)
.
6-12 19.3 d 0.014 - 0.443 d 0.529mbine (0.002) (0.271)
8 7.54 - 0.007 - 0.019 - 0.614 0.219 0.44(0.016) (0.009) (0.102) (0.118)
10 - 59.1 - 0.048 0.188 - 0.518 0.701 0.87(0.022) (0.050) (0.091) (0.182)
13 0.216 0.005 0.004 - 0.524 0.232 0.55(0.026) (0.015) (0.094) (0.771)
14 51.3 0.030 - 0.016 - 0.213 - 0.459 0.33(0.065) (0.018) (0.123) (0.504)
NOTES:a Standard error values are shown in parentheses beneath the respective esti-
mated regression coefficients.I) ti r00
1
a 11r means Districts 1, 2, 4, 8, 10, 13, and 14 taken together.° Data are for 1965-66.d Data not available.e Not included in final regression equation because F value in the final test
for significant reduction of residual variance was less than 0.005.
ACTION 2 45
Table 15. The R2 statistic in the last column is an estimate of theportion of the variation in the dependent variable that can be explainedby the determining variables included in the equation.
Each of tIle variables that might have affected changes in achieve-ment between 1965-66 and 1966-67 will be discussed in a separatesub-section. The results from each type of statistical analysis arenot applicable to discussion on each variable and, therefore, are notreferred to in each sub-section.
1
GRADE. There is reason to expect that grade level would be as-sociated with benefit from compensatory education. There may becritical periods during whtch compensation for educational depriva-tion is relatively easy, ani others when it is very aifficult. Thus,it may be that if educationally deprived children are given remedia-tion soon enough they may be able to overcome most if not all of theirdisadvantage. This concept is consistent with the opinion expressedin an earlier report on the first year of Title I (Reference 3, p 39).
In this study, the effect of grade level was examined by an analy-sis of variance, The analysis of variance showed no significant dif-ferences in average change among grade levels.
The simple correlations between changes in achievement andgrade level are shown in Table 16. Although most of the correla-tions are negative, the results are not statistically significant.
The average achievement level in 1965-66 and the average foreach measure of change in achievement between 1965-66 and 1966-67are shown in Table 17. This table also shows the relationship be-tween Standard T- score, national percentile, and grade equivalence.
For example, the average T- score for the 78 observations ongrade 6 was 42.6. This average corresponds te the 23rd nationalpercentile. A student obtaining a Standard T-score of 42.6 at themid-point in the 6th grade would have a grade equivalence score of5,3 compared to the national average of 6.5. A student with a scoreof 5.3 would be considered 1.2 years behind in achievement level.
The results in Table 17 show no apparent relationship betweengrade level and changes in achievement. They also show no apparentrelationship between amount of educational deprivation and gradelevel.
67TMP-115
Table 15. Estimated regression coefficients relating change in achievement at the
first decile to selected variables.°
DistrictConstant
Terms
MobilityRatec
PercentNegroc
I Mean'Achieve-i ment
nd-AtteCEonceStudentRate
EffectiveFunding
(Title I)
Pool 0.29 - 0.018 .- 0.020 - 0.330 0.210 0.0063 0.13(0.010) (0.007? (0.061) 10.074) (0.0063) '
1 4.62 - 0.031 - 0.021 - 0.175 0.156 0.16(0.028) (0.0;2) (0.186) (0.087)
2 1.29 - 0.020 - 0.034 - 0.662 0.212 0.22(0.019) (0.015) (0.189) (0.334)
3 - 77.1 d 0.043 e 0.826 0.09(0.052) (0.465)
4 41.7 - 0.038 - 0.061 1.24 0.183 0.25(0.050) (0.032) (0.466) (0.430)
5 3.51 d - 0.007 - 0.090 d 0.02(0.013) (0.226'i
6, 12 48.2 d 0.001 1.06d 0.47
combined (0.005) (0.607)
8 22.2 - 0.051 - 0.041 - 0.672 0.128 0.31
(0.025) (0.014) (0.158) (0.183)
10 - 45.5 0.113 0.156 - 0.170 0.477 0.45
(0.055) (0.123) (0.225) (0.447)
13 50.0 0.006 - 0.008 0.458 0.310 0.21
(0.052) (0.030) (0.187) (1.54)
14 106 - 0.120 0.009 - 0.037 1.07 0.17
1
(0.106) (0.030) (0.209) (0.823)
NOTES:a Standard error values are shown in parentheses beneath the respective esti-
mated regression coefficients.b "Pool" means Districts 1, 2, 4, 8, 10, 13 and 14 taken together.
c Data are for 1965-66 only.d Data not available.e Not included in final regression equation because F value in the final test for
significant reduction of residual variance was less than 0.005.
SECTION 2
Table 16. Correlations between grade level and changes inachievement scores.'
r
DistrictChange of Achievement
AIX,i
A D,i
AQ,1
AQ,a
Number '1
of Grades j
1 - 0.15 - 0.23 - 0.20 - 0.05 39
2 - 0.17 - 0.12 - 0.12 - 0.09 56
3 - 0.07 - 0.11 - 0.12 - 0.03 38
4 0.00 0.05 - 0.18 0.06 27
5 0.01 - 0. i2 - 0.16 0.02 28
6 0.55 0.06 0,77 0.93 4
8 - 0.16 0.10 - 0.14 - 0.18 55
10 - 0.28 - 0.09 - 0.28 - 0.14 13
12 - - - - -13 - 0.17 0.01 - 0.02 - 0.26 32
14 - 0.22 - 0.04 - 0.20 - 0.16 18
All Districts - 0.13 - 0.08 - 0.14 - 0.08 314
NOTE:° Correlations were computed under the assumption that the relation between
change in achievement and grade (if any) is a linear relation betweenchange in achievement and the familiar numerical identification of grades(i.e., grade 1, grade 2, etc.). Although some of the individual correla-tions are significant at the 5 percent level the percer.tage of the total thatare significant is approximately what would be expected by chance alone.
Tab
le 1
7. S
umm
ary
of a
chie
vem
ent l
evel
196
5-66
and
cha
nges
inac
hiev
emen
t bet
wee
n 19
65-6
6 an
d 19
66-6
7, b
y
grad
e.
Ave
rage
Ach
ieve
men
t Lev
el19
65-6
6A
vera
ge C
hang
eN
o. o
fO
bser
vatio
nsG
rade
Sta
ndar
dT
-sco
reN
atio
nal
Per
cent
ileG
rade
Equ
iv. °
A R
A D
1A
Q1
A Q
3
139
.1.
141.
3+
1.9
7+
2.2
7+
0.9
9+
3.01
7
21
44.0
282.
2-
0.33
- 0.
22+
0.2
4-
0.97
26
340
.918
2.6
+0.
16+
1.0
6+
0.08
-0.
3442
442
.723
3.5
4:Q
.06
+ 0
.53
+ 0
.28
-0.
0979
541
.620
4.3
- 1.
01-
0.56
- 1.
24-
1.07
47
642
.623
5.3
- 0.
54+
0.0
5-
0.75
l-
0.60
78
740
.718
5.8
- 0.
66-
0.04
- 0.
20-
1.26
7
842
.825
7.0
- 0.
40+
0,1
1-
0.38
- 0.
5320
1045
b+
0,5
00.
00.
00.
c1
11
.48.
846
.235
b-
1.23
- 0.
18-
1.40
-1.
426
1251
.255
b+
0.9
0+
1.3
00.
0+
1.3
01
NO
TE
S:
a G
rade
equ
ival
ence
was
com
pute
d fr
omta
bles
pro
vide
d by
the
MA
T a
nd 1
TB
S T
ests
and
is b
ased
on
mid
-yea
rev
alua
tion.
For
exa
mpl
e, th
e gr
ade
equi
vale
nce
of 1
.3 fo
r th
e fir
st g
rade
is0.
2 be
low
the
1.5
leve
l whi
chis
the
aver
age
for
mid
-yea
r fir
st g
rade
s in
the
natio
n.b
Rib
lishe
rs' c
onve
rsio
n ta
bles
for
com
putin
g gr
ade
equi
vale
nce
wer
e no
t ava
ilabl
e to
the
cont
ract
or.
I.
SECTION 2 e, 49
The possible effect of grade level was studied in the regressionanalysis by relating the error component (i. e., E in Equation 2) tograde level. This was done in an attempt to determine if the changesin achievement that could not be explained by the variables includedin the regression equation could be explained by differences in gradelevel. There was no evidence that changes in achievement were re.lated to grade level.
POVERTY. Oi ..e. of the criteria for inclusion in Title I is low in-come level of the pupils' families. Although allocation of Title I dol-lars to school districts was based on number of pupils from familieswith income below $2, 000 the criterion of low income within a schooldistrict and, therefore, qualification of the school for Title I fundsvaries throughout the country. Our classification of poverty for eachschool was the criterion used by each of the respective school dis-tricts.
Schools in the sample were classified as high, medium or lowpoverty relative to all Title I schools in that district. In view of thepurposes of Title I funds more money might be expected to be allo-...;ated to schools with the most poverty.
'The analysis of variance and the simple correlations betweenpoverty and changes in achievement do not show any reliable relation-ships between poverty level and changes in achievement test scores.
MOBILITY. High student mobility is a condition likely to dilutethe effects of Title I. It is unlikely that students who transfer toanother school will have the same type of CE program.
High mobility makes it difficult to relate Title I dollars withchanges in achievement because many of the students taking whatwe have called the "post test" might have been exposed to very littleof the CE programs in the school in which he is tested.
The results from sample data give inconclusive evidence on theeffects of mobility. The regression coefficients for this variableare generally negative but only some of them are statistically signi-ficant (see Tables 14 and 15).
The analyses of variance showed no significant differences inchanges in achievement between high and low mobility rates. Thecorrelation coefficients in Tables 12 and 13 do not show a negative
50 GTMP-115
relationship. Since the regression and correlation coefficients haveopposite signs there is reason to believe that mobility is directlyrelated to other variables which have a positive effect on chan&es inachievement. For example, it is possible that more Title 1 dollarswould be allocated to schools with high.mobility because they arelikely to have a high percentage of pupils from families with low in-come.
The effects of high mobility should be investigated further inPhase II. It will then be possible to use analysis of covariance whichis a more powerful analytical tool than used thus far since it makesadjustments for covariates.
PERCENT NEGRO. The relationship in change in achievement topercent Negro was examined in light of the topical interest in Negroeducational problems. The analysis below suggests a strong rela-tionship between percent Negro enrollment and changes in achieve-ment. Five levels of percent Negro were defined for analysis ofvariance: 0 to 19 percent, 20 to 39 percent, 40 to 59 percent, 60 to79 percent, and 80 to 100 percent.
The results in Table 18 show that the 0 to 19 percent group re-sponded best, while the 40 to 59 percent group responded the worst.The positive change for the 0 to 19 percent group is statisticallysignificant at the 10 pezeent level. The LieLaye change in the meanand first quartile for the 40 to 59 percent group are statistically sig-nificant at the five percent level.
The non-linear relationship between percent Negro and change inachievement indicated by the results in Table 18 explains why somany of the simple correlatioia coefficients shown in Table 19 aresmall. There does appear to be a significant linear relationshipbetween percent Negro and absolute achievement level but not a sig-nificant linear relationship between percent Negro and change inachievement.
The effect of these levels on changes in achievement (as measuredby A X, A D and A Q1 ) was studied in one-way analyses of varianceusing 277 observations and in a two-way analysis of variance using
SECTION 2
Table 18. Average of changes in achievement scores, by percentNegro in school .
5 1
Number of Cases
Change in Mean:
Mean change in b43Z
Standard errorSignificance level'
Change in First Quartiie:
Mean change in baiStandard errortalSignificance levelo
Change in First Decile:
Mean change in AD.,Standard error ADiSignificance levela
Percent Negro in school
0-- 1 9 20-39 40-59 60-79
56 34 45 35
0.5 - 0.4 0.9 - 0.40.29 0.45 0.41 0.360.10 0.40 0.05 0.30
0.9 - 0.02 1.4 - 0.50.40 0.50 0.39 0.460,05 b 0.01 0.30
1.1 - 0.1 0.3 - 0.00.38 0.68 0.54 0.480.01 b b b
I 80-100
108
- 0.20.210.40
0.30.240.30
0.40.30b
NOTES:a Probability of obtaining a value as large or larger than the sample mean by
chance if the expected change in achievement score between 1965-66 and1966-67 were zero.
b Greater than 0.50 percent.
80 observations.* The analysis of variance results are shown inTable 20. The results using the first decile are not statisticallysignificant but several of the results using the mean and first quar-tile are significant. The difference among the five groups arestatistically significant. The difference between the 0 to 19 percentgroup and the 40 to 59 percent and between the 0 to 19 percent groupand the 20 to 100 percent group are both statistically significant.
* Ideally, it would oe preferable to handle all of the factors simul-taneously in a multiplz factor analysis of variance. However, thiswas not feasible; there ere numerous cells with missing data, andthe resulting partial layout was not balanted.
52 67TMP-115
Table 19. Correlations between percent Negro in a school andachievement scores.°
I District No.ofObs.
InitialAchievement
LevelAMean AD1 AQ1 AQ3
1 39 -0.20 -0. 20 -0.21 -O. 20 -0. 12
bI
2 56 -0.46 -0.17 -0.04 -0.25 -0.21
4 27 -0.93 0.00 -0.04 -0.19 0.06
5 28 0.05 -0.08 -0.11 0,04 0.20
6 4 -0.89b 0.76 0.60 0.45 0. 07
8 55 -0.33b -0.13 -0.11 0.07 -0.15
10 13 -0.38 0.47 0.34 0.41 0.35
12 4 -0.64 0.49 0.77 0.68. -0.55
13 32 -0.10 O. 09 O. 01 -O. 17 0.21
14 18 -0.34 -0.14 -0.19 -0.5513 0.11
NOTES:aPercent Negro for District 3 were not available for 1965-66.
bSignificant to 5 percent level.
It is difficult to specify the i recise cause as towhy changes inachievement are related to per,..ent Negro. However, the relation-ship is strong enough tha+ the topic deserves further analysis whenmore data are available on the allocation of Title I funds for CE.
Thee, ",..,,considerable variation among school districts in percent
Negro students, therefore it might be thought that the above resultsreflect simply differences among districts. To control for the effectsof district, a two-way analysis of variance was performed using onlytwo levels of Negro composition and the eight districts which met
/
SECTION 2
Table 20. Significance levels of differences in mean changeamong groups based on percent Negro.°
Contrasted Groupsb AR AD
1AQ
1
1. All 5 Groups 5% 25% 0.5%
2. Group I and Groups 11-V (pooled) 10% 50% 5%
3. Group I and Group III 10% 50% 5%
4. Group I and Group II 50% c 50%
5. Group III and Group V c c
6. Groups 1-11 (pooled) and c c 50%
Groups III-V (pooled)
Notes:aPercent Negro in the pre-year was used for the grouping: Group "I,
0 to 19 perc:nt Negro; Group 11, 20 to 39 percent Negro; Group III40 to 59 percent Negro; Group IV, 60 to 79 percent Negro; Group V,80 to 100 percent Negro.
bBased on the S-Msthod of multiple comparisons (see Reference 15).cGreater than 50 percent.
the data requirements. Because we wished to test whether the ob-served difference among levels of percent Negro was maintainedwhen district was controlled, only the categories of percent Negrowhich differed the most were compared. For each district havingsuch data, the five observations with the lowest percent Negro wereselected in each of the contrasting categories of 0 to 19 percent Negyoang 40 to 59 percent Negro. In both changes in the mean and changesin the lower quartile the results are highly significant. That is, thetwo levels of percent Negro have significantly different means, re-inforcing the earlier results. The non-significant results for the"interaction" effect indicate the relationship between achievementand percent Negro is similar within each of the eight districts.
54 671MP-115
Change in Percent Negro
Just as school racial composition correlates with achievementtest score level (e. g., correlation of - 0.30 in Table 12), so changesin racial composition may be expected to correlate with changes inachievement level.
The correlations between change in percent Negro (post versuspre) and changes in achievement test level are shown in Table 21.Only a few of the correlations within particular districts are solarge as to be statistically reliable and these do nor present a con-sistent picture. To look more intensively into the variable of changein racial composition, a second type of examination was done. Therewere 53 school-grade observations from 26 schools for which thechange in percent Negro in the school was 5 percent or more. Ofthese, 41 increased in percent Negro and 12 decreased. The changesin mean and decile achievement test scores for these observationsare shown in Table 22. Relatively large changes in percent Negroin either direction are associated with less favorable changes inachievement scores.
PERCENT SPANISH. There are several school districts whichhave a sizeable proportion of students of Spanish extraction. Theanalysis did not reveal any significant relationships between percentSpanish and changes in achievement. However, the analysis is notdecisive. The statistical tests were based on a small sample andthere was little variation in the percentages of pupils from Spanishspeaking families within each of the school districts.
MEAN READING ACHIEVEMENT LEVEL AT THE BEGINNINGOF TITLE I PROGRAMS. Compensatory education is aimed primar-ily at students who have suffered from cultural disadvantages andcan, therefore, be expected to have relatively low achievement testscores. It might be expected that the emphasis of the CE programs(particularly remediation programs) would be toward students withlower achievement test scores rather than those with highest scores.Thus, the nature of the program would suggest that a negative cor-relation might exist between change in achievement level and originalachievement level. It appears that.this is the case. However, inter-pretation of the meaning of the correlation must be qualified by therealization that there is a built-in correlation due to statistical re-gression between original levels on a variable and changes in thatvariable from one time to another. That is, on any given measure-ment of a variable "errors of measuremeat" are reflected in extreme
c
-
SECTION 2
Table 21. Correlations between change in percent Negro in a school
and achievement scores.°
II District
-No.of
Obs.
InitialAchievement
Score
-A X
IA Di
1
AQ3
1
2
4
5
6
8
10
12
13
14
39
56
27
28
4
55
13
4
32
18
b-0.34
0.30b
0.34
-0.01
60.89
0.05
0.17
-0.27
0.05
-0.03
0.11
-0.05
-0.12
-0.31
-0.76
-0.07
-0.40
-0.03
-0.04
-0.05
-0.03
0.01
-0.02
-0.45b
-0.60
-0.03
-0.22
0.13
0.02
-0.43
0.13
-0.04
0.23
-0.28
-0.45
0.00
-0.44
0.20
-0.01
-0.38
0.04
-0.01
-0.42b
-0.20
-0.07
-0.16
-0.20
-0.50
-0.08
0.06
Notes:aPercent Negro for District 3 were not available for 1965-66.
bStatistically significant to the 5 percent level.
_
67TMP-115
Table 22. Average change in achievement test scores, by typeof change in racial composition.
Observations NAveragea Mean
JAverageA Decile
Large increase (5%) in percent Negro 41 -0.8 -0.4
Little change in percent Negro 219 -0.2 +0.5
Large decrease (25%) in percent Negro 12 -0.6 -1.2
TOTA L 272 -0.34 +0.31
scores and a portion of these will "regress" toward Ihe mean onlater testing (Reference 14, pp 321-324). This phenomenon is main-ly a problem in evaluating the correlation between LII.Z and the premean a. e., the mean for 1965-66h Errors in measurements of
D1, Qi and Q3 for 1965-66 would not likely be highly correlated with-in measurement of X in 1965-66 and, therefore, the spurious* corre-lation between changes in D1, Q1 and Q3 and the pre mean is likelyto be small.
The average initial reading achievement level was a highly sig-nificant variable in the regression analysis. The regression coef-ficients are negative in all twenty-two regression equations shownin Tables 14 and 15. This result is in line with the negative cor-relations shown in Tables 12 and 13. The regression coefficients onthe pre mean (r) in Table 15 are affected by the spurious correlationdiscussed above.* Regression coefficients and simple correlationcoefficients both indicate that this variable is more highly correlatedwith changes in achievement at the first decile than any other statevariable analyzed during Phase I (e.g., see row 2 in Tables 12 and13).
*It is not possible to develop a. precise estimate of the effect of thespurious correlation from data that are available. A technical dis-cussion of this problem is presented in the last part of Appendix C.
SECTION 2
..................
57
These results suggest that the relationship between initial readinglevel and expected changes from CE should be investigated furtherin Phase II. As with all statistical results, the researcher must becareful in drawing firm conclusions until he has analyzed possiblereasons for spurious correlations.
MEAN ATTENDANCE RATE IN SCHOOL PRIOR TO CE. Theschool attendance rate in 1965-66 has a small positive correlationwith AR, b D1, A6Q1 and 6 Q3 when computed from the combined data
for all school districts (see Table 12). However, the same corre-lation computed from data within each of the school districts shows
both negative and positive correlations (e.g., see correlation forDistrict 13 in Table 13). The regression coefficients are generallypositive but only a few are significantly different from zero.
A positive correlation between attendance rate and enhancementwould suggest that it might be well to orient CE programs towardsincreasing attendance. The returns per dollar of Title I might behigher than if all money is oriented towards academic skills. How-ever, the correlations observed in the Phase I analyses might betested in follow-on work before drawing any firm conclusion con-cerning attendance and CE programs.
Change in Attendance Rate
Attendance is another type of pupil performance often regardedas a possible criterion for educational programs. It is assumedto vary with children's motivation to learn and to be a prerequisiteto benefiting from school work.
To date within our study no control for other factors, such asweather or recording procedures which might affect attendance mea-sures has been used. Further, districts differ fairly vnarkedly indefinitions of attendance. Therefore, the present mea.ures should
be regarded as fairly crude, especially for inter-district calculations.
The correlations between change in attendance figures between
year s 1965-66 and 1966-67 and change in achievement statistic sduring these same years are shown in Table 23. Most of the corre-lations appear low and only two are statistically reliable to the 5percent level. There does not appear to be any clear relationshipbetween change in attendance rate and change in achievement testscores.
58
...1.*.,..........61.
67TMP-115
Table 23. Correlations between change in attendance mite and changesin achievement scores.
Districts A7 I A n1
I AQ.I
I AQJ
I Number I
1 0.21 0.34° - 0.04 0.17 39
2 0.05 0.06 -0.05 -0.01 56
3 -0.14 -0.26 -0.25 -0.06 38
4 0.06 -0.21 0.00 0.13 27
10 0.19 0.46 0.10 -0.01 13
13 0.03 0.20 0.06 -0.05 32
14 0.18 0.48° 0(.21 0.19 18
Note:
°Statistically reliable to the 5 percent level.
SIZE OF GRADE AND SIZE OF SCHOOL. The statistical evidenceshows no significant relation between change's in achievement andnumber of pupils in the grade or number of pupils in the school.There is little evidence to encourage further investigation of thisvariable in follow-on works.
SUMMARY OF ANALYSIS OF SPECIFIC STATE VARIABLES.This study presents statistical evidence to support a priori reasoningthat certain environmental conditions have an important effect onthe results of compensatory education. This suggests that CE pro-grams should be designed in light of environmental conditions. Timein Phase I did not permit analysis of possible trends in achievementat inner-city schools. Analysis of specific environmental variablessuch as grade, poverty and mobility would be more meaningful if thetrend factor could be studied at the same time.
In this sample statistically significant differences were associatedwith percent Negro and achievement level at the beginning of compen-satory education. The analysis did not reveal any significant asso-ciation between the other state variables and enhancement in achieve-ment.
SECTI ON 2 59
The differences in average changes between 1965-66 and 1966-67in the various districts were statistically significant. It is not pos-sible to say if these differences are due to trends, differences inpupil/school characteristics, or fliff-r-ncc,s in rE3 rIrrI"nrrIcz,
The percentage of the variation in changes in achievement that canbe explained bx the variables included in the regression model is verysmall. The It for the regression using change in the first decile asdependent variable and computed from data for seven districts is only0.13 (see Table 15). That is, 0.14 is the best estimate of the percentof the variation in 6 D1 that is explained by the five variablesmobility rate, percent Negro, initial achievement level, attendancerate, and "effective Title I dollars per student."
There are many state variables other than those studied here whichmay have a relationship to enhancement in achievement. Future ef-fort in this area would do well to consider these additional variablessuch as changes in proportions of classes given achievement tests,the form of achievement tests used, and the percent of minority grouppupils (combining Negro and Spanish-speaking) in a grade.
60
:
67TMP-115
SECTION 3
PRELIMINARY ANALYSIS IN IDENTIFICATION OFDISTINGUISHING FEATURES OF SUCCESSFUL CE PROGRAMS
The first phase of this study focused on changes in achievement
test scores between 1965-66 and 1966-67 and the relationship be-tween these changes and selected pupil-school-environmental char-acteristics. The more detailed questions concerning dist!.nguishing
features of successful CE programs were deferred. However, itwas possible to make two types of examinations of effects of CE
programs preparatory to later analyses. One of these was the in-clusion in the correlation and regression analyses of a variable in-dicating average Title I expenditures per pupil by district. The
second examination involved case studies of specific CE activitiesat particular grades in two districts undertaken as prototype studiesof estimating cost and analyzing results of CE programs. Results
of the exercise suggest how relationships between CE programs and..
enhanced pupil performance must be examined.
DISTRICT LEVEL TITLE I EXPENDITURE PER PUPIL-11 DISTRICTS
To obtain at least a crude measure of the overall level of perpupil expenditure for CE, Title I expenditure figures for each dis-trict as a whole were used. For this level of expenditure to be ap-propriate to the achievement test results, the amount of the Title Iexpenditures in 1966-67 proportionate to the length of the academic
year elapsing before the post test was assigned. The measure of
Title I dollars per student was based on (1) total Title 1 dollars ex-pended within the district for 1965-66 and 1966-67, and (2) totalnumber of pupils in. district schools receiving Title I funds. Ef-
fective Title I dollars per student for each sample observation wascomputed as the sum of average dollars per pupil in 1965-66 plus a
portion of the average dollars per pUpil in 1966-67. The portion
used for 1966-67 was the fraction of the academic year that hadelapsed up to the time of the 1966-67 test for each specific grade
unit.
SECTION 3 61
At the level of specific school districts there seems to be a con-gruence between changes in achievement test score and effectiveTitle I dollars per pupil (see Table 24). The three districts withthe highest level of Title I funding have the largest gains in achieve-ment a"..: the first decile.
The correlation coefficients in Table 12 reveal an apparent rela-tionship between gain in achievement and effective Title I expendi-tures. The correlations between "effective Title I dollars" and eachof the four measures of changes in achievement are significant atthe 5 percent le-vel.
These positive findings must be qualified, however, by threeconsiderations. (1) The correlations for the lower quartile andlowest decile i. re lower than for the mean and upper quartile. (2)Observations within a district are not fully independent, since fora given grade level in a district all tests were administered at thesame time, and the same amount of Title I expenditures will apply.Since it is known that changes in achievement scores differ signi-ficantly among districts, much of the observed relationship may bedue simply to difference among districts.* (3) The regression co-efficients for the mean and the lowest decile after correction formobility rate, percent Negro, attendance rate and mean initial a-chievement level are small relative to their standard errors. Thissuggests that the apparent relationship between district level TitleI expenditures and achievement change might be accounted for byother variables which are correlated with amount of Title I dollars.
Correlations between district level Title I expenditures and achieve-ment changes within a district (e. g. , see Table 13) have quite limitedmeaning because the only source of difference in amounts of Title Iexpenditures is difference in testing data among grddes. It is not sur-prising that these correlations were not statistically significant.
In all, this crude attempt to determine the overall relation betweenkvel of Title I funding and achievement change produced evidence sug-gestive of a positive relationship; but this evidence is based on highlyaggregated data and cannot be regarded as highly reliable. Analysesof more detailed data within particular districts are presented below.
* For this reason it is hard to establish the degrees of freedom ap-propriate for testing the statistical significance of the correlationsbetween achievement test changes and Title I funding,
62
-
Table 24. Average Title I funds per pupil and changes
in achievement.
67TMP-115
District°
_
Change in Di Change in 7 EffectiveTitle I $ per pupil
6 4.20 1.28 140
12 1.78 0.43 135
13 1.32 1. 12 137
14 0.95 0. 14 55
10 0.55 -0.21 60
8 0.26 -0.38 53
3 0.22 0.36 51
1 0.05 0. 13 62
2 0.04 -1.35 20
4 -0.68 -1. 03 64
5 -0.70 -1. 16 64
Note:
°Districts arranged by order of magnitude on observed changes in D1 .
PURPOSE OF TWO CASE STUDIES IN ALLOCATION OF
PROGRAM RESOURCES TO THE GRADE LEVEL
One of the main objectives of this study is to determine relation-
ships between changes in pupil performance and characteristics of
compensatory education (CE) programs, including resource expendi-
tures and pupil exposures. In order to develop reliable estimates of
the contribution of CE to enhanced achievement, it is necessary to
measure or estimate resource expenditures at the same level at
which pupil performance measurements (e.g., achievement and at-
tendance) are made. Pupil performance measurements are available
for grades, but CE program descriptions and financial data were
found mostly at the district level. When these kinds of information
are recorded at school and district levels of aggregation, the effects
of CE upon pupils are obscured because of the small percentage of
SECT1 ON 363
students receiving GE and because of the large variation in CE re-
sources among grades within the schools. Therefore, it is neces-sary to estimate the distribution of program resources from school
districts to schools and grades and to identify the several types of
CE programs implemented for selected grades in sample schools.
*ago...
Wide disparities exist among the sample school districts in the
amount, degree of detail, and level of aggregation of data. It was
obvious that substantial effort would be required to extract, sum-marize, standardize, and process the information, and assign values
to the variables that will be utilized in the regression analyses to be
performed during Phase II. Two of the sample school districts wereselected to determine the feasibility of assigning CE resources to
grades. This analysis should yield a better understanding of varia-
tions in types of CE programs, duration of exposure of pupils, and
specific amounts of CE resources expended in these two districts.More importantly, the analysis should provide guidance as to the
preferred method for assigning these resources to the grade level
for all sample school districts during Phase II of this study.
'Different approaches were tested in the two districts selected.These differences in method reflect substantial differences in the
data obtained. One approach, used in District 10, employed the CE
program information that is readily available at the district level
and from this attempted to identify the type of programs authorizedby school and grade and to distribute CE resources to schools and
grades within the district based on these authorizations. The second
approach, used in District 13, started at the grade level within a
school and attempted to identify the specific programs that were im-
plemented and to estimate the resource requirements to perform
these activities.
Each approach required that reasonably complete data be already
accumulated or that c.dditional data be easily obtainable by phone or
mail. A brief inventory was made of the principal types of data ob-
tained from the several school districts. Each set was evaluated and
ranked according to relative completeness of pupil performance data,
including achievement test scores and attendance records, the variety
of compensatory education programs represented, and the quantity
and quality of financial data on hand.
The first approach assumes that programs are carried out in
reasonably good agreement with the original proposal or projectdescription and that there is little variation of program character-
istics among recipient schools except as indicated in the program
64671MP-115
plan. The second approach attempts to catalog , le distribution of
program resources which actually occurred at the grade level ina school. It avoids the assumption of little variation of programsamong schools and, therefore, requires more detailed informationabout participation, staffing, costs and changes of program- in aschool in order to distinguish the CE increment from the regularschool program.
Although the second approach is the more time consuming of the
two, its use was desirable in one of the pilot analyses because ofobserved departures of programs from original proposals amongmany of the sample school districts. In many instances, this de-parture from a prescribed program resulted from an inability duringthe first year of Title I to hire staff personnel in the middle of theschecq year or to obtain equipment on short notice. The co- 'aict of
a give_ program during the following year, was more likely to be
in accord with the budget and plan of the school district.
INFORMATION REQUIRED
Regardless of the approach employed, the objective was to obtain
very detailed knowledge of programs, schools and pupils. Frag-ments of information must be assembled from many sources todevelop and verify each estimate. The types of inforrnition desiredinclude:
1. Program Descriptions objectives, activities, personnelassigned, materials and supplies used, pupils (by schools
and grades), time and duration of program, pupil participa-tion and exposure.
2. School/Pupil Characteristics ethnic composition, economic
status, relative academic level, staff composition and turn-over, special classes.
3. Attendance Records by school and grade: average daily at-.
tendance, average dailymembership, gains, losses, per ent
attendance and absence, unusual factors or events influencing
attendance.
4. Financial Records district budget and expenditure reports,expenditures by school and project for regular and special
programs, expenditures by time periods, sources of funds.
SECTION 3
5. Evlluation Reports objectives, activities, staff and pupilparticipation, project expenditures by time periods, staffand pupil performance, measurement devices and theircharacteristics.
ANALYSIS OF DISTRICT 10 CE PROGRAMS
65
This section describes the process and the wide range of infor-mation employed in allocating program expenditures to the gradelevel for District 10 sample schools. Detailed program and expen-diture data are included.
Description of the Anaiysis
This has been an effort to pursue an approach which employs pro-gram information from school district plans, reports, and records asthe basis for determining the incidence and intensity of CE activitiesand for estimating resource expenditure for individual grades in sampleschools. District10 was selected for this experimental effort becauseit was one of the few visited which records expenditures by school andbecause it was one of the few visited which records expenditures byschool and because relatively detaild program information had beenobtained. The initial step was to assemble description information oneach CE project in the set of sample schools. This involqed search ofmany documents of program descriptions, Title I applications, schooldistrict budgets, expenditure reports, and evaluation reports.
Next, descriptive and quantitative summaries were prepared foreach project including objectives, principal activities, and resourcestImployed. Where appropriate the descriptions indicate (1) the num-
schools in the district and in the sample which had each pro-jet,. (2) the number an.d grade levels of pupil participants, (3) totaland per pupil expenditulres, and (4') hours of pupil exposure.
Finally, when the grade levels served by each project had beenidentified, reported expenditures were allocated to appropriate gradesby project and time period for each sample school.
The following sections describe the sample schools, some of theircharacteristics, and the bases for their selection; present the CEprogram descriptions; and summarize the results of this exercise.
66 67TMP-115
Sample Schools and Their Characteristics
Table 25 lists the characteristics used as criteria for selectionof the ten sample .:lemeritary schools and indicates the CE programsfor each school in 1966-67.* When selecting the sample from ap-proximately 55 schools eligible for Title I programs, an attempt wasmade to choose a sample which would possess a wide range of school
characteristics. In addition to these criteria, schools were chosento represent the mix of CE programs of the district. Eligible schoolsranged in enrollment from 150 to over 2500 pupils and the sampleschools chosen cover a large part of that range.
A high proportion of eligible schools had all-Negro pupil popula-
tions; however, some schools were selected which had the largestproportions of non-Negro pupils among the eligible schools. Theselection was based on data compiled in the winter of 1965-66 to
establish the initial eligibility of schools. The sample of ten schoolsincluded three which had substantial proportions of nen-Negro pupils.Since that time, substantial change of pupil populations has occurred,and in 1966-67, only one of these schools reported having any non-
Negro pupils.
About 16 percent of District 10 pupils were from low-incomefamilies; in eligible schools this percentage ranged up to 55 percent.tSample schools had from 22 to 49 percent of economically deprivedpupils. School personnel point out that high mobility the move-ment of a pupil from one school to another is a frequent character-istic of schools with substanti-1._ proportions of pupils from poorereconomic circumstances. Indices of mobility were computed forthe sample schools and are shown in Table 26 with attendance andracial concentration percentages. Considerable variation of pupilmobility exists, fr i.n year to year and from school to school, but
no distinct pattern is indicated.
* The selection criteria do not agree with similar information insubsequent talles due to different reporting periods.
tAs computed for the 1965-66 Title I application, based on annual
family income of less than $2,000.
1
I
)
SECTION 3
Table 25. Sample selection criteria and compensatory educationprogramsDistrict 10.
r---;:...I.........:_......,........................
67
1")a)
a.
r^mpensnteiry Frificntinn Program
...in
._ 1
00 0
2 2. C) .17. c >2 b a
0o 0
o t E ...vi 4. V iLiRA0
a) < o c c V) -0 .tY 4)0 C.) e E ti 0 x 0 20) I.13 > 0 ID UU g 2 ._
v,> .c 0
.... an = ...... VIC 0i.. ... 'V
C En ti 2 CL "5) 2' 1 t 0u .3 12 c E C 0 2 0 0
10 ...I CL. E X
8O a.
c -.g I.=
1 186 27.3 79
2 570 35.5 100
3 1561 35.2 100
4 265 31.8 100
5 1618 25.3 100
6 272 37.0 88
7 29:3 22.2 54
8 776 22.0 100
9 506 32.1 100
10 545 49.1 100
X X
x
x
Figure 9 permits comparisons to be made among racial con-centration, mobility and attendance. These comparisons do notindicate obvious relationships among these characteristics. Neithermobility nor attendance appears associated with racial distribution.An inverse relationship between pupil mobility and attendance mightbe anticipated but these data do not support that expectation.
68 67TMP-115
Table 26. Attendance, mobility and racial distribution ofsample schools (percentages) in District 10.
63-4
Attendance
64-5 65-6 66-7 63-4
Mobi li tya
64-5 65-6 66-7 63-4
Negro Pupils
64-5 65-6 66-7
95 92 89 92 32 35 37 28 20 41 69 100
2 89 92 89 91 29 41 18 62 100 100 100 100
3 89 89 86 90 14 29 34 47 100 100 100 100
4 85 87 87 91 23 28 28 35 100 100 100 100
5 91 92 84 92 39 13 54 25 100 100 100 100
6 83 87 87 88 64 43 26 74 29 52 74 100
7 90 98 90 92 47 53 35 47 7 23 90 44
8 88 90 85 93 43 13 22 17 100 100 100 100
9 87 87 89 91 16 32 21 22 100 100 100 100
10 83 87 85 86 33 29 25 30 100 100 100 100
Note:
a (Gains + Losses) + ADM.L
Improved attendance is an objective of many CE programs. Thefollowing attendance data were obtained by school and grade forfour consecutive years ending with the 1966-67 term: initial re-gistration, gains, losses, end-of-year membership, average dailymembership, average daily attendance, average daily absences,percent attendance, and percent absences. Attendance rates weresummarized and examined to detect any systematic change thatwould be helpful in analyzing the effects of CE programs. Table26 showed average annual attendance rates for the sample schools.
SECTION 3 69
SCHOOL
1
2
100
-
50 ...ea, ..4
0
100
50-
o .
.../.0
ee /*V
6
7
100
50 i
0
50
0
4. a.anfowaleel,
/.. .. f./
co Me 0 460.11b
3
100
so-
8
1M150
moomp=mNIMIND.. i.e."' 4 ......
....,",,,.../..
o
100 IN&
al .. ay..am, ...4 9
50 50
4 00° ,IP ella . 00
0 0
5
100
50
0
01111
411.4P GO.dP Sib *.
.%/% %./ qk
%. .%,
II I t64 65 66 67
Ig.....' ... 411.4ra ....
146.
dr,..0 -I1111 44b
... Imenur
Nombo alo as al op 6.4.
/09. ft.0111
10
KEY (%):
----ATTENDANCENEGRO
50
0
I. 494..0". a... .6.11111
es go404.
,..*lab 'tame.
1 I 4
64 65 66
MOBILITY
Figure 9. Racial distribution, mobility and trends ofattendance (percent) in sample schools.-
70 67TMP-115
A mixed pattern of increases and decreases in attendance ratesis apparent. Eight schools recorded decreased attendance for the1965-66 school years. All ten sample schools reported increasedattendance during the 1966-67 term. This is the only year of the
period in which all of the sample schools recoided imprnved at-tendance rates. To avoid a hasty conclusion that this improvementresulted from CE programs, a more probable cause was soughtand found. Extensive storm damage was sustained in this schooldistrict early in the 1965-66 school year and widespread disruptionof school activities was experienced during a period of weeks. Theimproved attendance of the following year more probably reflectsthe resumption of normal attendance. Additional analysis of attend-ance data has been reported in Section 2.
Monthly summaries of attendance data were not available at thecentral school office but probably could have been obtained from in-dividual schools at the expense of additional time. If monthly datahad been obtained, rather than average annual data, it might havebeen possible to omit the storm period from consideration and makea more sensitive appraisal of the impact of CE programs, and otherfactors, upon pupil attendance.
Program Descriptions
Sample schools had from one to four CE projects in 1966-67.
One project, Quality Instruction, was supported by school districtfunds, eight projects were funded under Title I, ESEA, and two pre-school projects were funded under the Economic Opportunity Act(EOA). The two EOA projects are listed for information only. Both
were pre-kindergarten projects; no allocation effort was requiredand very little project description information was available. Inaddition, there was a district-wide surmner CE program, calledReading-Enrichment-Recreation (RER), which is not listed in Table25 because pupil participants were not identified by schools.
PROJECT: QUALITY INSTRUCTION. Purpose: To improve thequality of instruction in selected schools which are experiencingsubstantial change of pupil population due to racial integration.
SECTION 3 71
Description: When the proportion of Negro pupils in a schoolexceeds one-third, the school may be selected for this project.More intensive services are allocated to the school which may in-.clude administrative or clinical personnel, classroom or specialistteachers, guidance counselors or teaching materials and supplies.This project is funded by the school district. Three of the sampleschools received small increments of support from this project in1966-67. It was reported that schools with this project seldom areeligible for Title I projects.
PROJECT: TEACHER AIDES. Purpose: To relieve teachers ofnon-professional and routine tasks in order to provide increasedopportunity for creative teaching, planning, individualized instruc-tion, and pupil counseling; to enable use of a greater variety andquality of instructional materials; to provide increased attention tothe needs of disadvantaged children.
Description: In 1965-66, 352 teacher aides were assigned to asmany classes in 53 schools, 1 per class with average class size of30 plus, in kindergarten and grade 1, including 17 "3-1" schools.*Each aide was employed 4-1/2 hours per day for ten weeks. In 1966-67, 379 aides were assigned to classes in kindergarten, grade 1, andgrade 2, in 54 schools, as follows: 345 part-time aides (4-1/2 hoursdaily, with 3 hours in classroom) were provided in 255 first grade,83 second grade, arid 6 combination first-second grade classrooms;full-time aides (6 hours with 5-3/4 hours in classroom) were as-signed to 31 kindergartens; 3 aides were also assigned to upperelementary grades as an experiment. During this year, the aidesparticipated for 27 weeks; all were high school graduates and overhalf had some college education. In 1965-66, the aides were sup-plemented by $350 of supplies and equipment per classroom in grades1 and 2, e.g., tape recorders, film strips, picture sets. In 1966-67, $300 of equipment and supplies were provided each of the 84second grade classrooms added to the aide project that year. Theaides enabled teachers to devote more time to instruction; also theyworked with children individually and in groups, especially in thearea of language development (which was part of their in-servicetraining).
*Schools having the Intensive Instructional Improvement project,see page 77.
,.
72
Table 27 summarizes Teacher
67TMP-115
Aide assignments in the sampleschools during the two program periods.
Table 27. Teacher aide assignments.
SAMPLE
SCHOOL
rDAIIP..7..-......
MARCH-jUNE 1966
K 1 2 Ca K
1966-67
1 2 Ca
1 1 2 2
2 1 3 1 3
3 2 10 1 11 2
4 1 1 1 2 1
5 3 7 3 7 7
6 1 3 1 3
7 1 2 2
8 1 4 1 4 4
9 1 2 1 1 2 2
10 1 3 3
Note:
°Combination of grades 1 and 2.
1965-66 l 966-67
Total Schools Served: 53
_53
Sample Schools Served: all 10 all 10
Total Pupils Served: 10,560* 11,461
Total Expenditure: $488,4721. $911,833
Expenditure Per Pupil Served: (10 wks) $44 (27 wks) $80
Hours /Pupil Exposure 225 K:8001-2:600
Estimate: number of aides x 30 pupils/classt Estimated from salary and equipment costs, plus estimatedadministrative overhead.
SECTION 3 73
PROJECT: ADJUSTMENT TEACHING. Purpose: To meet the varyingand special needs of grade 3 underachieving students from low-incomeareas.
Description: Adjustment teachers provided individualized tutoringto small groups of 8 to 10 pupils in any needed subject area. Eachchild in the project participated 40 minutes per day. Underachievingchildren were selected on basis of test scores, cumulative records,and teacher observations. This project was initiated in the fall of1966 and was reported to have continued for 20 weeks.
1966-67
Total Schools Served: 8
Sample Schools Served (School 7): 1
Total Pupils Served: 430*
Total Expenditure: $60,000t
Expenditure Per Pupil Served: $139
Hours /Pupil Exposure: 67
PROJECT: REDUCTION OF CLASS SIZE. Purpose: To Iftelp makepossible individualized instruction by reducing the size of classes.
Description: Class size was reduced in grade 1 of twelve schoolsthrough the use of portable classrooms. In 1965-66, expenditureof $847,477 was made to obtain 17 portable classrooms. In the fol-lowing year, 15 additional teachers were employed to use them.
* Estimated: 9 pupils x 6 hours/day X 8 schools
Estimated: 8 teachers at $7,500 (average salary and benefits)
7467TMP-115
Total Schools Served:
Sample Schools Served (Schti-,1
Total Pupils Served:
Total Expenditure:Expenditure Per Pupil Served:
Hours/Pupil Exposure:
1):
1966-67
12
1
720*
$257, 248t
$357
450**
PROJECT: CLINICAL READING. Purpose: To improve writing,reading, spelling, work habits, and powers of concentration in
grades 3 through 6.
Description,: Clinical Reading Centers were established by the
school district, before ESEA programs, in elementary and secondary
schools. Title I funds permitted expansion of this project. In 1965-
66, 10 after-day centers were established, serving 15 pupils each
for one hour each day. In 1966-67, 8 after-day centers were cone.
tinued and additional day centers were established, two full- and
two half-day centers. In the day centers, groups of 32-35 pupils
receive 45-60 minutes/day of instruction instead of another class
activity which would be missed least. The day program servedpupils in grades 3 - 6, for 28 weeks (1966-67) at 5 hours/week. The
after day programs served grades 4 and 5, 'for 24 weeks (1966-67)
at 5 hours/week; class size was reduced to 8 - 10 pupils in 1966-67.
The following summary describes the incremental portion of the
program, made possible by Title I support. The Center in thesample school was funded by the school district but is representa-tive of other centers which were funded by ESEA.
*4t
.11=1,
Estimate:Estimate:
c ons tructionclas s room.
** Estimate:classes).
12 schools x 30 pupils x 2 classes/day.
15 teachers x $7, 500 average salary plus 1/10 ofcosts (based on estimated 10 year life of a portable
30 weeks x 5 days/week x 3 hours/day (two hal-.-day
SECTION 3
1965-66
75
1966-67
Total Schools Served: 10 12
Sample Schools Served (School 2): 1 1
Total Pupils Served: after-day* 151 81day 89
Total Expenditure:* $49, 555 $107, 379
Expenditure Per Pupil:* $328 $631
Hours /Pupil Exposure:* 75 50
PROJECT: LANGUAGE ARTS TEACHING CONSULTANTS. Purpose:To provide imaginative approaches to meeting individual needs ofunderachieving pupils, in all phases of listening, speaking, reading,and writing.
Description: Two language arts teaching consultants worked intwo schools with small groups of children who were achieving belowgrade level, in grades 1 through 6. The project encompassed allphases of listening, speaking, reading and writing, using innovativeapproaches and materials not used in the regular classroom. Groupsconsisted of about 8 children from each grade level, with meetingsdaily during class hours, plus one hour after school for specialtutoring and recreational reading. The project also included fieldtrips; a.,-litory and visual examinations; and 'parent involvement.
r4.
These data are recorded here as reported by the project super-visor even though the basis for computation may differ from otherprojects described in this section.
76
1965-66
67TMP-115
1966-67
Total Schools Served: 2 2
Sample Schools Sel7ved (St.hr,rd A): 1 1
Total Pupils Served: 120 88
Total Expenditure: $15,900 $32,741
Expenditures per Pupi' d $133 $372*
Hours /Pupil Exposure: 50t 300t
PROJECT: CENTRALIZED LIBRARIES. Purpose: To enable schoollibraries to serve both school and community. They are designedto be accessible to the neighborhood after school hours and duringthe summer, as well as to provide varied instructional and enrich-ment materials as part of the school program.
De.. .- ;ion: The project was initiated in 1965-66 and continued
du: Ing trim following year. It involved the assignment of librariansto elementary schools for the first time. Libraries have been, orare being, developed in 15 schools employing portable buildings, re-novations, new construction, and limited service to two schools
without librarians or furniture,. Five libraries began providing ser-
vices in January 1967; the remainder are in the process of develop-ment. A center was established to process books for six of these
libraries.
* Per pupil expenditure for those directly participatir; a figureof $77.59 was given which "inclr-3.es all children in grades onethrough six in the two schools involved in the pro,iect because it is
felt that all were a.fected directly or indirectly by the project."
tIn 1965-66, project started in one school 6 weeks before close
of school year; in other schools, 4 weeks before close. Estimatesassume 1 hour during school day and 1 hour after school, daily
for 5 and 30 weeks respectively, allowing for holidays and absences.
_
SECTION 3 77
1965-66 1966-67
Total Schools Served:
Sample Schools Served:
Total Expenditul-e:*
5
(Ce.1-setrd ) 7
$460,584 $441,092
PROJECT: INTENSIVE INSTRUCTIONAL IMPROVEMENT. Purpose:To concentrate services in 17 elementary schools in low-incomeareas where pupils consistently have tested low in readiness andachievement, among the most educationally disadvantaged in thesystem.
Description: This project was in operation for about 21 weeks ofthe 1966-67 school year, starting in January 1967. Services in-cluded the areas of administration, instruction, health, and socialwork. Adjustment teachers aided selected third grade pupils andconsultants participated "to intensify and vitalize learning experi-ences. " Teachers aides were assigned to grades K, 1 and 2. After-school tutoring was available, especially in reading; selected pupilsparticipated thrice weekly for an hour in groups of 4 to 10. A totalof 1961 pupils received up to 50 hours of tutoring.
Each school in this project had physical education, art and musicprograms. Instructional materials, designed for disadvantagedpupils, were provided $100 per classroom in grades 3 through 6
plus $1 per child in the tutoring class. Administrative assistants insix of the largest schools, with enrollments exceeding 1300, enabledthe principals to devote more time to improving instruction. Oneperson devoted half-time each to the duties of administrative as-sistant and adjustment teacher. Social and attendance workers visitedpupil homes; counseling was available for parents and referral tocommunity social and welfare agencies. Extensive medical and den-tal examinations were made; funds for correcting defects were ob-
tained from community service agencies.
Teachers and principals determined and requested specific kindsof help from consultants, identified problem areas, and recommended
These are only part of the total cost of the project, to be completedin the future. Adequate information on costs of developing, stocking,and operating of these libraries is not available.
78 671MP-115
instructional materials. There were 5 primary and 4 upper elemen-tary consultants, a ratio of one consultant for 50 teachers. Teacherswere given released time to "pursue professional tasks independently."
Additional staff included:
14 adjustment teachers for 3rd grade6 administrative assistants/adjustment teachers
10 consultants2 music consultants, grades 4, 5, 64 itin( rant art teachers, grades 4, 5, 6
18 physical education teachers6 health staff (4 nurses, dental hygienist, health coordinator)
15 social services (8 visiting teachers, 6 attendance workers,1 coordinator)
1 director
76 Total
1966-67
Total Schools Served: 17
Sample Schools Served (Schools 5, 8 and 9) 3
Total Pupils Served: 16, 500
Total Expenditure: $685, 014*
Expense Per Pupil Served: $41.50
Hours/Pupil Exposure: 630t
PROJECT: ENGLISH AS A SECOND LANGUAGE. Purpose: Toteach English to children with other native languages, who often hearno English spoken at home, and to provide a better understandingof regular classroom activities.
Description: Teachers worked with small groups, using audio-visual-lingual techniques. The project started in the summer of1966, continued during the regular school year 1966-67, and wasconducted during the summer of 1967 as an integral part of the
' Does not include funds spent in 3-I schools under other Title Iprojects e.g., teacher aides were in the teacher aide project budget.t
Estimated on basis of 21 weeks X 5 days per week X 6 hours perday (assuming no absences or holidays).
SECTION 3 79
Reading-Enrichment-Recreation (RER) project, discussed below.In the summer sessions, each child participated 2 hours a day for 6weeks. In the 1966-67 school year, participation was generally 1hour a day, although some variation according to individual needswas allowed.
Summer1966 1966-67
Summer1967
Total Schools Served: n/a 9 n/aSample Schools Served: n/a
(School 7)1 (avg. 35
pupils /class )n/a
Total Pupils Served: 303 320 365
Total Expenditure: $51,182 $65,981 ,Exp. Per Pupil Served: $169 $206 ,Hours /Pupil Exposure 60 150 60
PROJECT: SUMMER READING-ENRICHMENT-RECREATION (RER).Purpose: To provide special educational, cultural, and recreationalservices to educationally deprived children in grades 2 through 12(grade 1 not included in 1966). Specific objectives, as expressed for1967, were to improve classroom performance in reading beyondusual expectations; to improve children's verbal functioning; to im-prove artistic awareness and self-expression; to improve the chil-dren's self-image; to increase their expectations of success in school;to improve the health of the children through physical fitness.
Description: An RER center was located in each of the 54 Title Ischools. The six-week project in the summer of 1966 consisted ofmorning sessions (3-1/2 hours including breaks) with one hour de-voted to each of reading and language skills, cultural enricirnent, andphysical education, and a 2 hour afternoon recreational program whichwas open to any child, regardless of whether enrolled in the morningclasses. Pupil-teacher ratios in the classes were as follows: remedialand language arts, 10:1; cultural enrichment, 15:1 (20:1 in 1967);physical education, with trained instructor, 25:1; (in practice, in phy-sical education, trained instructors were not always available andclass size sometimes exceeded 30). Total elementary instructionalstaff in 1967 was 447 teachers and 11 consultants for an overall ratio
*Part of overall RER budget.
80 67TMP-115
of about 18:1. Each child went on 3 field trips, cultural enrichmentfield trips, with this experience being worked into the content of thereading program. The project provided innovative instructional ma-terials and equipment, delayed in 1966. The 1965-66 evaluationstates, "...techniques, materials and activities which duplicatedthose used during the regular school year were least successful inproducing the desired behaviors." Experience with art, music anddrama (including pupil participation) were provided. The afternoonrecreation program concentrated on group sports and games. Se-condary students with knowledge of playing an instrument took partin instrumental music program three afternoons a week. Lightsnacks were served to elementary pupils during morning breaks.
The summer program, 1967, was similar to that in 1966 exceptthat (1) grade 1 pupils could also participate; t2) the session lastedfor 6-1/2 weeks, plus 1 week of pre-service training for staff; (3)the elementary cultural enrichment portion of the program wasfocused more specifically on language arts; (4) the secondary RERprogram was changed more radically so that emphasis in the academicpart of the program "shifted more or less exclusively to the languagearts; especially reading, although music also continued to be a strongfeature, including jazz performances." A wide selection of readingmaterials was made available (although again delayed), e.g., paper-back books for secondary pupils. Substantial attrition of these bookswas anticipated, even sought. One supervisor believed that disap-pearing books would indicate greater pupil interest in reading. Someteaching equipment was employed, such as tape recorders.
The supervised recreation portion of the project was continuedwithout substantial change and employed the equipment that had beenpurchased for the program of the preceding summer. In 1967, itwas extended beyond the end of the academic program (to August 18).The English as Second Language project was included in the summer1967 activities.
SECTION 3 81
Summer 1966 Summer 1967
Total Schools Served:* 54 54
Sample Schools Served: all 10 all 10
Total Pupils Served: 9,000 9,939tTotal Expenditure: $2,534,189** $1,302,213
Expenditure Per Pupil Served: $282 $131
Hours/Pupil Exposure* 90 90
Expenditures, by School and Program
This school system was one of the few visited that records expendi-tuxes for individual schools. Expenditure data were furnished by ac-counting categories, for the regular school programs of 1964-65,1965-66, and 1966-67 and for the compensatory education projects of1965-66 and 1966-67. Table 28 is a sample of the type of informationobtained. Tables 62 through 70 in Appendix G contain information forthe other 9 schools. Table 29 summarizes total expenditures, byschool and year, for the regular and CE programs. The prinicipalimpact of this information is that it reveals considerable variationsof p tr-pupil expenditure, even for what is described as the same mixof programs. For the regular program of 1964-65, the largest per-pupil expenditure was more than double that of the smallest. Similar
Elementary and secondary schools combined.
t Of the 9,939 enrollment, about 72 percent enrolled in the centerwhere they regularly attend classes; 19 percent enrolled in anothercenter; 7.7 percent were children attending parochial schools; 1.2percent were children attending other private schools. About 1,800were secondary level pupilsfor these, there was a high attritionrate; perhaps 50 percent left the program before its end. The bestmaintenance of attendance at secondary level was at grades 7 and 8.
*
Including about $860,000 for playground or sports equipment.
Three hours/day for 6 weeks (not including afternoon or extendedrecr eation pr Og rams ).
..?"-
r1 7
,, -.
7'7
,.."
1,tr
,
Tab
le 2
8.E
xpen
ditu
res
for
regu
lar
and
com
pens
ator
yed
ucat
ion
prog
ram
sSch
ool 1
.
Dis
tric
t 10
Sch
ool 0
1R
egul
ar S
choo
l Pro
gram
sC
ompe
nsat
ory
Edu
catio
n P
rogr
ams
Exp
endi
ture
Cat
egor
ies
Tea
cher
Aid
esA
djus
tmen
t Tea
chin
gR
educ
ed C
lass
Siz
eP
roje
ctQ
ualit
y
64-6
565
-66
66-6
765
-66
66-6
765
-66
66-6
7
##
#$
#$
#$
#$
#$
)
Prin
cipa
ls, C
onsu
ltant
s, S
uper
viso
rs1
8647
190
061
1147
430
Tea
cher
s6
3126
86
3329
18
5506
23
2481
229
9332
28
Libr
ary,
Aud
io-v
isua
l&
TV
Per
sonn
el18
Sec
reta
rial-C
leric
al1
3536
136
231
4070
Tot
al S
alar
ies
4345
145
920
7060
625
2929
9332
28
Tea
chin
g S
uppl
ies
& O
ther
Exp
ense
s25
135
638
864
731
718
85
Tot
al In
stru
ctio
nal E
xpen
ses
4370
246
276
7099
431
7633
1051
13
Hea
lth S
ervi
ces
Pup
il T
rans
port
atio
n
Pla
nt O
pera
tion
& M
aint
enan
ce79
7776
4310
726
27
Fix
ed C
harg
es10
913
114
3
Foo
d S
ervi
ces
Stu
dent
/Com
mun
ity A
ctiv
ities
Tot
al S
uppo
rt E
xpen
ses
7977
7643
1072
610
913
117
0
Tea
chin
g E
quip
men
t13
551
913
920
9
Tot
al E
xpen
ses
5167
953
919
8185
538
0435
8054
92
----
-.~4
i001
."""
*401
"0"4
1°,1
44
th
Tab
le 2
9. S
umm
ary
of p
rogr
am e
xpen
ditu
res
of D
istr
ict
10.0
Sam
ple
Sch
ool
1964
-65
Reg
ular
Bud
get
Tot
alE
xp. $
$/P
upil
151
679
407
211
2599
215
329
1951
200
460
938
256
529
2636
178
668
356
289
784
713
332
816
1981
218
913
7535
237
1095
944
180
1965
-66
Reg
ular
Bud
get
IC
E B
udge
t
Tot
al$/
Pup
ill
Tot
alE
xp. $
Exp
. $$frupil
1966
-67
Reg
ular
Bud
get
CE
Bud
get
Tot
alE
xp. $
$/P
upil
5391
928
338
0420
1133
8819
923
427
41
3043
9020
722
437
15
6723
126
928
0611
3103
8123
611
690
9
6842
025
610
528
39
9071
030
812
965
44
1404
9219
258
658
1350
3526
648
3110
1034
3019
548
549
8185
5
1428
95
3926
95
8159
6
3928
71
9308
4
1164
98
1868
01
1838
56
1344
73
437
264
305
334
253
288
361
294
412
273
Tot
alE
xp. $
$/P
upil
9072
48
3594
666
3226
625
4136
120
3668
324
2163
167
1885
658
1938
131
1078
924
4864
10
Not
es:
°Exp
endi
ture
s of
the
sum
mer
pro
ject
(R
ER
) w
ere
not r
epor
ted
by s
choo
ls a
nd a
re n
ot in
clud
ed.
bP
er p
upil
expe
nditu
res
are
base
d on
AD
M fr
om th
e ye
ar e
nd a
ttend
ance
rep
ort.
CA 73 e 0 z 03
1
. fr
84 671MP-115
but lesser disparities are shown for the subsequeat years. Forschools with the same CE projects, per-pupil expenditures alsovary in the same way. For example, while schools 4 and 10 eachhad the same CE project in 1966..67, the per-pupil expenditures inthe former were double those of the latter. This variation is notdue to different numbers of teachers or other resources added.Schools 5, 8 and 9 had the same two CE projects in 1966-67, yetthe largest per pupil expenditure was about 25 percent greater thanthe smallest. Similar differences are shown for 1965-66.
These differences raise some important questions. The reli-ability of estimates of CE program cost by grade and school basedon district level program information depend on the degree of stand-ardization of programs among pupils served. If equal amounts ofa resource are added by a project for each school but numbers ofpupils differ, it is not really meaningful to assume that the resultsof the project will be similar among schools.
Moreover, even with a similar per-pupil allocation, resources mayactually be employed in different ways. The statistics show variationsof per-pupil expenditures between schools; they do not show variationsof the employment of resources within a school or class. Consider aproject in which a reading teacher instructs five pupils of a class. Herimpact on the five pupils is not altered by the large or small number ofpupils who constitute the remainder of the class. At the same time,allocation of the costs of teacher and materials to the whole class doesnot reflect the real impact or cost per pupil of the project. If alloca-tion is made on the basis of the five pupils served directly, no recog-nition is given to the widely claimed "spillover effects." Allocationto the grade level may involve unknown amounts of error in estimatesof exposure to project activities.
Not only are there substantial variations of per-pupil expendituresbetween schools which have the same mix of projects but these "build"on very substantial differences of the regular budget per-pupil ex-penditures. Other variations are shown by school, from year to year,not all of which are increases, as public information records to bethe rule for schools in general. Half of the sample schools had spentless per pupil in 1965-66 than in the preceding year.
There are at least three factors which account for much, perhapsall, of these differences of per-pupil expenditures. Frequent andconsiderable variation of school registration or ADM is indicated byend-of-year pupil accounting data obtained for this study. School
-
...................OW,TWIN **011111ft,e,loy,....... ........1*10.9M.04.0.1.0.1* .................*44....11*./........
SECTION 385
administrators plan budgets based on anticipated school enrollmentfor the coming year. They cannot control the number of pupils whochange schools. Rarely do they change the boundaries of school ser-vice areas to offset enrollment fluctuations; they do not wish to causefrequent disturbances of pupii-schooi- community relationships.School area boundaries are changed when the physical capacity of aschool is exceeded as may be caused by growth of pupil populations.Another factor is the trend of increased salaries in recent .fears.The larger regular budgets of all sample schools in 1966-67 werecaused, at least in part, by salary increases which averaged $1, 500
per year for teachers. Finally, teacher seniority and salaries ac-count for another element of variation.
Such variations of expenditures will complicate any attempt tocompare programs on a cost-benefit basis, i. e., to seek the cost cfobtaining a unit of increased pupil performance via alternative CEapproaches. Salary differences among teachers, changes of salaryscales, and changes of expenditures for materials or equipment aresubject to control of school administrators and can be taken into ac-
count. Fluctuations of pupil enrollment are not controllable and area source of error or reduced precision in evaluating the benefitsand costs of educational activities.
Allocation, by School and Grade
When the school district accounting system permits summarizingexpenditures by school and program, the methods for allocating these
expenditures to grades are relatively simple. The difficult part ofthe process is that of determining which program activities were di-rected at each of the several grades of a school. This task involvedsearching through many documents in order to assemble a reasonablycomplete list of assignment of programs to grades. Frequent dis-crepancies and disparities were found. For example, expendituresreported in different sources often were in conflict; some were esti-mates for which the basis was not always clear, others were partialor total expenditures but for different time periods.
After compiling and checking the information on hand, preliminarydrafts of project descriptions were prepared. Unresolved questionswere referred to district and project administrators who were veryhelpful in providing additional information and explanations. When
the information was considered reasonably complete, the projectdescriptions were written. From this search was obtained the grade
86 67TMP-115
level assignment of the several projects. The final step was thearithmetical exercise of distributing stated expenditures at the schoollevel to the grades served. Except when project information indicatedotherwise, expenditures were allocated to grades in proportion to thenumbers of pupils in that grade for each year.
Table 30 is an example of the results of this exercise and containsestimates of regular and CE program expenditures, by school, grade,and year.* Information for the other 9 schools in contained is Tables71 through 79. These allocations must be regarded as estimates, atleast until verified by school district administrators. Caution mustbe exercised when comparing these estimates with changes of pupilperformance. The time periods during which these estimated funds
were expended do not necessarily coincide with that of the pre- andpost-test measurements. Some refinements or reallocations may be
appropriate before undertaking more sophisticated analysis, such asregression studies. This may require assuming specific rates of ex-penditure of resources month by month rather than a constant rate ofexpenditure through the school year. The following discussicn isbased on the constant-rate assumption.
Effects of CE Programs on Achievement
Analysis of achievement test scores has been reported in Section 2.Only a brief recapitulation is made here in order to complete thiscase study. Figure 10 summarizes changes of first deciles and meansof reading test scores of grades in sample schools for which the a-chievement test data were suitable for statistical analysis. For ex-ample, consider the changes recorded for grades 3 and 4 of School 1.
* These estimates, on a per-pupil basis, do not agree with the per-pupil estimates of project descriptions above. They were computed
on different bases and from different data. These estimates arebased on reported expenditures and were computed from the totalnumber of pupils in a class. The estimates of the project descrip-tions wer, made chiefly from budgetary cost information and attemptto indicate expenditures for pupils who participated directly in aproject.
Totals are reported expenditures; columns may not add to thetotals due to rounding.
SECTION 3
Table 30. Program expenditures by grades of School
87
1965-66 I Regular
Programs
$
Teacher AidesAdjustmentTeachers,Reduced ClassSize
$
Total CEPrograms
$Grade ADM
K 29 7411 994 994
1 57 14565 1953 1953
2 25 6388 857 857
3 24 6133
4 26 6644
5 21 5367
6 7 1789
Sp. 22 5622
Totals 211 53919 3804 3804
1966-67 Regular
Progrcms
$
Teacher Aides,AdjustmentTeachers,Reduced ClassSize
$
ProjectQuality
$
Total CEPrograms
$Grade AD M
K 30 13567 1085 910 1995
1 35 15828 1266 1062 2328
2 34 15376 1229 1031 2260
3 19 8592 577 577
4 18 8140 546 546
5 20 9045 607 607
6 14 6331 425 425
Sp. 11 4974 334 1 334
Totals 181 81855 3580 5492 9072
Note:aPer-pupil expenditures for regular and CE projects of the ten sample schools are
given in Table 28.
88
-64
3.3
5
1---3
3
6
I.
1 2 3 4 5 6 7 8 9 10
SCHOOL
3
3
3 4 GRADE
5
34 5
3
3
67TMP-115
3
3
3
-.5 4
1 2 3 4 5 6 7 8 9 10
SCHOOL
Figure 10. Observed change of reading achievement for test period
September 1965 to September 1966District 10.
24
21
18
15
12
9
SCHOOL 1GRADES 3-6
6
3
65-66 66-67
SCHOOL 6GRADES 3-6
SCHOOL 2GRADES 3-6
65-66 66-67 65-66 66-67
TFT DATE E=1 REGULAR PROGRAMS CE READING
Figure 11. Exposure to CEselected schools.
t1.0 CE OTHER
SECTION 3 89
The score of the third grade pupil at the first decile was 5 unitsgreater on the post-test while the score of a classmate at the meandid not increase. Scores of fourth grade pupils at the first decileand mean were lower on the post-test than when tested one yearearlier. This figure records about equal incidence of negative andpositive changes of these measures. No consistent pattern of changeis shown, by school or grade. These measures, and others reportedin Section 2, yield no indications of significant change of readingachievement.
No relationships were found between change of pupil test scoresand type of CE programs or level of expenditure. Table 31 givesthe relevant data for grade 3 in nine schools. The change in themean and at the first decile were not more favorable in those caseswhere per pupil expenditures on CE were the highest. This absenceof the desired results is not unexpected; the following paragraphspresent some of the reasons.
First, achievement test scores analyzed did not match well withthe grades which had CE projects. Test data were not obtained forsome grades. In particular, all sample schools had the TeacherAide project in grades K through 2 but test results were not obtained.Some of the test data re cived could not be made suitable'for statisticalprocessing. For example, either pre- or post-test scores were notavailable for some grades. Table 32 compares the grades for whichpre- and post-test scores were used to measure changes with thegrade levels which participateA in CE activities.
Next, not all of the CE projects could be expected to have impactupon reading performance. Only one of the ten projects was orienteddirectly toward improving reading achievement and it served limitednumbers of pupils. Two projects focused upon English language artsof which reading is an essential part. Two other projects includedreading content; one was started after the post-test date, the otherwas a summer program for which pupil particfpants from the sampleschools cannot be identified. The remaining five projects providedadditional personnel, materials, or equipment for use in other in-structional activities. Table 32 sl_ows that only three of the sampleschools (2, 6, 7) had reading-oriented CE projects in 1965-66.
Figure 11 demonstrates the exposure of pupils to CE activitiesin three sample schools. The relative magnitudes of regular andCE expenditures are shown on the vertical axis.
Tab
le 3
1.E
xpen
ditu
res
Gra
de 2
(19
65-6
6) a
nd a
chie
vem
ent d
ata
for
Gra
de 3
(S
epte
mbe
r 19
66)
of D
istr
ict 1
0.
Sch
oola
Reg
ular
Pro
gram
Tot
alC
EA
DM
Exp
endi
ture
s/pu
pi I
Pre
&D
1A
Q1
AQ
3C
EC
E +
Reg
.7
818
545
2108
101
2120
434
.5-1
.46.
0F
.i.o
-5.2
410
360
921
4321
262
31.4
4.8
5.7
3.8
2.5
163
8485
725
3429
036
.00.
14.
81.
5-2
.0
920
729
2106
8525
269
31.1
3.0
2.0
5.0
1.0
712
763
2481
4654
331
36.6
1.0
0.6
3.0
-1.7
1013
715
1704
7622
203
25.5
4.0
0.0
6.0
3.3
213
924
2433
7632
215
34.2
-0.9
-2.0
-2.0
-1.0
551
120
6337
218
2926
432
.5-1
.7-2
.0-3
.0-3
.0
614
375
2954
5851
298
33.2
-3.0
-5.2
-4.0
-1.0
346
085
6559
245
2719
4N
otav
aila
ble
Ave
rage
of a
bove
20,7
9828
4697
3225
332
.80.
61.
11.
0-0
.7
Not
e:aS
choo
lsar
rang
ed b
y or
der
of m
agni
tude
on
obse
rved
cha
nges
in D
i.
SECTION 3
Table32. Comparisons between grades with CE projects and grades forwhich achievement data could be analyzed.
91
eftrAINre. ir-eNn WILIIPLi ArLIICVMS.ACAITVIVAUC.) n...i . vvniLn m..n1LvLovu.1,41
TEST DATA WERE ANALYZED
r`CI A MCC 1A/L11/"LI LIMN..-m.1....1 vvill....11 I ir-,..,
CE PROJECTS
ji Pre (9/65) Post (9/66) 65-66
-MO
1 66-67
3,4 3,4 K-2 K-2
3,4 3,4 K-2, 1-6, K-2, 1-6
3-6° 3-6°
- - K-2, 1-6 K-2, 1-6
4 3 3 K-2 K-2
5 3,5 3,5 K-2 K-2, 3-6°
6 3 3 K-2, 1-6° K-2, 1-6,
3 3 K-2, 1-6° K-2, 1-6,
1-6°
8 3 3 K-2 K-2, 3-6°
9 3 3 K-2 K-2, 3-6°
10 3,5 3,5 K-2 K-2
Note:
°Projects which had at least some reading emphasis.
School 1 illustrates the situation of minimum pupil exposure,about one month, in a CE project which added limited amounts ofpersonnel, materials or equipment to the school program. A pupilin School 6 who was in a 3rd grade in 1965-66 and progressed tothe 4th grade for the following year might have received 5 or 6
months of exposure in a language arts activity. In this school,the CE expenditures were greater in absolute and relative terms.A similar pupil in School 2 could have received about the samelength of exposure in a project with definite reading orientation.Per-pupil expenditures for CE projects were greater in School 6
than in other schools of the sample (see Table 28).
Finally, because achievement tests were given in the fall(September 1965 and September 1966), only a limited amount ofexposure to CE activities could be represented by the test data.
92 67TMP-115
For projects initiated soon after Title I funds were made available in1966, pupils could have received as much as five months of exposureprior to the post test. If they participated in the summer reading pro-ject also, a maximum of 7 months exposure c.)uld be obtained. Sinceno project was a full-time activity, it is improbable that more than afew pupils received as much as 7 months of part-time exposure to CEactivities.
Summary of District 10 Case Study
This case study required about eight man-weeks of effort after thedata had been obtained. A return trip of about one week would be re-quired to verify the expenditure estimates. Similar studies of otherschool districts probably would require greater effort and time.
Several observations can be made about this study of CE programsof one school district:
1. No obvious relationships were found among pupil mobility, theproportions of Negro pupils in the sample schools, and at-tendance rates.
2. It was not possible to determine if CE projects had any impacton pupil attendance rates. More detailed data would be requiredto eliminate the impacts of other factors.
3. Half of the ten CE programs of this school district added modestincrements of personnel, materials, or equipment to the activi-ties of a school. It will be very difficult to discover the impactof these resources on pupil performance, as measured by testsof reading achievement.
4. Change of reading achievement may be very difficult to measureat the grade level. Only one CE project had a primary and di-rect focus on reading achievement, four others had varyingdegrees of reading orientation. Only a fraction of pupils in agrade, probably less than 25 percent, received direct impactof these projects.
5. Considerable variation of per-pupil expenditures was noted.In some cases, the variation of the pe- apil expenditure of theregular school program from year tc ar exceeded the CEincrement for either year.
6. No significant change of reading achievement was found and norelationships were discovered between change of reading testscores and type of CE activity or level of expenditure. Therewas little reason to expect substantial change in view of Vie un-limited amounts of pupil exposure represented by the pre-posttest period.
7-
SECTION 3 93
7. When detailed project information and expenditures dataare known by project and school, the allocation processis not particularly difficult. If either type of informationis lacking, it may be impossible to obtain reasonablyaccurate estimates. Without such accuracy, the probabilityof gaining meaningful results from cost-benefit compari-sons is extremely limited.
ANALYSIS OF DISTRICT 13 CE PROGRAMS
This section summarizes CE activities of District 13, describesthe selection of sample schools, presents the method used to esti-mate CE resources at the grade level, describes the characteristicsof CE projects, summarizes CE project expenditures for two gradesin each of two schools, and compares these expenditures with changesin achievement test scores.
Background of District CE Activities
Compensatory education in School District 13 consists primarilyof the comprehensive program funded under Title I, ESEA supple-mented by other federal, state, and local CE activities. For the1965-66 school year, 26 activities were organized into foiir projects:Pre-kindergarten, Remedial and Corrective Basic Skills Develop-ment, Cultural Enrichment, and S.....pportive Auxiliary Services.*Twenty-three public elementary schools and nine public secondary,continuation or adjustment schools with total enrollment of 25, 500students were included, plus four elementary parochial schools andone secondary parochial school, totaling about 2, 000 students. Thepublic schools served census tracts in which 7 percent to 22 percentof the families reported incomes under $2, 000 in 1959. In the fol-lowing year, one public elementary school and one public secondaryschool were added. Title I funding of the pre-kindergarten projectwas terminated. Budgets for these projects are shown in Table 33.
Title I projects and funds were approved early in 1966. Imple-mentation was delayed by non-availability of staff or equipment andby construction or procurement lags. Deviations from original plans
4.4"r The district reported 37 CE activities. These were summarizedto 26 for this study.
94
Table 33. Title 1 project budgets in District 13.°
67TMP-115
1
Project1
1965-66 1966-67
Pre-kindergarten $ 243,000 0
Remedial & Corrective Skills 1,075,000 911,000
Cultural Enrichment 486,000 79,000
Auxiliary & Supportive Ser. 668,000 994,000
Overhead & Fixed Charges 130,000 227,000
TOTALS $2,602,000 $2,211,000
Note:a Data were obtained from District CE plans of December 1965 and
July 1966, and from the 1966-67 Title I budget status report ofJune 1967.
were sometimes required, but expenditures for 1965-66 were re-ported as reasonably close to the budget amounts. In 1966-67 areduced budget was approved and commitment of financial supportwas not received until after the beginning of that school year. Thiscaused some disruption to the current program including some re-ductions to meet the lower budget level.
In addition to the Title I projects, this district has pursued otherCE activities with federal, state, and local support. In 1963-64and 1964-65, a pilot project was conducted in five schools, and pro-vided a wide range of activities at a funding level of $50,000 per year.A second State CE program, initiated in 1966-67, provided a three-part effort to reduce elementary school pupil-teacher ratios, buildfacilities at nine elementary schools, and provide a demonstrationreading program in one junior high school. In 1964-65 a counselingproject, supported by National Defense Education Act and local funds,was operated in several secondary schools, and provided servicesapparently similar to the Title I program.
41
SECTION 395
Sample Schools and Their Characteristics
Data on the following school and pupil characteristics wereexamined: percent eligible pupils, percent Negro pupils, percentSpanish speaking pupils, school enrollment, numbers of instruction-al personnel, and pupil mobility. Insofar as possible, schools wereselected to represent a wide range of these characteristics in orderto determine whether relationships exist between the outcomes of CEactivities and characttrisitics of the recipient pupil population. Thesecharacteristics are summarized in Table 34.
Seven elementary and four secondary schools were selected. Theseschools had enrollments from 423 to 1,918 pupils and employed from17 to 86 teachers. School populations consisted of from 11 to 91 per-cent Negro pupils and from 7 to 84 percent Spanish speaking pupils.From 8 to 19 percent of the pupils in these schools were from familieswith income under $2, 000 per year. Pupil mobility* in the sampleschools ranged from 16 to 79 percent.
bescriptions of CE Activities
This district has pursued a wide range of CE activities, some pre-ceding and others in addition to those made possible by Title I, ESEA.To estimate the allocation of resources to grades requires knowledgeof the several CE activities of the district; but their large numberprecludes detailed discussion of each one. The following paragraphspresent brief descriptions of District 13's CE activities. Thirty-sevenactivities, identified by the district, have been summarized to twenty-six activities for this study. In the process of summarizing, someactivities were combined and others, which did not serve pupils inelementary and secondary 1.des, were excluded, e.g., pre-schooland adult education acti-
Remedial and Corrective Basic Skills Activities
PRIMARY REMEDIAL READING. To provide remedial reading as-sistance to children of average or better ability but who are underachievers. This program has been provided for grades 2 and 3 for
*Defined as: (total students entering or withdrawing from a school
during the 2nd through the 9th month of the school year) (averageschool enrollment during the school year).
Tab
le 3
4.S
ampl
e sc
hool
cha
ract
eris
tics,
196
5-66
.
Sch
ool
Cha
ract
eris
tics
Sch
ool N
umbe
r
I2
34
51
67
89
1011
Enr
ollm
ent
825
601
546
851
423
830
821
1085
1603
1360
1918
Inst
ruct
iona
lP
erso
nnel
2420
1732
1529
3252
7862
86
Pup
il M
obili
ty (
%)
7977
6156
3978
5638
4632
16
Neg
ro P
upils
(%
)8
2456
8011
4491
7059
6923
Spa
nish
Spe
akin
gP
upils
(%
)14
2111
1284
407
1432
1623
Per
cent
of P
upils
from
fam
ilies
with
inco
me
unde
r $2
000/
yrin
195
9
810
1212
1819
1910
1510
9
.I
, ,,,
SECTION 3 97
several years with district funds; Title 1 support permitted expansionof this program to provide additional teacher specialists for eligiblepublic and parochial schools and to include pupils in grades 1 and 4.
READING ADJUSTMENT CLASSES. To help students in grades 7and 8 who have average or better intelligence but whose readingachievement is at least 2 years retarded. Reading problems werediagnosed and pupils placed in groups of 8 under 2 teachers withspecial training or experience. Teacher aides and additional equip-ment were provided. Title I fun.Ls supported this project in 3 schools.
REMEDIAL READING. Similar to the above project, to meet needsof pupils in grades 7 to 12 with lesser reading retardation, in groupsof 15 to 18 pupils. Already a distr'ct-supported project, ESEA fundsprovided additional classes in 6 schools.
ENGLISH AS A SECOND LANGUAGE. To help non-English speakingpupils to develop fundamental English language skills and to preparethem for regular school programs. An aural-lingual approach wasused at beginning, intermediate, and advanced instructional levels.A class of 18 pupils was provided with ._' teacher aide or assistant,tutors, materials, and equipment. Some portable classrooms wereobtained. This is an expansion of an existing project.
CLASSES FOR EDUCATIONALLY HANDICAPPED PUPILS. A re-mediation project to provide more attention to problems of academic-ally handicapped pupils of average or graater ability but retarded inbasic skills. Small groups of pupils (15-20) in elementary gradeswere assigned to a teacher on a full-day schedule. A resource teach-er, materials, in-service training and portable classrooms wereprovided. Title I funds enlarged an existing activity.
Cultural Enrichment Activities
CULTURAL ENRICHMENT. A number of activities were providedto motivate and stimulate elementary and secondary pupils of dis-advantaged areas, inclueing: art and music classes, enrichedsummer session and Saturday classes for talented pupils, assemblyprograms and cultural exchange, study trips, oral communicationsskills festival, equipment for remedial use and cultural enrichmentin business education, and equipment for instruction in home manage-ment and personal development.
98 67TMP-115
LIBRARY SERVICES. ESKA. funds were used to expand a modestdistrict program-to_provide and equip libraries in elementary schoolsand secondary target area schools and to provide personnel forgreater library availa.bility.
CLASSES FOR MORE ABLE PUPILS. More able pupils in grades3-6 were grouped into an enriched and accelerated program in fourcenters. A resource teacher, field trips, visiting teachers, specialequipment, and in-service training wr:re provided.
Auxiliary and Supportive Services Activities
HEALTH SERVICES. Speech and hearing teachers were hired to in-crease speech and hearing therapy in Title I schools. Additionalnurses and a clerk were added to increase nursing time provided atTitle I schools.
COUNSELING PROJECTS. Five program activities provided addi-tional professional or clerical personnel and materials to intensifycounseling toward solving academic, mental, or social problems ofpupils and parents.
AUXILIARY TEACHER SERVICES. After a pilot project, additionalteachers were appointed to assist classroom teachers by: substitutingfor the regular teacher to permit small group instruction; lessonplanning; providing remedial reading instruction to individual pupilsor small groups; serving in the library; confering with parents; andassisting in school/parent/community activities.
TEACHER ASSISTANTS, AIDES, AND CLERKS. Assistants, aides,and clerks were provided in selected secondary and elementary tar-get schools to relieve teachers of non-instructional classroom dutiesor clerical work and to permit more individualized instruction.
KINDERGARTEN AIDES. After a successful pilot project, parenthelpers were employed as teacher aides to alleviate some of theproblems of overcrowded classes in project schools. They helpedto prepare instructional materials, to improve the classroom environ-ment, and to supervise classroom and playground activities.
SECTION 3
EXTENDED TIME AND TRAINING ACTIVITIES. Three programactivities provided for orientation and in-service ft-a...fling of new
teachers and for continuing professional advancement through con-
ferences, courses, and workshops.
99
ADMINISTRATIVE SERVICES. Target schools have more office
referrals by teachers and administrative transactions. Vice prin-cipals were appointed in target area elementary schools to strengthenadministration and permit the principal to devote more attention to
curriculum development.
CENTRAL OFFICE ACTIVITIES AND FACILITIES. Two projects
were defined to provide for centralized district office services and
facilities. One described capital outlay for remodeling, construction,
or relocation of portable classrooms employed in projects describedabove. The other provided funds for administrative and support ser-vices of central offices which were necessary for the management,coordination, and evaluation of the overall CE program.
Other CE Activities
1963-64 PILOT PROJECT. A pilot project was initiated in 1963-64
in three elementary and two junior high schools to improve communi-cation skills and motivation, to aid pupils in understanding theirabilities, to strengthen teacher understanding of pupils' problems,to improve school relations with parents, and to provide pre-schoolexperiences.
STATE CE PROJECT. In 1966-67, a state-district supported pro-ject was pursued in nine elementary and one secondary school to
reduce pupil-teacher ratios and to obtain portable classrooms. Ademonstration reading program was initiated in one junior high school.
NDEA COUNSELING PROJECT. During 1964-65 and 1965-66, acounseling project was supported in several secondary schools byNDEA funds to raise educational and vocational aspirations, to in-crease home-school contacts, and to motivate pupils to remainschool. Activities included extended individual and group guidance
sessions with pupils and a parent participation program.
Description of the Analysis
The approach pursued in this case study was designed in accord.-
ance with the type and volume of available program information and
100 67TMP-115
expenditure data. As defined at the district level, many CE "activi-ties" actually were resource inputs to a CE project. It was necessary,therefore, to examine several sources of information such as projectplans, activity descriptions, budget documents, and evaluation reports,in order to arrive at an operational description of a project at theschool and grade ley&
Financial re...ords did not distinguish expenditures by schools.Special accounting for Title I funds employed resource categories(e.g remedial reading teachers, teaching equipment) and schooladministrative divisions: elementary, secondary, student services,curriculum services, post high school, and overhead. The CE pro-.jects were subdivided by the district according to types of servicesprovidea: pre-kindergarten, remedial and corrective development,cultural enrichment, and auxiliary and supportive services. It was thennecessary to define the CE activities pursued in a grade, to cumulatea catalog of resource inputs from several sources, and to estimatetheir costs. Use of district-wide cost factors was required, such e saverage teacher salaries and per-pupil expenditures for instructionalmaterials or equipment. School and district personnel reports werea primary source of resource data.
A data form was designed on which to record information soughtfor subsequent regression analyses. This form permitted aggrega-tion of info,mation at three levels: all eligible schools, a particularschool, and a selected grade. The following time periods were usedas needed; the school years ending June 1965, 1966, and 1967; sum-mer sessions 1965 and 1966. In addition, since Title I began in themiddle of school year 1965-66, a separate time period, Spring 1966,was used for ar:tivities which occurred only in the latter half of thatschool year. The 26 CE activities identified in this study were thencategorized into ten types of projects to pe7mit comparisons on abasis more consistent with projects of other districts. They were:
1. Remedial reading and language arts
2. Other specific instri..ctional projects (e.g., math, science)3. Administrative changes, such as team teaching or modular \
scheduling
4. Vocation-related projects5. Staffing increments not related io specific instructional
pro:acts
SECTION 3
6. Non-instructional equipment and supplies
7. Staff development
8. Health and guidance services
9. Community-parent services10. Cultural enrichment.
Finally, characteristics of staff, students and activity costs wereincluded as follows:
1. Number of equivalent full-time teachers, by period2. Number of direct pupil beneficiaries
3. Ave7age size of participating ?upil group
4. Average hours/week per participant5. Weeks of project exposure per pupil
6. Instructional salaries7. Other expenditures.
101
A brief description was prepared for each of the 26 CE acti.itiesand the data form was completed for each grade in the sample schools.An attempt was made to provide the required quantitative informationat the lowest organizational level for which program information wasavailable. When possible, these quantitative descriptions were basedupon specific information at the grade level. When grade level datawere lacking, school or district level information was used to allocateexpenditures to grades, employing the following assumptions:
1. The number of children in a grade was defined as the total ofpupils listed in active enrollment reports of that grade pluspupils in ungraded classes allocated to that grade by progr..1mcoordinators.
2. Allocated staff positions were considered filled.3. Services described at the school level were applied on a per
pupil basis, as appropriate.4. Budgeted amounts were used to estimate expenditures.5. Information provided in CE program descriptions was accepted
as describing the CE activities.
1
102
The allocation process revealed that many uncertainties still ex-isted on the part of the authors with respect to information concerningservices received, despite the fact that this district was chosen forhaving relatively complete records.
Illustration: Primary Remedial Reading
This illustration demonstrates the methods for estimating CE re-sources at the grade level for a specific activity. The particular casereflects the general observation that the actual application of CE re-sources varies widely among schools and grades in terms of absoluteamounts of CE activities undertaken. Further, considerable varia-tion also exists within a grade through time.
For a number of years prior to Title I, District 13 has conducteda remedial reading program for second and third grade "underachievers"of average or better ability. This program provides specially trained.remedial reading teachers in all elementary schools. These teachersare assigned on a half-day, nine-week basis through the 36 week schoolyear. They work with groups of 10 children for one period each day.These sessions are in addition to regular classroom work in reading.In theory, a child participates in the program for nine weeks, returnsto his regular schedule for nine weeks, and then, if necessary, re-ceives nine more weeks of remedial help.
With the addition of Title I funds, six remedial reading teacherswere added to intensify existing services to eligible public and pa-rochial schools. From February to June, 1966, this supplementalprogram was divided into three six-week sessions, serving grades1-4. Pupil participants were selected by staff members of theschools. During the 1966-67 school year, a state-funded programprovided additional teachers for this remedial reading program.These teachers were not assigned to target area schools, but toother scIlools in the dist-ict. To perform the required calculations,the following steps were taken and assumptions made:
1. Estimates of instructional staff were derived from districtrosters of Primary Remedial Reading teachers. Each'eacher is assigned to a school on a half-day (three periodsor less) basis. Normally the school year is divided intofour nine-week periods of instruction. Thus, it was assumedthat staff were added in increments of one-eighth (one-fourthyear x one-half day) of a Full-Time Equivalent (FTE) teacher.
SECTION 3 103
How , the 18-week first semester of Title I was dividedinto 3 equal parts, which yields a FTE factor of one-twelfth(one-sixth year x one-half day).
2. Salary extimates were based on average regular teachersalary in 1965-66, $8,100.
3. Salary costs were apportioned on the basis of the number ofpupil participants in a grade.
4. Other costs of the program were considered insignificant, asreported by the program director.
5. Pupil participation by grade and school was obtained fromreIN-Irts made by the teachers.
Table 35 contains the results of the above calculations for this CEactivity in one sample school. It illustrates the amount and variationsof CE resources among grades and time periods. Similar tabulations,not shown here, of grade level expenditures for other activities andschools indicate that similar variations cften occur. One reason forthese observed variations is the fact that, although some activitiesmay apply equally to all eligible pupils, others apply only to certainpupils, grades or schools. Some activities contin-ted at the samelevel with the initiation of Title I support; others expanded.
These variations were compounded by changes in the numbers ofregular classroom teachers and in pupil enrollment. Examples ofthese variations are included in the following detailed studies of theCE activities and achievement changes of selecteu fifth and sixth gradecohorts of pupils.
Case Study of Sixth Grade Cohorts
Until the grade level descriptions are more fully verified, it ispremature to examine in depth the relationship between CE inputs andpupil performance. However, it is possible to illustrate the types ofcomparisons that could be made by the following case study. In thestudy, summaries of CE expenditures were made for fifth and sixthgrades in each of two schools where a sharp contrast in achievementtest results was noted. The cha.:ges of mean achievement were asfollows (1966-67 scores minus 1965-66 scores in Standard T units):'
The Reading subtest of the Stanford Achievement Test was used forthe sixth grade and the Paragraph Meaning subtest o: the SequentialTest of Educational Progress for the- fifth grade.
104
-
67TMP-115
Table 35. Estimated CE resources for primary remedial readingSchool 3.
TimePeriod' Grade
Ful 1 -Ti me
EquivalentTeachers
PupilParticipants
Estimated
I Expenditurec
1 0 0 0
1964-65 2 0.2 21 $1300
3 0.1 10 700
4 0 0 0
1 0 0 0
1965-66d 2 0.2 19 1300
3 0.1 12 700
4 0 0 0
1 0.1 20 500
1965-66 2 0 0 0
Title I peri od 3 0 0 0
4 0.0 10 200
1 0.1 21 700
1966-67 2 0.2 31 2000
3 0.3 30 2000
4 0.1 10 700
Notes:
°Project not pursued during summer peliodsbThe activity was designed for groups of 10 pupils to receive 5 hours/day ot re-
medial instruction for 2 alternate quarters/period,. During the Title 1 portion of
1965-66 three 3-week classes weie added, and in 1966-67 there was consider-
able irregularity in schedulingc Based on a constant average teacher salary. Other costs were considered
negligible.d Listed here are activities which extended throughout the academic year.
eListed here are activities restricted to the special time periods established when
Title 1 began in the spring of 1966.
SECTION 3
IN Mean AchievementSchools Grades
5 6
5 -1. 1 +1 . 2
7 -3. 4 +5. 7
Difference +2. 3 -4. 5(School 5 m'-zus School 7)
105
There are three observations which should be made wit!, respectto this table:
1. At the sixth grade level, School 7 showed a considerable gaincompared to School S.
2. At the fifth grade level, both schools showed a decline inachievement, with School 7 showing a greater decline thanSchool 5. The difference between the declines experiencedby School 5 and School 7, however, is quite small.
3. Between the fifth and sixth grades a reversal of the relation-ship between School 5 and School 7 occurred. While at thefifth grade level, the difference between the schools is notlarge, the absolute total of the differences for both.gradesis substantial. However, it should be noted that the com-parisons for 5th and 6th grades are not independent. Theearlier fifth grade class in the 5th grade comparison is thelater 6th class in the sixth grade comparison. Thus, anunusual early fifth grade class could, by itself, have accountedfor the change in the direction of the relationship between the-
fifth and sixth grade classes. This lack of independence,while it affects the reliability of conclusions reached on thiscase study, nevertheless illustrates how CE expenditurescan be compared with achievement test results. Other com-parisons, such as the absolute magnitude of the changes, arenot appropriate due to different test dates, test intervals, ortest forms.
In developing estimates of resource inputs, the grades were viewed
as cohorts, i. e., groups of pupils which receive a sequence of ex-perience through time. The ideal cohort would have stable member-ship throu6 time, but these groups did have changes of membership.For these comparisons, the 6th grade cohort refers to classes which
move from grade to grade and were in the 6th grade when tested inrespective years. Therefore, when considering a sixth grade cohort,
106 67TMP-115
the services received when this group was in the fifth grade must alsobe borne in mind.
Differences of CE services provided to the two cohorts were ex-amined in the grade tested and in earlier years. The types of com-pensatory education activities provided to the sixth grade cohorts atSchools 5 and 7 are shown in Table 36. There were many activitiesserving both groups of sixth graders. However, for many of thesethere is no basis for differentiating between the experiences at thedifferent schools; information was district wide and therefore pre-*sumably applied equally in the two schools.
For this illustration, it was assumed that the types of compensa-tory education most likely to be related to reading achievement testscores are: (1) Remedial Reading and Language Arts, (2) Other Spe-cific Instructional Programs, and (3) Staff Increments. Consequently,the analysis was restricted to these three types of CE activities; thespecific activities of these types which were considered in this analysisapp_ar in Table 37. The procedures used in combining expendituresare given below, and include the various assumptions which had to bemade.
The following procedures were used in summing expenditures for6th grade cohorts. The expenditures discussed are summarized inTable 37.
1. To make descriptions at the level of the district or the schoolapplicable to particular grades it was necessary to allocatesuch services on a per pupil basis. Table 38 shows demo-graphic characteristics of the successive 6th grades at Schools5 and 7. For the purpose of comparing services in the ele-mentary schools of different size, vve assume no differencein per pupil level of service unless there is contrary evidence.
2. It is assumed that the CE services which occurred in the yearpreceding the achievement tests are those which have impactupon the achievement test scores. For the sixth grade thetwo sets of achievement test compared are those administered
ert
Whether services were designed to be equal for all pupils or equalfor all schools would make a difference. It was assumed that serviceswere provided on a per pupil basis.
Tab
le 3
6. C
E a
ctiv
ities
ser
ving
six
th g
rade
coho
rts
and
leve
ls o
f det
ail o
f ava
ilabl
ein
form
atio
n.
CE
Act
ivity
Per
iod
of S
ervi
ces
Pro
vide
dLe
vel o
f Det
ail
Sch
ool 5
64-6
5 65
-66
66-6
7
Sch
ool 7
64-6
5 65
-66
66-6
7
Eng
lish
as S
econ
d La
ngua
geE
duca
tiona
lly H
andi
capp
edR
eadi
ng C
linic
Aud
io-V
isua
l Equ
ipm
ent
Impr
ovin
g In
stru
ctio
n in
R&
CS
kills
aF
acili
ties
for
Rem
edia
tion
Cla
sses
for
Mor
e A
ble
Libr
ary
Ser
vice
sC
ultu
ral E
nric
hmen
tC
entr
al C
E S
ervi
ces
& O
verh
ead
a
Inte
nsifi
ed V
isiti
ng T
each
er S
ervi
ceG
uida
nce
Clin
ic°
Add
ition
al N
ursi
ng S
ervi
ces
Aux
iliar
y T
each
er S
ervi
ces
Ele
men
tary
Sch
ool A
dmin
istr
atio
n S
ervi
ces
Tea
cher
Ass
ista
nts
& A
ides
Res
ourc
e H
elp
to N
ew T
each
ersa
Add
ition
al C
leric
al A
ssis
tant
sIn
-Ser
vice
Edu
catio
n°C
urric
ulum
Dev
elop
men
t°S
tate
P/T
Red
uctio
n P
roje
ct
xx
xx x
xx
xx
x xx
xx
xx
xx
xx
xx
xx
xx
xx
xx
xx
xx
xx
x x
xx
x xx
xx
xx
xx
x xx
xx
xx
xx
xx
xx
xx
xx
xx
xx
xx
xx
xx
x x
Sch
ool
Gra
deG
rade
Sch
ool
Sch
ool
Sch
ool
Gra
deS
choo
lS
choo
lD
istr
ict
Sch
ool
Dis
tric
tS
choo
lS
choo
lS
choo
lS
choo
lD
istr
ict
Sch
ool
Dis
tric
tD
istr
ict
Gra
de
Not
e:a
Exp
endi
ture
s fo
r th
ese
activ
ities
wer
e as
sum
ed to
be e
quiv
alen
t bet
wee
n si
xth
grad
e co
hort
s of
1966
-67.
.1IM
P
108 67TMP-115
Table 37. Summary of CE expenditures for sixth grade cohorts.
1
School 5, Sixth Grade School 7, Sixth Grade
e-F A,tivitiPc 1965-66 1966-67 1965-66 1966-67
English as a Second 4)et 1 CAAI JUV $1500
Language
Readi ng Clinic $1200
Educational ly Handicapped 1600 $8100 8100
Classes for More Able 2400
Auxiliary Teacher Service 300 200
Elementary SchoolAdministration Service a
Teacher Assistants and Aides 200 200
Additional Clerical Assistance 200 100
TOTALS $1500 $3800 $8100 $12200
Estimated Number of PupilsReceiving Service 54 54 68 61
Expenditure Per Pupil $28 $71 $119 $200
Difference BetweenSuccessive Cohorts $43 $81
Note:
° Negligible .
SECTION 3
Table 38. Demographic characteristics of the sixth gradesof Schools 5 and 7.
Characteristic
1965-66 1966-67
School 5 School 7 School 5 I School 7
Number of regular & adjustmentsixth graders (February) 46 62 45 58
Sixth grade pupils as percent oftotal enrollment 12% 8% 11% 8%
Sample size for achievementtests (January 1966 andOctober 1966) 34 62 42 62
Percent of sixth graders tested 74% 100% 93% 107%
[Number of classroom teachers 1 1-1/2 1 1-1/2
in January 1966 and in October 1966. For the first test, thetime interval for which to establish the level of serc:rice isfrom January 1965 to January 1966. This period includes noTitle I expenditures. For the second test, the time of serviceis from October 1965 to October 1966. This period is thesimpler to examine since it is approximately equal to academicyear 1965-66 and the summer of 1966.* This includes the"first year" of Title I. Thus, the analysis consists of a com-parison of the January 1966 scores of the cohort which was inthe 6th grade in the 1965-66 school year with the October 1966scores of the cohort which was in the 6th grade during the1966-67 school yc;ar. Since each cohort was assumed to havebeen supported by CE programs during the year preceding thetest, programs initiated after January 1966 were not consideredas having affected the earlier cohort for purposes of the com-parison, even though they were in fact applied to that cohortlater in the school year.
T For convenience, this assumption is made for all but the regularclassroom teachers' salary, which is calculated more precisely.
110 67TMP-115
3. Although English as a Second Language at School 5 was notfunded by Title I.in grade 6, this 7...:tivity was expanded byTitle I elsewhere in the district and is included in the analysis.The rate of expenditure was the same at School 5 for the pastseveral years. The service for the sixth grade cohort wasestimated as approximately 15 percent of the service given tothe school.* Even though average teachers' salary changedover years, a constant 1965-66 salary level was used for eachyear to prevent variz.tion due to cost of the service rather thanquality or amount (Referenc- 9).
4. For the Educationally Handicapped and Adjustment classes atSchool 5, there was no service prior to January 1966. Thus,there were no expenditures for the earlier cohort. For thelater sixth grade cohort, there was no service at the sixthgrae level in 1966-67. There was, however, service to thatsame group during the last half of the preceding school year,when that cohort was in the fifth grade. This service con-sisted of a class of 18 pupils drawn from the fourth throughsixth grades. More exact information about participation wasnot obtained. Therefore, one-third share of the expenditurefor this service was assigned to the later 6th grade cohort.At School 7, the same level of service existed for the fifthgrades in 1964-65 and 1965-66. Therefore each cohort re-ceived the same amount of service.
5. In School 7, Classes for the More Able were begun in Febru-ary 1966. Therefore, this service was not considered teapply to the earlier sixth grade cohort; the service for thelater sixth grade cohort was that received during the Title Iperiod of 1965-66 by the fifth grade. School 5 did not havethis service.
More pIccise estimates of the amount of service to the later andearlier cohorts would be the proportion of the total school which wasin the fifth grade in 1965-66 and the average of the fifth grade in1965-66 and the sixth grade in 1965-66. These estimates, based onFebruary enrollment figures, would be 14 percent and 13 percentrespectively. Because it is not known whether special classes orkindergarten classes were eligible for this program, an approximatefigure constant for all cohorts, was used in the present illustrativecalculations
SECTION 3 111
6. The Auxiliary Teachers Service, which began in February1966, was considered to apply only to the later sixth gradecohort of each school. Each school had a half-time auxiliaryteacher. The share of the cost which applies t^ th- Ath gradecohorts would be about 12 percent for School 5 and 8 percentfor School 7, based upon the percentage which the 6th gradeconstitutes of the school.
7. For Teacher Assistants and Aides who were assigned on 7,
per-school basis, the appropriate percentages that each sixthgrade was of the total school were applied to obtain arralloca-tion for the sixth grade cohorts. For School 7, informationabout teacher assistants in 1965-66 had not been receive(1, bythe time this analysis was performed so that the assumptionthat there were no services in that time .period is questionable.The services from teacher assistants and from aides wereadded together.
8. For Additional Clerical Assistants, the same proportions wereused to allocate service to the sixth grades. The difference inlevel of service favoring School 5 results because School 5 isa smaller school and it was assumed that there was more perpupil benefit for a given level of service at the smaller school.
9. The state project fo reduce pupil-teacher ratio began in 1966-67, and did not directly provide a sixth grade teacher.'
10. The amount for Elementary School Administrative Serviceswas negligible for the 1966-67 sixth grade cohort at School 7,the only school to which it applied.
The primary interest with respect to the expenditures of CE fundsis the change in expenditures per pupil through time. Based on the
*One teacher at Schooi 7 was actually paid from state compensatory
education funds. Since the project concerned was aimed at reducingpupil-teacher ratio in a range of grades, the particular grade assign-ment is incidental, for accounting purposes. It is more reasonableto regard the proportion of all teachers supported by CE funds as thelevel of CE services. This would amount to 20 percent in School 7and 10 percent in School 5 in 1966-67. To simplify the presentation,this CE project is included in Regular Classroom Teacher Services.
112 67TMD-115
above a4sumptions9 the amounts expended in the schools in the yearprior to the achievement tests were computed and are shown in Table37. To determine the expenditures per pupil, it was necessary todetermine the appropriate number of pupils serviced; this was per-formed as follows. For the earlier cohort, the average of (1) enrol-lment in the regular, Educationally Handicapped (EH) nid Adj,ictm,.nt5th grade classes in 1964-65 and (2) enrollment in regular, EH, andAdjustment sixth grades in 1965-66 was used. For the later cohort,enrollment in regulars EH and Adjustment fifth grade classes in1965-66 was accepted. The number of pupils serviced is shown inTable 39, and the cost per pupil for each cohort is shown in Table 37.The.level of CE services, as represented by expenditures per pupil,rose almost twice as much for the 6th grade cohort of School 7 be-tween 1965-66 and 1966-67 ($81 per pupil) as it did for the samecohort at School 5 ($43 per pupil).
Any comparison cf services to pupils provided by different schoolswould be incomplete if it were resti.cted to CE programs and failedto consider possible differences in regular classroom instructionfurnished by those schools. In order to provide for the support fur-nished by regular classroom teachers, it was necessary to cost theirservices for the relevant time period, i. e., the year preceding thetwo achievement tests. Since this service represented a.larger sumof money than was involved in the CE program, the approximationsof the support period used for the CE programs were considered toogross, and a more precise period was employed. In order to coincidewith the support periods for January 1966 and October 1966 test dates,the time periods used were:
Earlier cohorts:
Later cohorts:
The number of teachers
1/2 fifth grade 1964-65 academic year + 1/2sixth grade 1965-66 academic year8/9 fifth grade 1965-66 academic year + 1/9sixth grade 1966-67 academic year
at each grade level each year was:
School 5 School 71964-65 1965-66 1966-67 1964-65 1965-66 1966-67
Grade 5 1-1/2* 1-1/2 1-1/2 2 2 2-1/2
Grade 6 1 1 1 1.-1/2 1-1/2 1-1/2
* Fractional teachers arise when pupils from two grades are assignedto a teacher.
Tab
le 3
9.R
egul
ar c
lass
room
teac
her
expe
nditu
re fo
r si
xth
grad
e co
hort
s.
Sch
ool 5
.S
choo
l 7
Six
th G
rade
in 1
965-
66S
ixth
Gra
dein
195
6-67
Six
th G
rade
in 1
965-
66 in
Tea
cher
sal
ary
per
coho
rtin
yea
r pr
eced
ing
achi
vem
ent
test
s
Est
imat
ed n
umbe
r of
pup
ils in
coho
rt w
hen
rece
ivin
g se
rvic
e
Tea
cher
sal
ary
per
pupi
l
$10,
100
(24
+ 3
0)
$187
$11,
650
54
$216
$14,
150
(32
+ 3
6)
$208
Diff
eren
ce b
etw
een
succ
essi
veye
ars
(196
6-67
1965
-66)
$29
$50
Diff
eren
ce b
etw
een
scho
ols
$21
Not
e:a
Six
th G
rade
1966
-67
$15,
744
61
$258
^,11
1
For
196
6-67
, the
enr
ollm
ent o
f the
six
th g
rade
coh
ort i
s: fi
fth g
rade
reg
ular
and
adj
ustm
ent p
upils
in F
ebru
ary
1966
.F
or 1
965-
66, t
he e
nrol
lmen
t of t
he s
ixth
gra
de c
ohor
t is:
1/2
(si
xth
grad
ere
gula
r an
d ad
just
men
t pup
ils in
Oct
ober
196
5) p
lus
1/2
(fift
h gr
ade
regu
lar
and
adju
stm
ent p
upils
in M
arch
196
5). T
he g
rade
dis
trib
utio
n of
Adj
ustm
ent c
lass
es p
rovi
ded
by E
xcep
tiona
l Chi
ldS
ervi
ces
Offi
ce w
as g
ener
ally
acc
epte
d, a
nd a
pplie
d to
the
enro
llmen
t fou
nd in
the
Rep
orts
of
Act
ive
Enr
ollm
ent.
CZ
114 67TMP-115
Thc cos,. of teacher services was priced at the average elementaryschool teacher salary in 1965-66, $8, 100,
The calculations of teacher service for the 6th grade cohortswere:
School 5 in 1965-66 1/2 x 12100 + 1/2 x 8100 = $10, 100
School 7 in 1965-66 1/2 x 16200 + 1/2 x 12100 = $14, 150
School 5 in 1966-.67 8/9 x 12100 + 1/9 x 8100 = $11, 650
School 7 in 1966-67 8/9 x 16200 + 1/9 x 12100 = $15,744
The increasing amount of service for the later cohort is due to thegreater amount of time spent at the 5-"a grade level, which has agreater amount of teacher service.
The calculations of expenditures per pupil are shown in Table 39.The greater increase in regular classroom expenditures pei pupilfor School 7 is due to the decrease in the number of pupils in the
cohort.
Table 40 summarizes the expenditures per pupil for CE and class-room teacher salary. In both amounts of expenditures per pupil andincreases over the preceding year, School 7 pupils are faring muchbetter than those in School 5. These results are particularly interest-ing because they are consistent with achievement results describedearlier in this section.
Case Study of Fifth Grade Cohorts
A similar analysis was made for grade 5 of the same schools. Inthis comparison, a different time interval for the CE service for theearlier cohorts is involved since testing occurred in October in suc-cessive years. Thus, the appropriate support period for the earliercohorts extends from October 1964 to October 1965, and that for thelater cohorts from October 1965 to October 1966. As for the sixthgrade comparisons, these time intervals were considered to includethe academic year and summer of 1964-65 and 1965-66, respectively.
Differences in CE expenditures between fifth and sixth grade cohortsoccurred for only one activity. For the Educationally Handicapped (EH)
SECTION 3
Table 40. Expenditures per pupil for sixth grade cohorts.
115
School 5 School 7
1965-66Cohort
1966-67Cohort
1965-66Cohort
1966-67Cohort
CE Activities
Regular ClassroomTeacher Salary
TOTAL
$ 28
$187
$ 71
$216
$119
$208
$200
$258
$215 $287 $327 $458
Difference BetweenSuccessive Cohorts
$72 $131
activity, the level of service to the fourth grade in School 7 was lessthan that received by the 5th grade.* (See Table 41. )
To convert the amount expended for a cohort to a per pupil levelof service, the enrollment of regular, EH, and Adjustment classesat the fourth grade level of the preceding year was taken as the sizeof the cohort receiving service.
The level of CE service increased at each school for successivecohorts. As was true for the sixth grade, the increase for the fifthgrade cohort was greater at School 7 than School 5. The difference,however, is only about a third as great as it was for the sixth grade.
*There is also a shift in level of service between 1965-66 and 1966-
67. Taking into account about 1/9 of a year (for the service in 1966-
67) at half the level of expenditure would result in increase of about$450 to the 1965-66 total at School 7. Similarly, for the sixth grademaking this alteration in time intervals would result in a subtractionof the level of service to the 1966-67 sixth grade cohort at School 7of $450. In neither case does this correction alter results appreci-ably.
116 67TMP-115
Table 41. CE expenditures for sixth grade cohorts at Schools 5 and 7.
CE Activities
School 5, Fiith Grade School 7, Fifth Grade
1965-66Cohort
1966-67Cohort
1965-66Cohort
1966-67Cohort
English as a SecondLanguage $1,500 $1,500
Reading Clinic $1,200
Educationally Handicapped 1,600 $4,000 4,000
Classes for More Able 2,400
Auxiliary reacherService 300 200
Elem. School Admin.Service a
Teacher Assistants andAides 200 200
Add. Clerical Assistance 200 100
TOTALS $1,500 $3,800 $4,000 $8,100
Est. No. pupi ls in cohortwhen receiv'Jng services 58 68 69 79
Expenditure per pupil $26 $56 $58 $103
Difference between suc-cessive cohorts $30 $45
Difference between schools $15
Note:
°Negligible.
SECTION 3
As in the case of the sixth grade cohorts, the costing of regularclassroom teachers was based on the ov.-3 year supporting periodprior to the date of the achievement tesi. Since achievement testswere given in October, the support perio..1 consisted of eight monthsof the preceding school year and only one month (September) of theschool year in which the test occurred. The time periods used were:
Earlier cohort: 8/9 fourth grade services in the 1964-65academic year + 1/9 of fifth grade servicesin the 1965-66 academic year.
Later cohort: 8/9 fourth grade services in the 1965-66academic year + 1/9 of fifth grade servicesin the 1966-67 academic year.
Levels of teacher service varied, in number of teachers assignedper grade level, as follows:
School 5 School 7
1964-65 1965-66 1966-67 1964-65 1965-66 1966-67
Grade 4 1 1-1/2 1-1/2 2 2 3
Grade 5 1-1/2 1-1/2 1-1/2 2 Z
Using the average teacher salary oi $8, 100, regular teacher servicefor the four fifth grade schools was as follows:
School 5 in 1965-66 8/9 X 8100 + 1/9 x 12100 = $8,544
School 7 in 1965-66 8/9 x 16200 + 1/9 x 16200 = $16,200
School 5 in 1966-67 8/9 X 12100 + 1/9 x 12100 = $12,100
School 7 in 19(6-67 8/9 X 16200 + 1/9 X 20200 = $16,644
* One teacher at School 7 was actually paid from state compensatoryeducation funds. Since the project concerned was aimed at reducingpupil-teacher ratio in a range of grades, the particular grade assign-ment is incidental, for accounting purposes. It is more reasonableto regard the proportion of all teachers supported by CE funds as thelevel of CE services. This would amount to 20 percent in School 7and 10 percent in School 5 in 1966-67. To simplify the presentation,this CE project is included in Regular Classroom Teacher Services.
118 67TMP-115
There is an increasing amount of service at each school, but when
the increases in pupil enrollment lre considered there is a sharpcontrast in changes in expenditures per pupil: School 5 rises andSchool 7 drops. (See Table 42. )
When the change in expenditures per pupil for regular classroomteachers is combined with the change from compensatory education,
School 5 is seen to have an increase in expenditures per pupil of $61,
while School 7 has an increase of $21. (See 'I. ,e 43. ) This contrastin favor of the fifth grade cohort at School 5 is also consistent with
the achievement test results.
Summary of District 13 Case Study
This case study concerned itself with two sets of cohortscohortsin the fifth grade in both the 1965-66 and 1966-67 school years andcohorts in the sixth grade in the same two school years. Two schools
were included in the analysis. For each tet of cohorts, a comparison
was made between changes in achievement test scores over the twa
years and changes in the amounts of selected resources per pupil ex-
pended on the two cohorts. The expenditures used included estimatesof both funds expended explicitly for selected GE projects and thosefor regular classroom instruction for the year immediately precedingthe test date, since it was assumed that the test scores should reflect
the services provided during that period.
The case study results showed that
1. For the sixth grade collorts, increases in expenditures perpupil at School 7 were greater than at Scnool 5. Similarly,increases in achievement test scores at School 7 were greaterthan at School 5.
2. For the fifth grade cohorts, the relationships between School 5
and 7 are reversed. Achievement changes now tend to favor
School 5, although the differe .ce is small. Similarly, the in-crease in expenditures per p-L.,3i1 was greater at School 5 than
it -was at School 7.
3. Although the difference between the changes in achievementscores Lir the fifth grade cohort at School 5 and School 7 wassmall, it represents a substantial reversal of the relationshipwhich existed between the two schools for the sixth grade co-horts. It should be borne in mind, however, that the two cohorts
were not independent, and thus this reversal might well have
been the result of the unusual performance of the cohort common
to both gm:des.
SECTION 3
Table 42. Regular classroom teacher expenditure for the fifth gradecohorts at Schook 5 and 7.
119
R
School 5 School 7
1965-66 1966-67Cohort Cohort
1965-66Cohort
1 966-67Cohort
Teacher salary per cohortin year preceding achieve-ment tests
Est. no. of pupils in cohortwhen receiving service
Teacher salary per punii
$8,544 $12,100
58 68
147 178
$16,200
69
235
$16,644
79
211
1
Difference between successiveyears (1966-67 1965-66)
Difference between schools
+$31
$55
-$24
Table 43. Expenditures per pupil rur fifth grade cohorts.
School 5 School 7
1965-66Cohort
1966-67Cohort
1965-66Cohort
1966-67Cohort
CE Activities $ 26 $ 56 $ 58 $ 103
Regular ClassroomTeacher Salary 147 178 235 211
TOTAL $ 173 $ 234 293 $ 314
Difference between successive $61 $21
cohortsI
120 67TMP-115
Approximately 12 man-weeks of effort was required to pursue thiscase study. This effort is probably representati-, of thec\time andmanpower required to develop CE resource descriptions for schooldistricts for which expenditures are not reported by school and grade.In the case study just described, the number and nature of the assump-tions required to allocate resources by grade and project were suchthat additional analysis and verification is certainly required. To dothis would require revisiting District 13 for a period of at least oneweek.
CONCLUSIONS FROM THE CASE STUDIES
Compensatory education programs of two school districts havebeen examined in detail in order to determine the feasibility of identi-fying problems in and defining the effort required for d.:tscribing CEprogram activities and estimating the expenditures of resources atgrade level, so that they might be compared with measures of pupilperformance.
Different anafytical approaches were required because of differ-ences in data obtained from school districts. In one case, districtlevel CE project information was combined with expenditure data byschool and project to make estimates of project expenditures by grade.In the other, such info-mation was not available by schools, and spe-cific detailed information had to be obtained for each kind of activityat the school and grade level. In each case study, wide variations of
s'ervices were found among schools and grades. TI- need for ex-tensive information projects and activities was confirmed. In eachanalysis, the a,isumptions and interpretations were such that the esti-mates derived should be used with caution until verified to the extentpossible by personnel in the school districts.
To illustrate the use of these resource estimates, selected com-parisons were made with measures of pupil performance. In onedistrict, no relationship was found between changes in eipendituresfor regular or CE project activities and achievern.;nt test changes.In the other district, an apparent relationship was found betweenchanges in educational services and achievement scores for a smallnumber of grades and schools.
No firm conclusions should be drawn about the relationships of CEservices and performance measures from these limited examinations.The variations of services received among grade units and of achieve-ment measures suggest that it will be necessary to seek resource-
SECTI ON 3 121
performance relationships at the grade level. Further, the complex-ity of data on CE programs and expenditures and the difficulty inobtaining the required information indicate that such analyses willrequire much effort in data gathering anu analysis.
The year-to-year variation in regular school program expendituresper pupil was found to be greater than the amount of expenditures forCE. This suggests that both types of expenditures should be con-sidered in the evaluation of the contribution of Title I funds towardsenhancement of achievement.
129 67TMP-115
SECTION 4
OBSERVATIONS, CONCLUSIONS, AND RECOMMENDATIONS
The primary objective of this study is to provide answers to the
following questions:
1. Has significant enhancement of pupil performance resultedto date from CE programs?
7 What schools, pupil and environmental characteristics areassociated with enhanced pupil performance?
3. What are the distinguishing features of successful CEprograms?
Section 2 describes the analysis performed to answer questionsi and 2. Section 3 describes the work accomplished in the firststep for developing answers to question 3.
The knowledge gained and experience encountered during thestudy have made it possible to make certain observations which arerelevant in evaluation of programs to improve education. Theseobservations are made in the first part of this section. In the secondpart, specific conclusions with regard to the basic objectives of this
study are presented. Only partial answers can be given at this timeto the first two questions, and except for analysis of level of expend-itui es, the third question has not yet been addressed. Although con-clusions from Phase I are tentative, they provide some insights intothe ultimate answers to the above questions. Furth...r, they provideguidance concerning fruitful directions that future research mighttake.
Finally, the last part of this section contains recommendations,based on the Phase I effort; which it is hoped will prove helpful toDHEW and other interested organizations.
SECTION 4 123
GENERAL OBSERVATIONS
The following observations were developed during field trips anddiscussions with officials concerned with Title I programs. Thelarge number of field .trips provided an opportunity to learn aboutthe students who are the prime target for CE, the school programs,and the views of local school officials.
Pupils Affected by Compensatory Education Programs
It was observed that although some CE programs are designedto benefit all pupils in Title I schools, many compensatory educa-tion programs are directed towards the most severely disadvant-aged pupils within each school and grade. This observation isillustrated by CE project descriptions in Section 3 and appears tobe supported by the statistical analyses of changes in achievement .
This concentration of dE service., raises a fundamental questionabout the stated objectives of Tit lc I, ESEA. Within the targetschools selected in the sample, a large percentage of pupils in eachschool are "educationally deprived." It appears that "educationallydeprived" pupils who do not happen to rank at lower achievementlevels in their respective schools are not receiving the same amountsof CE services as pupils who rankat the lower levels. This mightmean that, within the stated objectives of Title I, many programsare not yielding maximum possible benefits.
If the performance of the lowest achievers increases in propor-tion to the CE program funds provided for these students, the con-centration of funds at the lower end of the distribution might be de-fensible. If, however, diminishing returns occur (i.e. , if achieve-ment gains decline for constant increases in CE program resources),then the policy of concentrating the CE program funds on these low-est achievers may be an inefficient way to spend such funds.
It was not possible in Phase I to study the possibility of diminish-ing returns because of the very limited data on expenditures perpupil. An effort that would include obtaining regular program ex-penditures as well as expenditures for specific types of CE wouldlikely provide a range of expenditures per pupil that would be a basisfor determining the level at which diminishing returns begin.
o Wow.. wwWw. Yo..whorw owryywewhohaw.....,Aw
124
Pupil MobilityOfficials in many of the sample school districts report that a
high rate of pupil mobility exists in the inner city schools. This hasbeen verified to a large extent by the statistics accumulated to datein this study. Where this problem is found to be serious, it niay benecessary to take explicit account of it either in HEW programs orindependently by the recipient states.
Possible problems created by this high mobility rate are easy tovisualize. If a severely disadvantaged child changes schools duringthe school year, it is quite likely that the CE resources expended toupgrade him will not have been utilized in an efficient manner, underpresent administrative procedures. Schools attended during the yearmay have different types of programs, different criteria as to pupilswho are Pligible for help, and different diagnoses of particularstudent problems.
Evaluation of CE at the Local Level
All of the school districts visited have designated personnel re-sponsible for evaluating.th.e effectiveness of educational programsin their districts. Some of these positions are relatively new, per-haps the direct result of OE or State interest in evaluation, or ESEArequirements. This should be viewed as a very beneficial impactof ESEA upon future educational programs.
Schools, however, are primarily concerned with the day-to-dayeducation of their students; they are only secondailly concerned,if at all, with the conduct of rigorous educational research. Yet,if effective CE programs are to be identified and if ineffective pro-grams are to be eliminated, some research sophistication must beintroduced. Studies must be designed from the beginning, to allowconclusions to be drawn without the need for the massive datamanipulation required in this study. Care must be taken to statevery specific objectives of an educational program and, whetheror not it is a CE program, studies should be performed as experi-ments in which variables and conditions are carefully controlled.Considerable effort is required to improve data reporting and re-cording, so as to make definitive analyses possible. The typeso-1 assumptions and interpretations required to obtain the basicdata of this study are inefficient and so complex that they mayeasily yield inaccurate information.
SECTION 4
-
125
Difference among School Records
The degree to which school districts observe recommended guide-lines for records keeping was observed to vary widely. Frequ,mtdifferences of definitions were found, causing additional diffi:..titieq
in obtaining data for comparisons of pupil performance and compari-sons of expenditure levels for CE programs among the school dis-:.ricts visited. For example, some districts considered an excused
absence as "present" for attendance records, others did not. Somecomputed pupil mobility for the school year, others for the calendaryear. Records of pupil "drop-outs" employed different categories.The occurrence of ungraded schools required extra care to identifythe imputed grade level of pupils, especially for the purpose ofanalyzing achievement test scores. Differences in definitions and
forms of records require considerable extra effort in evaluation of
programs such as Title I. As in this study, data can be adjusted tobe comparable but it is probable that they adjustments reduce pre-cision of the statistical measures. Sometimes the desired data arenot available and approximations had to be accepted, e.g., the sub-
stitution of enrollment for ADM data.
Definitions of and Cost Accounting for CE Programs
Many of the school districts visited defined some of their CE
projects in terms which complicate or hamper efforts to evaluate
the true cost of the project. Some projects were defined in terms
of activities or objectives, others as resource inputs. For example
several "projects" were defined in terms of resources such as
teacher aides, community counselors, resource teachers, and in-
structional materials centers. These "project" definitions permit
simple accounting of funds but complicate identification of the full
measure of resources employed.
Financial records are seldom kept by school or project, andavailable budget and expenditure records vary in form and level of
detail. The treatment of resource inputs as separate projects tends
to hamper evaluation by fragmenting the district CE program, by
adding unnecessary complexity, and by obscuring a portion of re-
sources allocated to a project. Much effort is required to "redefine"CE projects in such a way as to clarify the appropriate measure of
performance (1. e., the objective) and to identify all of the input
resources.
126 67TMP-115
Relating Title I Expenditures per Pupil to Changes in Achievement
Although achievement test data are kept by school and grade,expenditures of Title I funds cannot be easily correlated with theachievement test results. Many school districts have test scoresfor individual pupils identified by name of pupil and the school inwhich he was enrolled but they do not have means to identify theamount of CE the individual pupils received. In some cases thetest scores are in the form of frequency functions identified only byschool in which pupils were enrolled. In this case, even if CEcould be identified for specific pupils it could not be correlatedwith achievement test scores.
Since most types of CE programs are usually oriented to only asmall percentage of the entire pupil population in Title I designatedschools the records currently being kept do not permit easy andprecise evaluation of the effectiveness of CE programs.
If schools receiving CE funds (1) would provide descriptions ofpresent and past CE projects and (2) provide a code along with eachpupil test score identifying CE programs in which the pupil hadparticipated the school district would have a wealth of information.Most schools would have less than 10 CE projects in the entireschool and a simple code could be used to identify the CE programson the test record.
Year-to-Year Variation in Test'Scores
There is a large year-to-year variation in mean test score forspecific grades within a school. The variation is much larger thanwhat would be expected from analyses of variation among pupilscores within a grade in a specific year.
In the entire sample the standard deviation of the difference inmean scores between two years was approximately twice as grea.::as would be expected bas0 on the standard deviation of test scoreswithin each year. This means that there is a large year-to-yearvariation in test ,,cores caused by such factors as quality of instruc-tion and conditions under which test is given. In essence, the dif.ference in test scores for the same grade in a schc31 in two differ.ent years appears to depend as much on differences in testing andquality of instruction as it does on differences in pupils.
SECTION 4 127
One of the causes of difference between 1965-66 and 1966-67 wasCE but this cannot, of course, account for both negative and positivechanges. The sample statistics used in this study suggest that onthe average CE accounts for only a small part of the variation be-tween the two years.
CONCLUSIONS
There are a number of conclusions that may be drawn from thestudy thus far. However, it is necessary to recognize that theseare tentative conclusions and subject to some degree of uncertaintybecause average change in achievement scores were found to besmall and sampling variation was large. Since Title I, ESEA, hadbeen in effect for a maximum of 1-1/2 years in the schools includedin the study, it was expectea that whatever benefits had accruedfrom Title I would probably be small. Moreover, the actual periodof CE exposure between 1965-66 and 1966-67 achievement tests formost students in this sample was less than one year. In addition,it was expected that sampling variation would be relatively large,due to uncontrolled factors that could cause significant differencesin test results.
Conclusions from the analyses in Section 2 are based upon resultswhichalthough statistically meaningful within the restrictive con-text necessarily createdinvolve rather small numerical values.The average change in test scores measured in grade equivalencewas less than 1 month. Hence it is possible that clanges in achieve-ment levels which seem to indicate a positive effect from Title Icould be attributable to sampling variation arising from uncontrolledenvironmental factors. On the other hand, a general downwardtrend in achievement at inner-city schools, if it exists, could under-state the amount of enhanceir At produced by CE programs.
Thus, to make an analogy with physical science the problem wasone of "detecting a weak signal in the presence of noise. " When thestatistical laws governing the noise are not completely known it isnot possible to give very precise estimates of the amount of signal.To continue with the analogy, the problem is compounded by thefact that the instruments being used to measure the signal (i.e.,standardized tests, in this case) are known to be imperfect in calibra-tion.
1
,
128 67TMP-115
The analyses described in previous Sections and the conclusionssummarized below represents the best estimate of the situation basedon the data that could be accumulated fnr thi s Q tidy. The fnl 1 (wringconclusions are useful not only as substantive findings, but as indi-cators of areas where further work would provide additional con-firmation of findings.
1. Compensatory Education programs seem to have enhanced theachievement of the students in the lowest part of their classes.*Although the performance of these students actually increasedby only a modest amount between 1965-66 and 1966-67, thereis reason to believe (see number 4 below) that they would havedone considerably worse in the absence of CE programs.(See pages 27-30.)
2. The most favorable changes between 1965-66 and 1966-67 werein those school districts which had the largest average expend-itures of Title I funds per pupil. Statistical analyses shows apositive correlation between change in test scores and amountof Title i expenditures. (See Tables 12 and 24. )
3. There appears to be a decline in mean pupil achievement levelsin the Title I schools. This conclusion was based on averagestaken over 314 school-grades in 11 different school districts.(See page 27 and Table 6.)
4. The year-to-year variation in mean achievement scores is sogreat that it was necessary to base conclusions on averagechange for several schools and several grades. This makesit difficult to interpret the importance of changes in scores.The most frequently observed change for individual grades wasapproximately two months expressed in grade equivalence.(See Appendix A and Figure 8. ) At the first decile positivechanges were more frequent than negative, but for the meanthe reverse was true. The average change at the first decilewas an increase in the neighborhood of one-fourth month ingrade equivalence .and for the mean the average change wasa decrrase in the neighborhood of one-half month in gradeequivalence. (See Tabies 6 and 7.)
*Enhancement is defined as a higher achievement level than would
have existed if there had been no CE programs. See Section 2,page 27.
SECTION 4129
5. Year-to-year changes in regular program expenditures for spe-cific grades in a school often exceeds the funds available fromTitle I. This makes it important to compare changes in achieve-ment with eXpenditures from both regular and CE programs.(See, for example, Table 29 and Figure 11.)
6. The changes in attendance between the two years were not signifi-
cant. There appears to have been several factors contributingto changes in attendance but these causes were unrelated to CEprograms. (See page 26. )
7. Changes in achievement between 1965-66 and 1966-67 were re-lated to the proportion of Negro students in a school. Thoseschools having less than 20 percent Negro student population didsignificantly better than the schools having a higher proportionof Negroes. Moreover, this relationship was similar in allschool districts. The schools in the 40-60% Negro range (1965-66)
showed the most negative change between 1965-66 and 1966-67.(See page 50 and Table 18.)Although changes in achievement were related to percent Negropupils there is no apparent relationship between changes inachievement and changes in percent Negro. (See page 54. )
8. The degree of enhancement seems to be related to the initialachievement levels of a grade school (i.e., prior to dEP); thatis, the greatest improvement came from the school grades withlowest mean achievement in 1965-66. The observation does nottake into account variations in the level of funding for CE pro-grams, thus, the . -lationship might be due to the poorer schoolshaving the highest intensity programs. (See pages 54 and 56. )
9. The eleven school districts did not change uniformly in meanachievement level between 1965-66 and 1966-67. One districtshowed some significam improvement and two districts showed
significant declines. Observed changes in each of the otherdistricts were not significantly different from zero. (See page 32
and Table 9.)
10. No strong relationships were observed between changes inachievement and poverty level, mobility, percent Spanish speak-
ing population, or size of school (ADM). However, it was diffi-cult to obtain adequate measures of poverty and mobility becauseof varying definitions employed by the several school districts.The validity of standardized achievement tests is uncertain when
the tests are given to pupils whose native language is not Englishand this abscures analysis of the relationship between percentSpanish and changes in achievement. (See page 54.)
130 67TMP-115
11. Despite the lack of significance in the mobility and attendanceregression coefficients, the consistency of their signs suggestsan area for further study. With only a few exceptions, themobility coefficients were negative (i.e., indicating a possibleinverse relationship between mobility and achievement changes)and the attendance coefficients were positive (i.e., indicatinga possible direct relationship between attendance and achieve-ment changes). (See Tables 14 and 15. )
12. Responses to CE were not uniform among grades within aschool. Most schools had some grades in which achievementincreased between 1965-66 and 1966-67 and some grades whichdeclined. This suggests that subsequent analyses should obtainand analyze data by grade level rather than using statistics onaverages for the entire school. (See pages 37-38 and 86-89.)
13. No consistent relationship was found between grade level andchanges in achievement prior to and after exposure to CE.Negative and positive changes occurred at both lower and high-er grade levels. (See Tables 16 a._ 17. ) Also, the residualsfrom the regression equations were not related to grade level.That is, the variation in achievement that cannot be explainedby the variables in the regression equations is not related tograde level.(See page 45. )
14. From examinations of CE activities at the grade level in districts10 and 13 wide variations among and within schools and gradelevels were found in expenditures per pupil (both CE programsand regular classroom instruction) and in types of CE activities.Determination, by activity, of CE services provided at the gracelevel was found to be feasible but time consuming due to the com-plexity of programs and the nature of school district records andaccounts. (See pages 120 and 121.)
15. In District 10 sample schools no relationships were discoveredbetween change-of reading test scores and type of CE activityor level of expenditure. There was little reason to expect sub-stantial change in view of the small amounts of pupil exposureto CE prior to the post test. (See Table 31 and Figure 11. ) InDistrict 13 a positive relationship was observed between changein achievement scores and educational expenditures in twogrades in two schools. (See pages 103 and 105, and 118.)
SECTI ON 4 131
The number of pupils in each of the 314 grades observed in this .
study varied from 16 to 598. Since one expects the mean score of a,large class to have less sampling variation than the mean score for asmall class two different types of averages were computed. One wasa simple average of the observed changes and the other was a weightedaverage based on the number of pupils in each grade. In general theweighted averages tend to confirm the conclusions that can be drawnfrom the simple averages. The confidence level for conclusions onthe negative change in the mean and quartiles is, however, somewhathigher in the case of weighted averages.
There appears to have been many causes of the changes in achieve-ment scores in sample schools between 1965-66 and 1966-67. Thefive state and CE program variables included in the regression equa-tions appear to explain only about 25 percent of the variation in changesin achievement. Since the only available measure of funding for CEwas very crude it is possible that better data on CE program variableswill explain considerably more of the year-to-year variation. On theother hand, there were many negative changes between 1965-66 and1966-67 that are undoubtedly caused by other than CE program vari-ables.
It is very difficult to determine the results that should be expectedat this time from the first two years of Title I. However, judgementson expected results must be made in order to make decisions on futureTitle I expenditures.
A: aough it can be costly to continue programs that yield smallreturns it can also be costly to not encourage programs that areyielding returns. It can also be costly to not experiment with pro-grams a sufficiently long time to obtain a firm evaluation.
The conclusions in this report are presented as an objective an-alysis cf available data. They do not present strong evidence ofsubstantial enhancement but they surely do not support the hypothesisof no positive effect on enhancement. The tentative conclusion fromthis study is that Title I funding has increased the amount of CE fordisadvantaged pupils and that the achievement level of pupils whohave been exposed to this increased CE has been enhanced.
RECOMMENDATI ON S
The first and second recommendations are concerned with re-search projects th,:t will give more definitive answers to the threequestions posed at the beginning of this research effort. The third
132 67TMP-115
is concerned with long-term efforts to improve evaluation of educa-tional programs. The fourth is concerned with data reporting schemesfor recipients of Title I funds.
Further Analysis of Study Objectives
It must be kept in mind that the three questions posed at the be-ginning of this study do not cover all relevant questions concerningTitle I. For instance they do not include questions on the cost-benefit ratio of CE programs, rate of increase in achievement asexpenditures are increased, or the relationship between achieve-ment in reading and the overall value of education to the individual.These are important questions, but they were beyond the scope ofthe present overall study and will not be considered in these recom-mendations.
Among the 314 observations in the sample there are several thatindicate there might have been substantial increase in achievementfrom CE. However, before selecting or rejecting observations onsome criterion of success, the following factors should be considered:first, the relation between the measure of success and the focus ofthe specific CE programs; second changes in funding of programsnot funded from Title I but which coincided with Title I; third, theportion of the pupils in each observed grade that were the primetarget of the CE program; fourth, the period of exposure to CE inrelation to the time achievement tests were given. It is iecommend-ed that follow-on research efforts, oriented towards identifying dis-tinguishing features of successful CE, be designed to collect data onall four of the above aspects.
On the question of whether significant enhanceme;:t has resultedto date from CEP there are two major aspects that require furtherstudy. Further studies should be oriented toward determining ifthere is a significant downward trend in achievement levels in schoolstypical of those now receiving Title I funding. Second, data on thesample schools must be analyzed to determine which pupils in theseschool3 actually received CE. Case studies of two school districts,presented in Section 3, reveal many instances in which CE projectswere oriented to only a very small percentage of pupils in a school.
On the question of what school, pupil and environmental factorsare associated with enhanced pupil performance, further studiesshould focus on collecting and analyzing information on the specificpupils receiving CE. The Phase I study collected considerable in-formation on schools, some information on individual grades, and.
SECTION 4 133
practically no information on specific pupils. For example, informa-tion on the mobility rate for the entire school might be a poor estimateof the mobility of the most disadvantaged pupils.
Further analysis should concentrate on two aspects of the question,namely, what characteristics are associated with local school dis-tricts' allocation of CE funds and what characteristics are associatedwith success per dollar of expenditure.
Longitudinal Studies
This study did not utilize the method of measuring the progressof individual students during exposure to specific CE programs, i. e.,the longitudinal approach. Such a procedure has great merit if timeand data are available. TEMPO recommends that DHEW initiate aproject that would apply the longitudinal approach to the problem ofevaluation of CEP. It could attempt to utilize existing data but prob-ably would need to obtain further information covering longer timeperiods. Because of the comprehensive data collection and analysistask required, this study should focus on a few CE groups ratherthan a large-scale population group.
Non-Profit Educational Research and Evaluation Organization
All of the school districts have created positions for researchpersonnel for evaluation of special programs. However, in general,it has been exceedingly difficult for these groups to attract the typeof analytical talent necessary to design the necessary informationsystems, to perform the analytical work and to recommend improvedteaching methods based on such analyses. One reason for this re-cruiting difficulty is the growing demand for such analysts through-out industry and government. It is unlikely that this demand will bemet satisfactorily for many years. A supplement (and possibly analternative) to individual evaluation groups in each of the school .
districts throughout the country is needed. No attempt will be madeto review the proposals that have been made to deal with this educ-ational evaluation problem. Instead, this report will simply addone more proposal to the growing list of alternatives.
It is the view of this contractor that a non-profit EducationalResearch and Evaluation Organization is needed and would be at-tractive to the kinds of analysts required. Because of the desire on
---.................ele,"8118,0"..........
134 67TMP-115
the part of the school districts to preserve their independence, di-rect control and total financial support by the U.S. Department ofHealth, Education and Welfare might be 1.-..ndesirable. Although DHEW,should take the lead in helping to e 't ablish such an institution, it aughtbe best for the states and school districts to support it financially.General research into innovative teaching meth.ods and evaluationcould be made available to all member jurisdictions and states ordistricts could request help in testing new techniques or in eval-uating various local programs,. This institution could also help mem-ber jurisdictions in the design of information fyysterns and thereby en-courage the development of some degree of standardization among thejurisdictions without involving DHEW. The Board of Directors shouldconsist of representatives from States, Local School Districts, Un-iversities, Industry and DHEW.
Other arrangements are feasible and might be at least as pro-ductive as the one suggested here. For example, an existing organ-izaton (such as the Educational Testing Service) might be encour-aged to assume the role of such an organization. Another alternativemight be to expand the program being developed by Ralph Tyler(the Program for the National Assessment of Educational Progress Y:to include these evaluation functions, but this might not be feasiblepolitically. Still another possibility might be to encourage the Re-search and Development Centers to take on the job.
The Research and Development Center at the University of Calif-ornia at Los Angeles already has focused upon the study of the theory,methods, and operation of activities to evaluate instructional pro-grams.
All these approaches would try to centralize the scarce talent ina single location. If some decentralization is deemed desirable, itmight be well to consider the Regional E..,ucational Laboratories.The 20 RELs already have established relations with, and are fam-iliar with, problems of the schools in their regions, and the sort ofevaluation program needed might be easier to establish through theRELs than through another agency. It is not sufficient to establishthe evaluation function; it is also necessary that appropriate data beaccumulated on a recurring basis and that evaluation results, stem-ing from these data, be incorporated into the processes of planningand managing programs, schools, and school districts. One of the
*Reference 10, pp 378-:380
. IUM MUM
SECTION 4 135
important tasks that the research organization could perform is tospecify data requirements for educational research and to developa comprehensive data bank with contributions from all participatingschool organizations. DHEW is urged to take the lead in encoragingthe States and Distr:.cts to emphasize evaluation of their programsand to establish some centralized mechanism to get evaluations done.Evaluations a r e essential if the dollars allocated to education are tobe employed effectively.
Data Collection Formats for Title I Recipients
The most severe problem encountered in this study was the lackof accurate and relevant data on CE programs and pupil-school-environmental characteristics. These data are not consistentlyand systematically recorded by the school districts. Such was thecase despite the several guidelines already provided by OE. Unlesssomething is done to remedy fhis situation, timely, accurate andmeaningful evaluation will remain difficult.
It is important to make proper arrangements so that data routinelycollected by local school districts can be utilized, thereby circum-venting the need for designing experimental progra. ';o evaluateFederally funded projects. While it is recognized that requests byFederal Government agencies, for uniform procedures raise pol-itical and practical difficulties the requirement for consistent anduniform records is considered worthy of further attention.
136
APPENDIX A
PRINTOUT OF THE DATA USED IN THE SURVEY
67TMP-115
This appendix contains a prin:out of the data used in this surveyfor the regression, co-variance, and stratification analyses. Onlygrades which had achievement test results for both 1965-66 and1966-67 were used as observations. To meet this criterion, it wasnecessary to discard many observations when a post-pre year matchcould not be made.
A total of 314 observations are shown covering 132 schools in 11district. Eleven of the 12 grades are covered there were no ob-servations for the 9th grade in any district. Grades 10 and 12 con-tain only one observation each.
An example with a brief explanation of each datum or yariable isshown on page 137.
The data are listed in ascending district, school and grade orderstarting on page 138.
....=0.4111111111111111111134ylingin"
04 ni
AA
A0)22 1)on 1023
Thi
s
14 42 3
7.4 10 n2 1)1 04 024 02b
A 44 4
A465 003 002 03? .019 1
OSO 079 029 025 09 25 105 mob
AA
AA
J
A
I IBM
car
d re
visi
on n
umbe
rfor
GE
use
onl
yS
tand
ard
Dev
. of t
est s
core
s: 1
965-
6 =
8.0
; 196
6-7
= 7
.9
+ No.
of s
tude
nts
test
ed 1
965-
6 an
d 19
66-7
Atte
n&ric
c +
2.5%
196
5-6
to 1
966-
7A
ttent
ir..n
ce 1
965-
6 w
as 8
9%
iV T
este
d O
ct. 6
5 a
Oct
. 66
At 7
5%ile
, pos
t - p
re y
ear
test
diff
eren
ce w
as -
1.9Ski.
T u
nits
At 2
5%ilt
, pos
t - p
re y
ear
test
diff
eren
ce w
as +
3.2
Std
. T u
nits
At 1
0%ile
, pos
t - p
re y
ear
test
diff
eren
ce w
as +
0.2
Std
. T u
nits
At t
he m
ean,
pos
t - p
re y
ear
test
diff
eren
ce w
as -
0.3
Sta
ndar
d T
uni
tsM
ean
achi
evem
ent t
est s
core
for
the
1965
-6 c
lass
was
46.
5 S
tand
ard
T u
nits
IlF
or th
is g
rade
, the
sch
eol r
epor
ted
an A
DM
of 2
3 st
uden
ts in
196
5-6
and
25 s
tude
nts
in 1
966-
7T
his
is g
rade
4 o
f sch
ool 3
, in
city
4In
196
5-6,
the
scho
ol r
epor
ted
that
2%
wer
e S
pani
sh s
peak
ing
and
1% in
196
6-7
'In 1
965-
6, th
e sc
hool
rep
orte
d th
at 2
4% w
ere
Neg
ro a
nd 3
8% in
196
6-7
A p
over
ty in
dex
(3)
indi
cate
s se
vere
impo
veris
hmen
t rel
ativ
e to
oth
er T
itle
I sch
ools
in th
is c
ity (
1 =
low
, 2 =
mod
erat
e)A
bout
42%
of t
he s
tude
nt b
ody
in th
is s
choo
l ent
ered
or
left
durin
g th
e sc
hool
yea
r 19
65-6
6T
he s
choo
l rep
orte
d an
Ave
rage
Dai
ly M
embe
rshi
p (A
DM
) of
120
0 st
uden
ts in
196
5-66
The
city
rep
orte
d th
at a
ppro
i. $1
2 m
illio
n in
Titl
e 1
fund
s w
as s
pent
in 1
965-
6 an
d $1
0.2
mill
ion
was
spe
nt in
196
6-7
The
city
rep
orte
d th
at th
e A
DM
of a
ll sc
hool
s ai
ded
by T
itle
I fun
ds w
as 2
22 th
ousa
nd s
tude
nts
Thi
s id
entif
ies
scho
ol 3
, of c
ity 4
iden
tifie
s ci
ty o
r sc
hool
dis
tric
t 4 fo
r th
e re
st o
f the
dat
a on
this
line
(or
IBM
car
d) -
AN
EX
AM
PLE
FO
R T
HE
INT
ER
PR
ET
AT
ION
OF
TH
E D
AT
A
%
01
01
0060
0256
0146
09
33
390
92
04
143
119
36702400b
*031029
A080
081
132
118 v3
.03
0b6
057
01
01
0060
0256
0146
09
33
390
92
05
139
130
384
.003
-004
.009
.015
AOdl
081
125
126
Va10
056
057
01
02
0060
0256
0146
Ov de4
399
49
05
115
139
404016
-004
-031
-016
A077WM
111
139
96
-07
056
037
01
04
0060
02'36
0146
08
bp p
51
53
05
07809
369008
-004
-004
.015
A072
089
073
088
'01-22
056
037
01
05
006U
0256
0146
10
72
3%S
41
04
126
112
337
*015
*005
*00503d
A067
079
113
099
vo
-03
056
057
01
n5
ow:1i,
02,260146
10
72
355
hi
05
124
112
370
-018
-004
-026003
A076
081
111
106
vo06
056
037
01
06
0060
0266
0146
04
rs 3
60
63
0038
046
344
.023
-004
-034
.047
A04207
035
046
Ve
-04
066
037
01
07
0060
0266
0146
13
41
299
99
Oh
155
379
.002
-004
-037
.04d
A076
090
150
104
va
06
037
01
Od
0060
0256
0146
14
41
390
41
OS
164
191
378023004021032
A076
067
158
171
Is4-05
056
037
01
09
0060
0256
0146
Ov
23
399
99
04
132
115
39402202302900d
A072
073
128
109
v4
m09
056
057
01
09
0060
0266
0146
09
23
399
99
05
127
114
389
*006
-012
-0140e3
A07h
065
122
116
94
429
Obb
037
01
10
0060
0266
0146
or
r2
310
19
04
090
076
362028
*019
.031
.o18
A072
084
084
074
va
-25
056
057
01
10
6060
0256
0146
07
72
310
19
05
081
078
361
.014
-004
.013
.022
A079
064
067
080
91
-42
056
037
01
11
0000
0266
0146
05
92
331
34
05
061
054
339040
-004
.041
.067
A074
086
090
053
88
.2206
037
01
le
0060
0266
0146
lid
57
369
74
04
066
056
401
-031005
*00 el.084
A096
073
06!
056
Y6
-10
056
057
01
12
0060
0256
0146
08
57
369
74
05
068
odd
365
.021
-004
.341
-004
A105
095
084
080
93
-12
056
031
01
12
0060
0266
0146
08
57
369
74
06
078
081
369
.022
.036035
.008
A094
083
077
079
v2
.03
026
017
01
14
00b0
0256
0146
08
45
292
94
04
091
094
333
.045
.005
.042
.068
A067
091
085
069
94
.09
056
057
01
14
0060
0256
0146
od
45
p92
94
05
099
086
370
.005002
-010
.o16
A095
00004
01 9k -09
056
037
TO
TA
L D
AT
A B
Y C
ITY
, SC
HO
OL,
AN
D G
RA
DE
01 14 0060 0256
0146 0 45 p 92 94
01 lb 0060 0256 0146
od 43 2 99 99
01 lb 0060 0256 0146
Od 43 2 99 99
01 16 0000 0256 0140
09 39 1
00 o0
01 16 0000 0256 0140
09 39 1 00 00
01 16 0000 0254 0146
09 39 1 00 00
11/17 000 0256 0146
09 43 I
13 16
01 10 000 0256
0146 01 6' 2 44
49
01 10 0060 0256 0140
or 66 ? 44 49
01 14 000 0256 0146
09 51
1 26 P1
01 19 0060 0256 0146
09 51
1?6 /1
01 19 000 0256 0140
01 51
I 26 PI
01 20 0060 0256
0146 09 36 3 87 46
01 20 0060 0254 0146
09 36 3 87 46
01 21 0060 0256 0146
14 62 3 64 70
01 21 0060 0256 0146
14 52 3 64 70
01 22 0060 0256 0140
10 30 2 65
69
01 22 0060 0256
0146 10 30 2 65 69
01 23 0060 0256
0146 06 37 2 79 80
06 080 090 385 010
036 .003 010 A 082
077 078 090 V4 .06 026 017
04 094 100 371) .000
.005 .00S '016
A 078 077 091 096 VS 0.4 06
057
OS 100 089 309 -005 *024
-014 -008 A 084 071
068 085 *3 .36 056 07
04 III 124 402 *030
*063 *033 011 A 099
092 108 121 v4 02 056
057
OS 104 Iii 409 4004 .024
-024 -002 A 100 111
105 117 12 .00 056 037
06 100 103 422 4014
4023 *001 *017 A 085 01
098 101 91 .03 0i4.6 017
05 161 191 423 -006 -021
004 -009 A 090 099
158 184 *4 0t, 050 037
04 097 106 359 .011
*005 *022 -002
A 074 067 097 109 91 -01 056 057
05 07 008 400 -048 -034
-069 -035 A 078 Odd
070 087 42 -50 056 037
04 156 125 377 .011
.005 .035 .018 A 087
077 155 121 92 ..36 056 057
OS 132 138 399 -012 =011
-034 .003 A 084 086
125 134 91 *03 056 03?
06 141 122 410 =010
=009 .008 .4121)
A 093 077 138 121 12 -20
026 017
07 351 335 375 *010 4002
.022 .012 A 061
067 290 268 66 =26
056 057
08 289 299 385 -011
-036 -018 -008 A 067
065 253 tab
65 -
15 0
6 12
6
07 492 519 383 -015
=026 =011 =006 A 074067 408 456 01 13 056
057
008 A 073 081
365 397 04 01 06 h26
Ob 470 436 378 007
6'025 -008
07 340 365 411 .000
.019 .000 007
A 082 076 312 327
Oa 333
306394 -002 -019 -016 .007
A 076 089 305 296
07 201 213 388 -011
4002 =025 =026 A 069
ot2 166 195
TO
TA
L D
AT
A B
Y C
ITY
,S
CH
OO
L, A
ND
GR
AD
E
YO 004
056
057
09
.14
056
126
06
-02
056
057
ar.0
Ca
.43
01 23
0060
0256
0146 06 37
2 79
HO
08
204 109 394
-006
.4008
-01A
4010
A076 075 165 172 c112 .01
056 126
02 01020
035.3
0328 05 88
3 69
67
07
le 02
07b Odb 438
'045
.018won
"067
103 056 067 07e 06 04
10S 106
oe ol
0204
0353
03e6 ob 88
3 69
57
07
le 03
066 07b 4ob013
.028065
-024
098 094 077 070 VO 02
105 106
oe 01
020v
0353
0328 05 88
3 69
67
07
ld 04
079 073 421
-018
.000
.029
4.038
094 0v1 072 071 v1 .00
105 11)6
oe ol
0209033
03ed ob 88
3 69
57
07
14 0S
063 082 424
-043
-040
-077
-053
076 077 0*6 070 0* 4404
ab 106
oe ol
0209
0353
0326 ob 88
1 59
57
07
12 Ob
074 052 358
.02?028
030
.033
086 068 061 oa Vu
Vg;
oe 02020
0353
03ed lo 37
3 99
49
00
00 02
180 179 450007068024
.040
127 106 175 15b
4 01
10S lob
oe 02
020v
0353
03d8 10 37
3 99
99
00
00 03
189 193 434004
058
,4012
-020
069 095 157 148 vj 0110 10
02 02
0209
0353
0326 10 37
3 99
49
00
00 04
179 184 41i
.007
4045
4010
.003
10' 094 157 157 v3 4)1
106 106
ne 02
0209
0353
0328 10 37
3 99
99
00
00 05
148 159 396
.008
.000
.000
.011
odd odo 130 148 93 .00
lob 106
02 03
02u9
0353
0328 07 44
2 50
58
28
23 02
088 135 466
m039
-040
-037
-063
112 095 082 125 v5 .00
105 106
oe 03
0209
0353
03e8 07 44
p So
58
28
23 03
097 114 474
-011025
4.004
-021
102 098 096 105 44 .02
10S 106
oe 03
0209
0353
0326 07 44
2 So
58
28
23 04
095 104 441028
4068016
.026
099 067 061 097 v5 00
105 106
oe 03
0204
0353
0328 07 44
2 So
58
28
23 05
100 104? 438
.4025
.000
-027
-oeb
092 086 098 105 96 -01
105 106
oe 03020
0353
0328 07 44
2 So
58
26
23 06
095 101 464
.046
.000
-032
-050
104 087 076 099 vb .00
105 106
02 04
02ov
0353
0326 01 72
2 35
37
2928 02
099 125 473
.-040
-077
.032
.063
Old 00 095 121 44 .00
105 100
02 04
0209
0353
o3ed 07 72
? 35
37
29
28 04
108 112 445
.015
4030
-017
.047
092 103 106 108 v3 .00
105 lob
ne 04
0209
035i
03e6 07 72
2 35
37
29
28 05
088 116 480
-060-05
44042
-078
099 090 080 098 v3 00
105 106
TOTAL DATA BY CITY, SCHOOL, AND GRADE
02
04
0209
0353
0328
01
72
235
37
29
28
06
098
108
462
=024
=040000
-029
103
094
093
089
v400
105
106
02
04
0209
0353
0328
01
72
235
37
29
28 o3 097
123483
-011
-031
-02000f
088
098
085
lob 94
.00
Io5 lob
02
05
020v
0353
0326
li
99
199 Q9
00
00
03
323
322 43h024
025
.012
.012
062
089
316
359
94
.01
105
106
02
05
0209
0353
0348
17
99
199
99
00
00
04
276
364
416004
.017
-015
.000
087
obS 27940
94
.01
105
106
04
05
020v
0353
0328
If 9
i99 99
00
00
05
261
306
424
-039
-040
-037
-060
061
065
256
342
v4
.01
105
lob
oe os 020 0353
0326
11
99
i99
99
00
00
06
251
439
-047
-040
-047
-0362
064
066
244
4.01
1us
Lob
02 ob
0209
0353
0328
08
12
399
99
oo
oo
03
116
091
443021
.026024
.019
084
087
114
089
v3
.01
106
306
02
06
0244
0353
0328
08
12
399
99
00
00
04
112
128
416
"018
"000
"035
-032
079
088
101
121
4300
105 ao
02
06
0204
0353
0326
08
12
399
49
00
00
Ob
100
106
408
"013000
"015
-020
u79 ob4
o91
102
43
fl:10105
106
02
06020
0363
0328
0d
12
399
99
00
00
06
100
209
419
031
.000
-010
-053
093
o7u
093
204
v4
.00
105
106
02
o7
0209
0353
0328
09
86
201
42
b3
64
02
149
176
441
.020004
.057
.013
087
080
145
146
v3
-01
105
106
02
07
0209
0353
03e8
09
$8
201
02
53
64
03
138
159
486
-016
=012
=010
-024
092
086
136
!CI v4
-01
105
106
Od
07
0209
0353
0328
09
86
201
02
53
64
04
134
161
458
*021010036
oel
107
096
132
126
'C:
105
lob
02
07020
0353
0326
09
lib
201
o2 53
54
05
109
148
448
"012017
020
-015
103 o79
106
107
45
-02
105
106
02
07
0204
0353
0328
08
86
201
0?
53
84
06
127
131
456019049
.051
=010
108
087
128
098
v400
105
lob
02
08
0249
0353
0346
04
67
P 50
50
OS
02
02
066
057
446
.024
.040
.024
*026
108
118
052
063
v4
.00
lob
106
02
od020
0353
0328
04
67
250
50.08
02
03
061
06p 501
6.041
-033
..044
-.402
096
115
057
061
45
.00
105
106
02
OH
0209
0353
0328
04
67
250
SO 08 oe 04
063
050
489
-038
-023
-065
-014
111
108
060
049
v5
- 03
105
106
02
OH020
0363
0348
04
67
2 So su ua ue os 065
058
45?
-032
-048
-051
-035
09813
065 obo
44
.00
105
106
TOTAL DATA BY CITY, SCHOOL, AND GRADE
4tu
Od 08 0209 0353 0326 04 67 2 SO so 00 02
Od 09 0209 0353 0328 06 20 3 99 99 00 00
02 o9 0209 035.; 0328 06 20 3 99 99 00 00
Od 09 0209 0353 032o 06 20 3 99 99 00 00
oe o9 0209 0353 0328 06 20 3 99 99 00 00
Od 09 0209 0353 0328 06 20 3 99 99 00 00
Od 10 0209 0353 0328 07 15 3 99 99 00 00
02 10 0209 0353 0328 0/ 15 3 99 99 00 00
Oe 10 0209 0353 0328 07 15 3 99 49 00 00
Od 10 0209 0353 0328 0 15 3 99 99 00 00
oe 10 0209 0353 o3ed of lb 3 99 99 00 00
02 11 0209 0353 0328 06 49 2 00 40 ob 08
02 11 0209 0353 0328 06 49 2 00 00 06 08
Od 11 0209 0353 0328 06 49 2 00 00 06 00
02 11 0209 0353 0328 06 40 2 00 00 06 00
oe 11 0209 0353 0328 06 49.p 00 00 06, 08
02 12 0209 0353 0328 07 17 3 99 99 00 00
Od 12 0209 0353 0328 07 17 3 99 99 00 00
Od 12 02u9 0353 0328 07 17 3 99 99 00 00
06 049 082 476 -074
-06g."-058
-071
102 096 050 10t "15 *00 10b 106
02 108 113 512 -101
4061
4.075
-120
003 067 107 104 V4 ...02 105 196
03 115 103 431 .027
.025.032
.045
003 083 100 106 96 ,01 105 106
04 108 101 406 .4009
4000
-010
-020
089 086 105 103 94 400 105 106
05 098 106 384 004000
017
-001
on 077 078 loo v3 .01 105 106
06 106 Oeb 40b -029
4028017
-063
o97 oTi o99 OS0 93 Al IOS 106
02 127 128 471 4001
4011
4013
-025
099 063 128 125 lob -01 105 lob
03 125 123 457 007
000
-004
.013
078 095 126 123 v4 .01 105 lob
04 117 129 467 .031
-060042
.010
08 odd 112 119 94 401 105 106
05 128.121 434 44010
4018
-021
-020
083 070 102 115 y!..t :co i05 106
06 108 141 439 03500d
-042
-046
07 072 101 119 v6 00 405 106
02 o95 120 442 005040024
-01J
095 100 094 114 v4 00 105 106
03 09 098 477 -001025
-020
-003
088 097 077 090 94 00 105 106
04 097 108 458 055
4061066
40441
098 092 101 093 94 .00 105 106
05 073 094 459 .037
-002
-013
-050
101 070 Os 063 94 .e0 los 106
06 064 077 455 -002
-005
.004
.003
091 085 060 074 v5 .02 105 106
02 166 127 443 -007
-040
.000
.000
078 088 161 118 93 .01 205 106
03 132 139 432 -012
.040
.012
.040
087 065 122 135 93 .00 105 196
04 115-122 422 -018
.000
-023012
119 093 107 119 V3 402 105 106
-10
TOTAL DATA BY CITY, SCHOOL, AND GRADE
S.
o2
12
0709
0353
0328 0?
17
399
99
00
00
06
13* 099
411
m020000
-002
-030
0.51
073
134
102 V4 .00 105
106
03
01
0016
0074
0056 04
350
48
00
04
045
36
453
-025
.010
-028
-021
1075
070
046
030 v3 .02 016
017
03
01
0016
0074
0056 04
350
48
00
06
037 039
419
.041
.019
.042
.065
1OM
080
037
038 v2 .07 016
017
03
02
0016
0074
0056 04
p32
oo
04
05s 058
470
+003
-034013
.014
1074
063
054
057 v4 -01 016
037
03
02
0016
0074
0056 44
232
ou
06
067 064
496
-036
-032
-028
-043
1089
079
066
062 94 -06 016
017
o3
03
0016
0074
0066 03
317
19
04
024 02/
402
4004
-045
4061
'002
1050
077
023
026 v3 '12 016
017
03
03
0016
0074
0056 03
317
19
06
024 029
437
o°002
4055009
-020
1050
055
02b
025 v4 ,010 016
017
03
04
0016
0014000 02
262
00
04
027 020
439
.045
.045
.015
.053
2067
074
026
022 v4 .07 018
017
03
04
0016
0074
0056 02
262
00
06023 022
456008020
.023
.026
2069
089
021
022 93 .15 016
017
03
05
0016
0074
0056 03
292
00
04
034 035
395
+002
-045
-042025
1061
083
033
035 94 .12 016
017
03
05
U016
0074006 03
292
00
06
036 031
392004
'000
'009
'032
1080
102
029
029 93 .02 016
017
03
06
0016
0074
0056 02
2a9
00
04
023 034
462
0034025
*049054
108400
023
031 96 -17 016
017
03
ob
0016
0074
0056 02
249
00
06026 026
459
-006
.016
'000
.002
1080
070
026026 W6 09 016
017
03
07
00)6
0v *=
00b6 06
228
02
04
073 057
454
.016
.046
.021
.014
2064
069
069
;:50 v5 .0b 016
017
03
07
0016
0074
o -b 05
218
02
06
073 053
455
+001
-007
-043
.034
1098
114
072
084 95 .07 016
017
03
Os
0016
0074
0056 06
100
02
04
085 083
493
0020
.014000
#017
1089
086
079
082 95 .12 016
017
03
08
0016
0074
0056 06
140
02
06
066 076
480029
-007
-028
-031
1090
075
05906 94 .06 016
017
03
09
0016
0074
0056 03
101
00
04
046 046
461
'002048
'4001
'000
1096
060
045y44 Vb 05 016
017
TO
TA
L D
AT
A B
Y C
ITY
, SC
HO
OL,
AN
D G
RA
DE
os n
90016 0074 oob6 oi
oi Io 0016 0074 0066 06
0016 0074 00b6 06
0010 0074 00b6 06
0010 0074 00b6 06
OS le 0010 0074 C0T.1 04
OS le
OS 13
OS 13
OS 14
OJ 14
OS 15
03 lb
OS 16
OS 17
03 18
OS 19
OS 20
OS el
OS
10
04
11
0.1
11
0016
0074
00b6
04
0016
0074
00b0 Oi
0016
0014
0066
02
001J
0074
00b6 Ob
0010
0014
00b0
05
0014,
0074
00w6
Od
0016
0074
0056
08
0016
0076
00b6
06
0016
0074
00b6 od
001*
0014 web(' 06
0016
0074
00b6
01
0010
0074
00b6
Od
0010
0014
0066
1/
r01 3
01
60
00 06 039 033
02 04 063 064
426
439
*008
.022
024 018 .011 1 066 069 037 034 43 .17 016 01?
.013 -024 .on a 087 or 041 049 V4 .0 016 017
360
02 06 obi 06b
416
-023
-008
-030
.025
1Odb 088 057 Qbb 41 16 016 01T
376
00 04 080 083
417
.008
-017
-042017
1074 086 075 079 14 .07 016 017
376
00 06 067 060
419
-020000
-012
-005
1089 074 064 057 14 +II 016 017
214
00 04 0b7 051
452015048008
1)17
1091 086 Obl 050 9.1 09 016 017
214
00 06 Ob4 o49
463
-021
-008
-037
-013
1080 079 048 048 42 423 016 ot7
3e5
or 04 014 020
673
.055
.001
-019
-099
1086 059 016 017 vi 907 016 011
325
02 06 021 019
447
-070
-125
-110
-0b1
10+9 052 022 old .12 .620 016 017
1e2
03 04 Ode 057
401
.026
.061065
.000
1116 082 072 041 v6 -10 016 017
122
03 06 068 064
490
-031
-059
-062
.046
107e 09e 067 059 vb .05 016 017
203
01 04 098 102
474
-016
.013
.000
-024
1100 09b 087 099 14 .11 016 017
203
01 06 109 114
446
012
030
-020
-002
1085 085 097 tO6 44 .00 016 011
2Ph
00 08 OW !b9
493
-002018
024
-018
1102 088 193 152 43 *09 016 017
328
03 08 198 212
447
.025025009
*025
1088 092 191 205 41 11 016 oir
347
01 08 181 198
464009
*4).11
.000
*005
2083 odb 184 174 le -11 016 olr
100
02 od 193 255
465
-001
4022
-011
.00s
1o86 odo 192 251 ve
.13
0:01
1 W
IT
104
01 08 284 304
473
-005
-006000
-010
2081 079 280 498 ve .17 016 017
06
01 10 630 609
488005
4000
.000000
1090 091 b98 m8b v2 .03 095 096
51.
-e7
TO
TA
L D
AT
A B
Y C
ITY
, SC
HO
OL,
AN
D G
RA
DE
1,.,
.....r
oorm
rous
en,
i
03
21
0016
0074
0056
11
106
01
11
535
586
506
-019017033O16
1090
093
534
579
93
+15
005
095
0.121
0016
0074006
11
106
01
12
526
492
512
9009013000013
1090
090
482
501
v3 -et)
095
096
04
01
0222
1200100
U 46 3
96
95
00
00
04
085
062
403
-003
-003
.015
-020
1079
065
083
056
91
-0*
105
106
04
01
022C
1200
1020
12
46
395
95
00
00
06
074
081
402003
058
006
+014
1084
064
068
077
V1
.04
105
106
04 oe 0222
1200
1020
06
99
317
15
04
04
04
044
062
435000
*015
*009
-022
1082
063
041
045
0b08
145
104
04
02
ORdd
1200
1020
06
99-i
17
15
04
04
06
047
039
412
.010
.017
.009010
1079
09e=
0.2
::4t.
1=1-08
105
106
04
03
0222
1200
1020
12
42
224
38
02
01
04
023
026
465003
.00203'
s119
i000
079
024
025
69
4,25
105
106
o*
03
0222
1200
1020
12
42
224
38
02
01
06
029
022
456031
031
-027
.069
1072
085
026
021
OV +25
105
106
04
04
022d
1200
1020
10
40
299
99
00
00
0*
08*
073
429
.009
.058
-020
.002
1060odb 03
064
69
.36
105
106
04
(140222
1200
1o20
10
40
799
99
00
00
06
078
081
385005
.000
.009
.001
1081
oue 076
085
09
.30
105
106
04
OS
0222
1200
lodo
04
99
329
24
02
04
04
026
032
426
-016
-056
.000
-015
1076
077
024
030
67
-0a
105
106
04
o5
0222
1200
1020
OJ
99
29
24
02
04
06
020
021
420
-015
-009
-054
-011
105b
069
017
019
67
-00
105
106
04
06one 1200
1040
16
40
29A
98
00
00
04
141
172
431011009009
+017
1066
061
130
165
VI
.11
105
106
04
06
0222
1200
1020
16
40.2
9A
98
00
00
06
130
149
390
.003
.000
-007
-004
1069
073
124
140 vl
.11
loS
106
04
07
0222
1200
1020
13
31
197
97
00
00
04
079
096
427
-016
-014
.000
028
1083
076
069
091
92
.09
105
105
0 *
09
0222
1200
lodo
04
45
345
38
27
31
06
020
027
393001
,0*9
4,003
.015
1Old
072
020
026
90
.03
105
100
04
10
OP22
1200
1020
0/
32
194
94
01
01
04
ob4
046
422
*012
*040
*000025
1071
070
048
046
10125
105
106
04
10
0222
1200
1040
07
32
194
94
01
01
06
050
030
432007050041
036
I070
086
040
030
V125
105
106
TO
TA
L D
AT
A B
Y C
ITY
, SC
HO
OL,
AN
D G
RA
DE
, 42.
tri
n*
it
0722
1200
1040
09
49
P.21
19
36
34
o4
081
obb
444
-021
.049000
-040
1 086
068
078
063
v0
-04
105
106
04
11
0222
1200
1020
09
49
221
19
36
3*
06
068
008
430
-030
-011
-028
-019
1 079
069
051
068
*0
-04
105
106
o4
le
0222
1200
1020
16
29
153
57
oo
ol
08
426
-024
4025
-032
-024
1 081
075
325
344
91
-14
105
106
0*
13one
1200
1020
10
37
114
12
12
17
08
455
.0011
-006
.014
-017
1084
073
247
230
08
-20
105
106
04
1*one
1200
10,10
13
64
368
76
01
01
08
409
-010
.000
.015
.025
1 084
071
203
200
64
-11
105
106
04
15
0224
1200
102o
12
31
398
91
00
00
08
405
.003
.000
-007
0001
1071
088
224
221
88
-01
105
106
04
16
0222
1200
1020
17
29
190
t,'
00
00
OH
416001
.030
-006
.010
1 082
073
348
340
V2
-09
105
106
0*
n7
0222
1200
1020
13
31
197
v7
00
00
06
066
Odo
436
-042
-071
-080
-013
1 067
066
051
4.179
v2
409
105
106
04
od
0122
1200
1020
01
50
104
04
02
01
04
obl
061
452
-015022
000
-013
1o79
072
047
057
92
14 los lot'
04 oe one
140
1020
0/
b0
104
04
12
01ob 04b 00
475
-032-00
-020
-042
1 083
076
042
051
9214
lob
106
04
09
0222
1200
1020
04
45
345
38
27
31
04
036
032
438
.020
-082
-048
.021
1060
083
028
031
90
-03
105
106
004
0050
0184
0157
00
107
10
00
00
04
463
-009000
-020
-041
0 074
085
052
053
036
037
Ob
02000
0188
01*/
06
107
10
00
00
06
395001
-054
4011
4067
0 078
081
046
046
036
04?
004
005o
0189
0157
08
39
399
99
00
00
04
084
070
419
w0013000
037
-.013
0 083
070
083
064
036
03?
004
0050
018H
016r
06
39399
99
00
00
06
098
082
392
-014028
.010
-016
0 094
067
068
090
036
047
05
08
0050
0188
015/
03
305
OS
00
00
04
406
-008020
000
-07A
0 095
074
029
034
036
037
0.
05
08000
0188OW
03
305
GS
00
00
06
398
.007030
-010013
0 019
081
038
037
036
047
.1-4
Ob
10
00bo
0148OW
07
363
73
00
00
04
417
-007
-040
.000
-015
o 062
071
063
090
036
037
N.
-tt
I I ....
TO
TA
L D
AT
A B
Y C
iTY
, SC
HO
OL,
AN
D G
RA
DE
(A
0 10 000 0188 01b/ 01
3 63 73 00 00 06
410
-011 -003 -015
-006 0 075 068 065
047
036
047
Ob 14 000 0188 01b/ 01
2 03 02 00 00 04
448-014 000 000
-020 0 084 080 105
083
036
031
Oh 14 000 0188 olb/ 0/
2 03 02 00 00 06
430
.005 .olb .005
.021 0 072 067 072
078
036
047
oh 16 000 0188 01b1 03
2 14 13 00 00 04
425
.016 .000 4000
.031 0 064 089 029
023
036
037
Ob 16 0050 0188 0157 03
2 14 13 00 00 06
379
032 .036 .048
.036 0 064 065 02b
022
036
047
05 26 0050 0188 001 14
3 99 99 00 00 04
415
.007 000 .037
02o o 079 o79 145
123
036
037
0 26 0050 0188 015/ 14
3 99 99 00 00 06
304
*000 *000 *000
-006 o 078 081 122
085
036
041
05 32 005J 0188 0151 06
? 89 90 00 00 04
426
-019 -040 -018
.023 0 080 088 061
066
'
036
037
05 32 0050 0188 olbr ob
2 89 90 00 00 06
398019 .000 000
011 0 061 071 052
052
036
037
ob 38 0050 0188 0157 08
I75 79 00 00 04
436.001 .000 000
013 0 076 082 097
106
036
037
ob 38 000 0188 0157 09
175 79 00 00 06
414
-014 -033 -006
-012 0 070 070 066
089
036
047
0 42 000 01804 015/ 11
199 99 00 00 04
k.59-001 000 00J
-025 o 099 09 137
135
036
037
05 42 0050 0188 01b7 11
1 99 99 00 00 06
465
4,007 034 -026
000 0 093 110 098
106
036
047
05 SI 0050 0188 0151 10
P 38 45 00 00 04
434.016 000 024
.026 0 088 081 109
107
036
037
0b 51 0050 0188 0151 16
2 38 45 00 00 06
418
4,015 4,028 -015
-021 0 077 091 085
090
036
047
0 Se oobo oios olb7 07.
3 So 55 oo 00 04
435027 -032 -037
-045 0 085 on 065
0111
036
037
02 52 0050 0188 015? Or
3 50 55 00 00 Os
400
.013 -058 -020
-020 0 010 076 072
066
036
047
05 56 0050 0188 0157 09
199 89 00 00 04
456
.4001 4037 000
-02H 0 085 074 095
074
036
037
Ob 56 0050 0188 0151.09
1 99 89 00 00 06
414
004 4005 .016
.006 0 018 085 078
073
036
047
TOTAL DATA BY CITY, SCHOOL, AND GRADE
111M
1111
0.0N
ut 63
0050
0188 0157 Ob
2 99 99 00 00 04
403
.030000
037
.0b3
0075
079
059 064
038
03?
05 63
0050
0188 0151 05
2 99 99 00 00 06
429
-015
-04b
-039
*003
0096
102
040 045
036
047
08 01
ooIs
0115 0122 04 13
3 63 60 oo oo 03 075
105
419022
4084
4020
4000
1088
076
072 08e 04u0
056
037
00 01
0015
0115 0122 04 13
3 63 60 00 00 06 070
071
426029
.038
.033018
1091
058
045 034 04
.35
056
037
0 02
001*
0115 0122 03 29
2 25 25 00 00 06 045
034
446
.015
.050033
012
1091
058045 04
036
037
ob 02
0015
0115 0122 03 29
2 25 25 00 00 03 031
032
435
-015
-004
-015
.004
1081
072
031 032
036
031
Od 01
0043
0122 0121 08 89
3 32 21 00 00 02 087
090
446
4008
'002
-020038
064
087
081 U91 /3
046
047
Od 01
004i
0122 old 1 0 69
3 32 21 00 00 03 081
080
450
.002
.000
-020
.013
060
061
078 080 94
046
047
Od 01
0043
0122 0121 0 b 69
3 32 21 00 00 04 088
078
447003
m037
-006026
071
087
081 074 95
046
047
00 01
0043
0122 0121 06 69
3 32 21 00 00 Oh 076
091
464
-022004000
.038
089
075010 09b 95
046
047
od 01
0043
0122 0121 0 o 89
j 32 21 00 00 06 089
075
453009
011
-020
-026
071
076
083 076 vb
046
047
od 02
0043
0122 0121 ob 42
3 51 51 00 00 02 087
105
467
'1041
"077
-020
-075
072
078
068 105 94
046
047
od 02
0043
0122 0121 00 423 51 51 00 00 03 068
092
467
- 015000
-037
-010
075
066
064 089 v4
046
047
00 02
0043
0122 0121 Od 42
3 SI sl oo 00 04 074
078
426
..,006
4000
-012
4000
064
072069 075 b
046
047
od 02
0043
0122 0121 001 42
3 51 51 00 00 05 089
089
471
-035
-037
-020
m063
081
071
o66 083 94
066
047
Od 02
0043
0122 0121 08 42
3 51 51 00 00 06 068
088
444007
4037
.000
.013
077
079
062 089 'lob
046
047
08 03
0043
0122 0121 08 41
162 78 00 00 03 105
136
430009
.037000
-013
064
057
103 136 94
046
047
TO
TA
L D
AT
A B
Y C
ITY
, SC
HO
OL,
AN
D G
RA
DE
II
Od 03
0043
0122 01g1 Od 41
162 78
00
00
0*
131
120
435
*002000
*006
*000
077
082
128
084 vb
046
047
Od 03
004J
0122 ola1 0 41
162 78
00
00
OS
10418
438
-012
*00u
-024
-025
074
073
099
120 105
046
(147
08 03
0043
0122 01e1 Od 41
162 78
00
00
06
096
103
460
006
'000
-020
.000
071
071
095
100 v6
046
047
Od 03
0043
0122 01e1 od 41
162 78
00
00
02
126
134
455
4007
*000
*000
.025
101
072
134
131 vb
046
1347
Od 04
0043
0122 0le1 o 41
307 08
00
00
02
138
072
452009
.000
*000
-.010
088
087
136
064 /3
046
047
Od 04
0043
0122 01e1 05 41
307 08
00
00
03
109
089
448006
.015
.000
.000
080
066
105
078 y4
046
047
Od 04
0043
0122 01g1 06 41
3or oa
00
00
04
115
082
425022
*040
*0370e2
088
091
114
074 v3
046
04/
od 04
0043
0122 01e1 Ob 41
307 08
00
00
05
104
082
476
*004
*000
*000
.019
083
100
095
074 vb
046
04,
Od 04
0043
0122 01e1 05 41
307 Oa
00
00
06
080
085
466
*015
*007000
*013
076
080
076078 4
046
047
Od Ob
0043
0122 olgl Ob 85
200 00
00
00
02
084
081
469
-036
-061
-043
-045
066
084
076
078 v2
046
047
Od 05
0043
0122 0141 Qb 85
200 00
00
00
03
071
070
437
'033
'000
'000
.057
067
080
065
055 9b
046
047
Od OS
0043
0122 0141 05 85
P00 00
00
00
04
096
ob7
415
*014
'000037
.015
070
076
088056 2
046
047
Od os
0043
0122 01e1 Ob OS
200 00
00
00
05
063
o87
433
.009
4000
-024
-009
066
080
056
086 Y3
046
047
08 OS
0043
0122 0141 05 85
200 00
00
00
06
060
064
461
-019
.019-04
.026
076
075
059
064 v3
046
047
Od 06
0043
0122 011 07 22
399 99
00
00
04
115
123
459
-013
.000
*000
.000
076
064
115119 4
04607
Od 06
0043
0122 olgl 07 22
399 99
00
00
03
153
110
399
'026077
000
4042
061
059
153110 6
046
047
0 06
0043
0122 0121 or 22
39999
00
00
04
126
143
391
**001
"005
*000013
064
o77
122145 b
046
047
08 06
0043
0122 01e1 o7 22
399 99
00
00
05
130
119408
*008
*000
*000015
071
07b
130
111 4
046
047
Od 06
0043
0122 Oldl 01 22
399 99
00
00
06
089
122
433
-009
'000
'000
.000
071
077
098
124 V4
04500
_. 4,-
-0
TO
TA
L D
AT
A B
Y C
ITY
, SC
HO
OL,
AN
D G
RA
DE
od
07
0043
0122
17111
04
64
101
01
00
00
02
061
obv
422
.050
.077
062 .069
093
007
061
060
v4
046
047
00
07
o04J
0122
0111
04
54
101
ul
00
oo
03
079
047
424
4034048
*044 034
04 Ob ors 042 94
046
047
00
od
of
07
0043
004J
0122
0122
o1d',
0111
04
04
54 54
t 1
oil
01
ol
ul
00
00
00
00
04 05
ob7
047
063
059
40
444
012
..013
0040
032
0215 .016
*024 .025
07
004
05
062
057
041
05b
051
93
44
046
044
047
047
06
07
0043
0122
0111
04
64
101
01
00
ou
06
067
041
472
-049
-037
*098 .026
073
082
051
042
93
04t 047
06
08
004J
0122
0111
oq
16
299
99
00
00
02
119
094
458
*002
6000
*000 -002
070
063
116
080
V2
046
041
Od
08
004J
0122
0141
Ob
lb 2
99
99
00
00
03
llb
109
453
.048
.037
*024 .054
079
059
097
101
43
046
047
00
08
0044
0122
0111
Ob
lb 2
99
99
00
00
04
117
094
399
.006
.000
.000 .013
075
064
104
090
93
046
047
06
Oh
004'
0122
0111
06
16
299
99
00
00
05
094
095
446001
*000
000 .013
076
080
072
091
43
046
047
Od
08
004J
0122
0111
Op
16
299
99
00
00
06
089
083
443
.016090
*024 .000
067 orb 083
080
93
046
047
Od 09
od 09
004J
004J
0122
0122
0111
0111
09
09
29 29
2 2
99
99
99
99
00
00
00
00
02 Or 04
03 166 128
411
437
4024
'001
040
4000
4061 000
4000 *013
079
066
097
064
116
143
177
126
92 93
Oct, 047
046 047
Od
09
Ou4J
0122
0111
09
29
299
99
00
00
04
141
154
396
*015
*000
*000 *028
076
076
133
148
93
046
047
Od
09
004J
0122
0141
09
24
299
49
00
00
05
10b
141
430007
.000
.024 .012
066
076
100
134 vb
046
047
Od
09
004J
0122
0111
09
29
299
99
00
00
Ob
106
092
443
.014
-026000 .026
080
078
102
087 V4
046
047
os 10
0o4s 0122
0111
11
12
199
99
00
00
07
397
360
408*012000
000 -025
084
075
38030
07
046
047
Od
10
0043
0122
0121
11
12
199
99
00
00
08
381
377
413
*023
*000
000 .028
060
078
367
365
91
046
047
08
11
004J
0122 olel
10
43
347
48
00
00
04
332
312
433006040
000 4000
085
000
330
J02 v1
046
047
Od
12
0044
0122
0111
lb
16
155
63
00
00
07
483
501
4b1
*012
4000
000 i*o25
09100
455
481
94
046
047
TOTAL DATA BY CITY, SCHOOL, AND GRADE
30V90 ONY 110014DS 'AID A9 VTVCI 1'0'101
960 560 II* Co RSO IRO TOO 990 2 cr0. 090*
000* 090* 992 6q0 FRO r0 00 00
66
960 560 RE. /0 90 290 TO t90 2 010*
090* 020* 010 IIE CLO 690 ro oo oo
66
66
66
Ed
OE
27
40
*0
6n*0
6090
E9E0
E9£0
0900
0900
01
60
OI
OI
960 SAO 66*
to 60 160 L96 IRO 2
2906,
010-090* tIO 9tE tRO 201
10
00
00
66
66 2 it
00090
(9E0 WO RO
OT
960
560 sT.
26 ,S0
950
E90
990
2/TO. 010. 00. OIO. 99E 090 AO E0 PI 01
ir,
06.1
19 p0
0090
(9E0000
10
OT
960
S60 72-
TA
900
LEO
410
990
2010.
OtO.
260.
0£0.
2EE
990
9£0
r0
00
00
66
*L
E9/re 040
£9£0
9900
90
OT
960
660 09*
Ro
21?
591
690
990
29?0*
500.
610.
910.
*9E
961
EAT
SO
00
00
66
66
26?
91
0090
E9F0
0900
60
OI
960560 66
20
tE?
691
490
190
20p0",0E0-
020-
/TO- q2E
422
TO2
£0
00
00
66
66
2q2
91
R0,0
(9E0
0900
q0
OI
960
SAO 6I*
Ao
REA
9F0
R9010
2SPO*
9E0*
190*
990*
4IE QE0
620
FO
00
00
66
66
2GE
?0
R090
(9E0
0000
40
OI
960
5A0 E2* 9R
190
090
790
110
2EE0 "'
SIO*
R00
920
919
290
Q90
90
00
00
66
66
E29
Q0090
(9E0
0000
20
OI
960
560 RE* ;to 9S0 990
ESO
990
2010-
020.
0200.
600.,29(
090 R90 10
00
00 AA 66
E29
Q0
9090
E9E0
0900
20
OI
960
91,0 92-
40,
*20
220
290
990
2F9 O.
s50...640-
qq0.
69900
920
tO
00
00
66
69
2g2
20
14090
E9E0
0000
To
0I
960 SAO TE* PA
120
970
1q0
190
2020
510.
940.
100.
09E
610
.,i0
PO
00
00
66
69
292
20
6040
C9E0 gq00
TO
OI
/90
910
PA 29(
86E FRO
TRO
If110600
£00*
600..cut 9I* *2*
II
00
00
ss 19
194
41
1210
2210
p400
91
p0
/90
910
ia RAF 999
*10
910
t 600-
600-
*00*
910-
(2*
(1*
619
II
00
00
66
66
1S9 OT1710 MO
(900
610
140
910
EA RA* Ogs 910
4510
t 600-
900*
ITO*
000*
Ott 909 76S
TT
00
00
61
of
I Pt oT
IPTO 2710
r*00
*I
g0
1,0
940
60
92?02 PRO
900
000.
000.
000.£00
1£9
Es2
(92
60
00
00
E? Ed E
69
10
1210
2210
rt00
£10
/40
940
TA
?I?
992
610
660
620.
000.
000.
TOO.
EC* Qt?
992
10
00
00
ER E2 E
64
10
1210
2210
(too El0
/90
990
W.
IL*
14*
900
9R0
vq0
990.
000*REO
919
RR* OR* RO
00
00
E9 SS
19I
Q1
I210
2210
Fw00 21
00
10
10
0068
0363
0408
05
30
399
99
00
00
05
010
069
390
-021
.015
-025
-039
2060
050
064
064
86
-04
095
096
12
01
0003
0020
0010
01
356
56
00
00
03
025
020
459
-014
-030
'013
'005
1092
092
025
026
695
096
12
02000
0020
00/0
00
12
299
99
00
00
03
092
080
446
-005023
'002
.4:03
A095
001
091
086,94
-33
095
lob
12
03000
0020000
05
27
152
51
00
00
03
082
090
448005
.010000
018
1Ode
091
079
089 94
.0.7095
096
12
04000
0020
0010
00
399
99
00
00
03
115
109
386
.031
.068
4037
.013
1073
074
121
108
095
106
001
0028
0260
0241 od 79
108
11
14
16
01
125
126
403
-004000
000
.000
1087
079
123
124
Yr
-07
056
057
13
01
0028
026n
0221
08
79
108
11
14
16
02
117
108
422
.018
.000
*027015
1081
089
lob
104
vr
-07
056
057
13
01
ooed 0260
0221
08
79
i08
11
14
lb
05
087
111
482
-013000
6014
-028
1082
065
074
067
vr -or
105
106
13
01
0028
0260
0221
08
1Y
108
11
14
16
06
102
093
411
.017
-016
-004
.035
A074
086
108
094
9r
.07
Olb
106
1.4
02
002d
0260
0221
06
11
124
75
21
23
01
104
085
1)81
.041
.076
.023.04
1079
080
100
082
96
.11
056
057
13
02
ooed
0260
0221
00
77
124
25
21
23
02
081
oar 40f,
-004
-028
.005
-004
1081
084
074
084
v6
.1106
057
lJ
oe 0u2d u2a0
0221
oa rr
124
25
21
23
05062 or 469
.031
-010
-003
-042
1081
065
052
054
vb
.11
105
lob
13
02
0028
0260
0221
06
77
124
25
21
23
oh
062
048
380
.050
.6419
.064
.041
A077
070
063
050
96
.11
016
lob
13
03
0020
0260
0221
05
61
256
56
11
13
01
096
115
412
*044
.000
.023
.091
1078
109
062
099
v6
.00
056
057
13
03
0028
0260
0221
05
61
256
56
11
13
02
075
081
438007
-010
-011
*002
1078
o97
064
078
96OP
056
057
13
03
002e
0260
0221
ob 61
P56
56
11
13
05
054
obi
473
-032
-019
-025
-022
1082
obb 048
037
96
"00
105
106
13
03
0028
0260
0221
05
61
- 56
56
11
13
06
050
040
363053
.032041
.058
A059
089 050
042
96
.00
016
106
TO
TA
L D
AT
A B
Y C
ITY
, SC
HO
OL,
AN
DG
RA
DE
iiitk
lA46
0111
11di
Agi
litah
litga
l
3amarwelOOHDS'AIIDAfivivalY101901 sot 60- RA 99f iOt iRO IRO I ?tO 000. ;00. 600.. PEt Z9E 62, II it 91 ti 69 2 ?r ,fte p?00 ot pi
901 SIT II- sA OSE 06f C10 p/0 I SOO. SIO. /10. 010. i6E OFt f't RO qf 2E is 65 E qt 91 1720 0920 .700 AO pi
901 ;It 10* LA IR? 1IE RRO COO I 920 LE0* 520* 0E0*.t6E Of 62E RO 0 go 41 0/ gp it mo owl 0700 go 09CI 910 41- 90 790 290 ER0 790 v 590* tE0* 1t0* ESO. 19i p50 790 90 p0 10 26 16 E 9Q p0 MO 0920 p700 10 rt
90' SOT t1m 90 120 650 RtiO qp0 t 6c0- LOO 6/0 4E0- 19t Tq0 Q0 RO 10 26 16 E cm p0 MO 0920 pPOO JO pi
LSO 90 t1 9A 290 060 RR° I10 1 0E0* 000* 970- 900* 99E PRO 560 70 RO 10 26 16 E qc RO IP20 0920 0700 /0 pi
LSO 90 41' 90 460 901 601 410 I Ii0* 000. q/0. 1,0. 96E QEI OQI 10 p0 i0 26 (6 E 9c RO 1720 0920 p?00 JO pi
gOt
puT
910
soT
40..
40-
poi
qA
160
690
290
610
Ei0 990
pi0 200
y
I
£10,
000.
00.
410.
920+
RIO.
900,
000.
91 Ott
gEt 601
001
coli
90
sO
pt Ot t, sei E c
f9 09 99 99 pi
0
140
1720
MO0920
0920
000
p200
90
90
0FT
ici0 40 OP' 40 TOT 990 900 2o0 I frOo 000. 920. 910. GOE 160 f60 20 Ft Ot tt tt C fij PO MO 0920 pp00 RO FI
/GO 9c0 40m cm FIT q60 010 990 1 940* 000* I70- siO* 69E 44E1 121 TO p, 04 iyt *it E pi p0 1220 0920 pp00 qo pt
901 9T0 0Im /A ItO 9E0 sso 190 v 000- 420 *AO* ROO 6/E Qt0 940 90 pR tp il II E 61 40 IPZO 0920 0700 Q0 rt
got sot 01- LA etc, Eto no ipso I 210 - 000k £40- Ito- tiff, 190 40 c0 op AO IT II C 61 tO MO 0920 14200 O0 FT
/.0 q0 01 LA EEO 6E0 LRO FLO I 90. 090. 210. 140. 99E REO 190 20 RR tR 11 II E 61 40 1220 0920 0200 q0 pi
/CO 950 Olm jog, AEO 920 00 IRO 1 000. 950. qQ0. 10. 4/5 Tq0 fQ0 TO Po 9R II IT C 6c 40 tno 0920 p?oo co rt
901 910 20* 9A fli0 610 6S0 flO V 690 9,0. it0. IFO. 6SE IRO ORO 90 60 71 SR OR 2 qq 00 IPZO 0920 0700 tO p1
901 501 20* 96 41.0 290 490 MO I 100. 000. 0s0. 110. tSt TOT 910 GO 60 Fi SW 00 2 95 PO 1220 0920 ;WOO tO ri
LGO 90 20* 9. 960 960 490 WO I 610* 000* 00). 9104 tot rtt tOT 20 60 21 5 i4 09 2 qs RO MO 0920 op00 40 pl
ISO 90 20 W. OII 021 FAO IRO 1 IcOm SEO 920", 470 E0t qti tgi TO 60 FI SR 09 Z 9Q O0 1720 0920 '4200 40 fT
th
13
11 Oodu 0260
0221 19
16
323 22 23 25 11
589
588
490
-021
-017
-037
-030
1103 100 522 564 V8 .05 loS 106
14
01 0032 0146
013.1 07
45
377 ol oo oo 03
091
080
360
.004
-020000
)010
1065 075 104 075 v2 -16 105 106
14
01 0044 0146
0133 or
45
371 ol 00 00 04
061
084
417
.000
.004
-025
.010
1076 003 089 083 v2 .16 105 106
14
01 0032 0146
0133 0/
45
377 s1 00 00 06
084
07D
435001
00c
-018
-020
1-
088 094 064 069 92
16 105 106
14
02 0032 0146
0133 01
32
304 03 00 00 03
085
111
365021
+010
.051
-008
1095 087 100 103 v3 -12 105 106
14
02 0032 0146
0133 01
32
304 03 00 00 04
099
096
460
-011
-005
-1)10
-015
1094 097 074 089 v3 -12 105 106
14
u2 0032 0146
0133 01
22
304 03 00 00 06
015
088
470004
014
010o25
1o9 4 101 Oa 070 v3 -12 lus 106
14
n3 0o32 0146
0133 08
16
344 42 00 00 03
082
087
376
-017
-032
-010
-010
i081 081 089 078 v4 -04 105 106
14
03 0032 0146
0133 08
16
344 42 00 00 04
078
094
440001
030005010
1095 085 069 085 v4 -04 105 106
14
03 0032 0146
0133 08
16
344 42 00 00 06
079
075
430
-020
.020
-010
-062
1112 odd 081 064 V4 .04 105 106
14
o4 0034 0146
0133 09
30
332 29 00 00 03
097
090
380-00038
000
-025
1073 066 081 085 81 04 105 106
14
04 ooJe 0146
0133 09
30
332 29 Go 00 04
107
085
407
.039
4090040
4046
A091 070 064 060 vl .04 105 106
14
04 0032 0146
0133 09
30
332 29 00 00 06
072
089
426
-002
-010
-016
-010
1030 099 077 076 91 04 105 106
14
1)
0032 0146
0133 02
35
397 96 00 00 03
066
035
344
4003
*000
-020000
1060 072 033 035 84 -01 105 106
14
05 0034 0146
0133 02
35
397 96 00 00 04
059
036
391
.005
.036020
.010
1079 02 033 031 94 01 105 106
14
05 0032 0146
0133 02
35
397 96 00 00 06
060
023413
-005
.006
-020
.045
1090 073 039 021 94 .01 105 106
0.%
14
06 003e 0146
0133 08
39
383 )14 00 00 03
094
085
381
.013
.010
-025
.025
)085 081 081 066 v3 .08 105 106
14
ob 0034 0146
0133 08
39
383 84 00 00 04
085
103
42001701004o
-020
1083 077 083 078 83 -08 105 106
-ct
TO
TA
L D
AT
A B
Y C
ITY
, SC
HO
OL,
AN
D G
RA
DE
'
4.ta
t-, A
t,
,1,4
4,11
,1,
s
. City
School "--Disffict 470
Title
1 ADM
4*<-
/ Distnct
Pre
Totoi
Title 1
District
S Post
0/ Total
Title 1
School
ADM
00.
School
Mob.
%°s,,
Pov. Index %
Neg.
Pre % Neg.
Post
v % Span
Spk Pre
°C.
% Span
Spk Post
Grade 0
Grade
ADM
Pre Grade
ADM
Post Mean
Reading
Pre
e
A Mean
(Port-Pre) / A 1st
Decile e
A (Post-Pre)
c.)
e e
Std. Dev,
Pre
/ Std.
Dev, Post
Q3 (Post-Pre)
0 Sam.
Size Pre Sam.
Size Post
0
Gr. Att.
Rate Pre
OG ".
Test DWI:
Pre
/ o
Gr. A Attend.
Rote Test
Date
Pose
.01.1.1.MM
156 67TMP-115
APPENDIX B
DESCRIPTION OF VARIABLES
Because school districts vary in their administrative proceduresand classifications, identical forms of information for all districtscould not be obtained. There was, therefore, some variation inthe types of material included under any particular definition. Forexample, total Title I funds may be based upon budgeted amounts orexpenditures; mobility may be calculated during just the academicyear or for a calendar year; etc. The definitions which follow rep-resent the kinds of information we desired to obtain, and the figuresused in thisreport should be interpreted as approximations whichmay vary somewhat among districts.
Variables Description
District Title I ADM
. District Total Title I $1965-66
District Total Title I $/966-67
Grade
The average daily membei4hip of allschools (public and parochial) in thedistrict receiving Title I funds.Usually this figure was obtained fromapplication forms for Title I.
Expenditures (or budget) for Title Ifrom February through August 1966.
Expenditures (or budget) for Title Ifrom September 1966 through May 1967.
The school system classification ofgrade used for interpreting achieve-ment results. Some non-graded classeswith their appropriate grade levels wereincluded but most special classes formentally retarded were excluded.
APPENDIX B
Variables
% Negro 1965-66% Negro 1966-67
% Spanish Speaking1965-66
% Spanish Speaking1966-67
Mean Achievement TestScore - pre
Description
157
The percentage of the students in aschool identified as Negro in the dis-tricts' racial census. In cases whereno formal census was done, estimatesby district personnel were used.
District census estimats of the per-cent of the school population whichcomes from families where the primarylanguage in the home is Spanish and forwhich it is reasonable to assume thatschool performance is hampered by aninadequate knowledge of English.
The Mean Standard T score in thereading subtest or composite achieve-ment test by a specific grade at aparticular school in academic year1965-66.
Mean Achievement Test As above for academic year. 1966-67.Score - post
Change in Mean Achieve- Mean Achievement Test Score postment Test Score minus Mean Achievement Test score
pre.
Change in 1st decile, 1st As above, for the respective points inquartile, and 3rdquartile
Sample Size 1965-66Sample Size 1966-67
the achievement test score distributions.
The number of pupils in a grade at agiven school for whom achievement testdata were obtained in 1965-66 (pre) and1966-67 (post).
Attendance Rate (ADA/ Average Daily Attendance (pupil daysADM) attended divided by number of days
school is in session) divided by ADM.
Change in Attendance Attendance rate in 1966-67 minus at-Rate tendance rate in 1965-66.
158
Variables
Test Date
School ADM; GradeADM
School Mobility
Poverty Index
Average Title I dollarsper Pupil
Weighted Averages
Description
67IMP-115
Month and year achievement tests wereadministered to a given grade.
The Average Daily Membership (1965-66); as defined by the Office of Educa-tion (Reference 11); aggregate daysmembership divided by number of daysschool is in session. In some casesenrollment figures were substituted.
The sum of (a) total number of studentsentering the school after formal open-ing of the school years, and (b) totalnumber of students leaving the schoolduring the school year, the sum thendivided by the school ADM. (If infor-mation required for this definition wasnot available, substitute measureswere used.)
A ranking of Low (1), Medium (2), orHigh (3) to indicate how each schoolcompares to other Title I schools ina school district. (High means greatestdegree of poverty. )
District total Title I dollars 1965-66 plusK times (District Total Title I dollars1966-67) divided by District Title I ADM,where K is the fraction of the 1966-67academic year that elapsed by the 1966-67 test date for that grade.
The observed values of changes in themeans (AX), the 1st deciles (ADO, andthe quartiles (Li(2) are weighted by theaverage number of pupils in the pre andpost year.
APPENDIX B 159
nweightedaverage =
E .1 + m2)in
i=1 (m1
+m2
).1
i=1
[change in achievement]i
where m1
and m2 are the number of
pupils in 1965-66 and 1966-67 respec-tively (in the ith observation). Thesum of the weights add to one.
160 67TMP-115
APPENDIX C
STATISTICAL MODELS FOR ANALYSIS OF ACHIEVEMENT DATA
CONFIDENCE INTERVALS FOR OBSERVATION BY GRADE
The confidence intervals for A mean are based on the fact thatregardless of the form of the underlying distribution, the distribu-tion of the sample mean is approximately normal for large samples._ _Let X
0and X
1be the sample means in two successive years (Pre
and Post) for a particular grade and school, computed from samplesof sizes n
0and n
1,respectively. Let AX = 4^,X1 - Xo . Then, as-
-suming independence of X1 and R
0(they represent two different groups
of individuals) liX will be normally distributed with mean Sand vari-
ance a2/n0
+ a 2 /n1 = (72 (1/n0
+ 1/n1) where 6 is the true difference2in the means, and a is the (assumed) common variance. (The as-
sumption of common variances seems reasonable after observingthe small variation of the sample variances, as shown in the rawdata. Then, a 90 percent confidence interval for the true difference
6 is equal to LTC ± 1.65 s (1/n0
+ 1/n1)1/2, where s is the pooled
estimate of a, given by
s
[
2(n
0- 1) s 02 + (n
1- 1) s
1
n0
+ n1 - 2(1)
2where s and s 12 are the pre and post sample variances, respectively.0
The pooled sample sizes are never less than 30, and usually muchlarger.
\
APPENDIX C
-
161
Approximate confidence intervals for A 1 s t decile (adj.) and LI1st quartile (Lq 1) are computed from the Chebyshev inequality, andby estimating the standard deviations of the first decile and firstquartile from the equations by
ard
Sd = (1. 7094)s/n1/ 2
1
S = (1. 36 26)Sin1/2ql
respectively.* Thus, proceeding as with the b mean above,Ad
1= Ld
11- bd
10(the Post-Pre difference) has a true mean d
and variance
2
d2 2a = (1.7094) a (1/no + 1/n
1)
1
Writing Chebyshev's inequality in the form
_ d I > 1/.. a ]
(2)
(3)
(4)
and choosing X so that 1/A2 = O. 1 (i. e., X = 07, the approxi-mate 90 percent confidence interval for IS d1 is
ad1
± vro (1.7094) s (1/n0 + 1/n1
)1/ 2
, (5)
where s is the pooled estimate given by (1) above. The confidenceinterval for ,Aq
1is identical except for the constant 1.36 26 in place
of 1.7094:
± Vili (1.3626) s (1/n0 +1/n1)1/2
*See Reference 1 2, p. 243, for the origin of the two numerical
constants.
(6)
'4k
16267TMP-115
RATIONALE AND MODELS FOR MULTIVARIATE ANALYSES
Multiple regression and analysis of covariance were the two multi-variate analyses used in Phase I. They are important tools in analy-
sis of non-experimental data. They are expecially helpful when there
are large differences in so-called environmental conditions amongsample observations. These techniques not only help to identify theeffect of specific environmental conditions on the variable of primeinterest but they help to reduce sampling variation and, thereby,make tests for statistical significance more discriminating.
The regression mo-lel used in Phase I, namely
Y = a0
+ al ($) + az (M) + a3
(%N) + a4
(L) + a5
(A) + E (7)
has major limitations because of the different levels of aggregationon data for the different variables. The dependent variable (Y) wasmeasured at the grade level and was defined as change in meanachievement score or change in reading more at the lowest decile,lowest quartile, or highest quartile. The Title I dollars per student
were computed from average dollars per student for an entire schooldistrict. The mobility rate (M) and percent Negro (%N) were comput-ed for the entire school. The initial achievement level (L) is theaverage score for a specified grade in the 1965-66 test. The attend-ance rate (A) was computed for the entire school.
For the variables which were measured at the school or districtlevel the observations by grade are, obviously, not independent.This violates one of the assumptions in regression analysis and itmeans that less confidence can be placed on the estimated regressioncoefficients. For example the coefficient on the $ variable in Tables14 and 15 was, in essence, computed from seven observations, and
not 240 as indicated by the number of grade observations. With only
seven observations it is very likely that this variable is acting as aproxy variable and the estimated coefficient represents the effects ofvariables other than level of Title I expenditures.
When the model s estimated from data within a specific schooldistrict the dollars valiable ($) was dropped. The mobility and at-tendance rate variables also had to be dropped in some cases becausedata were not available.
'1
APPENDIX C
In analysis of covariance the variables M and A had to be deletedfrom the model in Equation 7 for the test of significant differencesamong schools in District 13. These variables are measured only atthe school level. This means that there would be no variation withinschools and it would not be possible to estimate the reqiiirerl statisticsin analysis of covariance for testing significant differences amongschools (see Table 44).
In analysis of covariance, the concepts of analysis of variance andregression are combined to provide a more discriminating analysisthan is afforded by either of the two component parts. The F testin analysis of covariance is designed to determine if there is a signi-ficant difference in the adjusted means of the depenctent variable amongspecified groups. The adjusted mean is calculated from the equation
adj (Y.) = Y. - E a. X. ,J J
where Y. is the observed mean for the ith group, a. are estimatedregression coefficients, and X. are the variables thak are known tocontribute to the variation in Y3 (see list of covariates in Tables 44through 47).
In essence, the observed variation in the dependent variable amongspecified groups is divided into the three components: (1) variationdue to the -:ovariates (X.), (2) variation due to factors associatedwith each group, and (3) variation due to the unknown randomcomponent. The F value in the F test is based on the ratio of thesum of the second and third components, to the third component.That is,
F =V2 + V3
V3
and the value of F in Table 44, for example, was calculated as
F = 1.416 = 4. 573. 23
In the example of Table 44, the F value is not significant at the5 percent confidence level and we cannot reject the null hypothesisthat the "school effect" contributes no significant influence on the
(10)
164 67TMP-115
Table 44. Analysis of covarianceaDistrict 13 schools (using change inlowest quartile as a measure of change in achievement).
ISource DF YY Sum-Squares Sum-Squares DF
(Due) (About)Mean-Square I
AmongSchools 10 65.5
ErrorWithin 21 132.4 71.0 61.4 19
(Total) 31 197.9 90.8 107.1 29
Difference for testing adjusted treatmentmeans 45.7 10
........--
3.23
4.57
Null Hypothesis. No difference among schools after adjusting:
F = 1.416 < F (5%) = 3.43
Note:a F or discussion of the use of covariance, see pages 34 and 163.
Table of Coefficients for Covariates:
VariableAmong Schools
CoefficientError (Within
Coefficient Std. Error T-Value
L -0.409 -0.495 0.107 -4.65
$ -0.006 -0.025 0.012 -2.11
Note:L is mean achievement level in 1965-66 and $ is effective Title I dollars perstudent (see definitions in Appendix B).
APPENDIX C 165
Table 45. Analysis of covarianceaseven school districts (using change inmean reading score as a measure of change in achievement).
e xi..I Source L.m,
1 1 Sum:,-Squares Sum-Squares DF(Due) (About)
Mean Square
AmongDistricts 6 169.8
Error(Within) 233 1214.3 350.7 863.6 230
(Total) 239 1384.2 434.0 950.3 236
Difference for testing adjusted treatmentsmeans 86.6 6
3.76
14.4
Null Hypothesis. No difference among districts after adjusting:
F = 3.85 > F (1%) = 2.90
Note:a
F or discussion of the use of covariance, see pages 34 and 163.
.
Table of Coefficients for Covariates:
Variable Among SchoolsCoefficient
'IL
A
$
Error (Within)
Coefficient St. Error T-Value
-0.170 -0.431 0.045 -9.52
0.158 0.078 0.062 1.25
0.013 -0.004 0.113 -0.34
Note:L is mean achievement level in 1965-66, A is attendance rate, and $ iseffective Title I dollars per student (see definitions in Appendix B).
i
166671MP-115
Table 46. Analysis of covarianceseven school districts (using changein lowest deci le as a measure of change in achievement).
cnurce DF YY Sum-Squares Sum-Squares DF
(Due) (About)Mean-Square
AmongDistricts 6 78.9
Error(Within) 233 2366,9 253.6 2113.3 230
(Total) 239 2445.8 238.1 2207.6 236
Difference for testing adjusted treatment
means 94.4 6
9.19
15.7
Null Hypothesis. No difference among districts after adjusting:
F = 1.71 < F (5%) = 2.14
Note:a F or discussion of the use of covariance, see pages 34 and 163.
Table of Coefficients for Covariates:
Variable Among SchoolsCoefficient Coefficient
Error (Within)
St. Error T-Value
L -0.136 -0.371 0.071 -5.25
A 0.240 0.098 0.097 1.01
$ 0.001 -0.019 0.018 -1.07
Note:I. is mean achievement level in 1965-66, A is attendance rate, and $ is
effective Title I dollars per student (see definitions in Appendix 8).
APFENDIX C 167
Table 47. Analysis of covarianceaseven school dktrkts (using change inlowest quartile score as a measure of change in achievement).
I Source DF YY Sum-Squares Sum-Squares DF(Nap) (Akvit)
Mean-Square
AmongDistricts 6 105.0
Error(Within) 233 1790.3 329.1 1461.2 230
(Total) 239 1895.3 334.9 1560.4 236
Difference for testing adjusted treatmentmeans 99.2 6
6.35
16.53
Null Hypothesis. No difference among districts after adjusting:
F = 2.60 > F (5%) = 2.14
Note:aF or discussion of the use of covariance, see pages 34 and 163.
Table of Coefficients for Covariates:,.......
Variable Among Schools Error (Within)Coefficient
Coefficient St. Error T-Value
L -0.149 -0.418 0.059 -6.31
A 0.164 0.050 0.081 2.54
$ 0.009 -0.019 0.015 0.36
Nate:L is mean achievement level in 1965-66, A is attendance rate, and $ iseffective T:tie dollars per student (see definitions in Appendix 8).
t.... ...................
168 671MP-115
dependent variable. By school effect we mean the net influence of allvariables associated with school. The associated variables includesuch things as type of Title I program, school facilities, and qualityof instruction.
In the analysis of covariance for the 11 schools in District 13, theamount of Title I dollars was not used for specifying different groupsbecause the only information on expenditures was for the entire schooldistrict.
In the analysis of covariance for the seven school districts it wouldhave been possible to specify different groups on the basis of expendi-tures. However, it seemed best to specify this as one of the variablesin the regression equation (i.e., one of the X. covariates in equation7 above) since numerical values for observadon on this variable arereadily available. For variables such as type of Title I programwhich cannot be easily expressed in numerical form, specification ofgroups based on type of program and the use of analysis of covarianceis an alternate to regression analysis.
The F values for Tables 44 through 47 are discussed in Section 2.It is advisable to consider more research in analysis of covariancein Phase II since time permitted only a few analyses in Phase I.
RESULTS FROM ANALYSIS OF VARIANCE AND ANALYSIS OF COVARIANCE
The summary of results from analysis of covariance were pre-sented in Table 10 in the main text. Tables 44 through 47 give fur-ther details.
Table 48 shows the results from analysis of variance among schooldistricts. The results from both the analysis of variance and analysisof covariance show a significant difference among school districts.
APPENDIX C
Table 48. Analysis of variance among school districts (LiX).
169
Sum of Squares DF Mean Square F Ratio
AmongDistricts 185 . 04 10 18.50 3.77
WithinDistricts 1508.35 307 4.91
Total 1693.39 317
F = 3.77 > F (1%) = 2.37 indicating there Li a significant differencein average change in the mean among school districts.
CORRECTION FOR EFFECT OF ERRORS OF MEASUREMENT
ON OBSERVED CORRELATION BETWEEN INITIAL TESTSCORES AND CHANGES IN TEST SCORES
Errors of measurement produce a negative correlatinn betweeninitial achievement level and change in achievement as measured,even if there is no true correlation. Similarly, any true correlationthere may be, will be obscured in the analysis of observed data.
Let the observed initial value, x, represent the sum of the truevalue,x*, and an error of measurement' el .
x = x*+ e .1
Similarly, let the measured final value, z, represent the sum of thetrue initial value, a true gain, g, and an error of measurement e 2.
z = x*+ g + e2 .
The observed difference between the two, y = z - x, becomes:
y = g + e2
- el .
(12)
(13)
The reason for the apparent negative correlation between x and y,even if x* and g are uncorrelated, is the presence of "+e 1" in theexpression for x and "- el" in the expression for y.
170 67TMP-115
Peters and Van Voorhis (Reference 13, p. 460) state that themagnitude of the correlation due simply to unreliability of measure-ment is
r = -x, y 1,11/2 (1-rx) , (14)
where rx is the reliability of the instrument used in measuring xand z. This means that if there is no reliability to the instrument,a correlation of -0.7 will be found between initial level and obs.arvedgain, and that as reliability increases the correlation due to mea-surement error alone will drop.
Thomson (Reference 14, pp 321-324) provided the followingformula to correct for errors of measurement:
=rx,,.%g
ar + x (1-r )xy a x
Y
1
aY
r [cr 2 a 2 (1-r ) - cr 2 (1-rxx y x z
where, using the terminology in (11) through (13) above,
2ae
1rx = 1.0 -2
ax
2ae
2r = 1.0 -z 2
az
(15)
(16)
(17)
The Thomson formula requires estimates of the reliability of themeasuring instrument and measures of the variance of initial value,(x) final value (z) and gain (y). At the present state of our studythere is no measure of the reliability of the test instrume.it for mea-suring achievement of school-grade units. In spite of the lack of suchestimates, it was desired to see how the measured fairly high negativecorrelation between initial mean achievement !evel and change in the
\
APPENDIX C 171
mean (i.e., -0.43 for the total of 314 observations) would be alteredwhen corrected for errors of measurement of various orders. Forthese calculations, it was assumed that el and e 2
are not correlatedwith each other or with x=:' cr g.
Usine Thomson's formula the "true" correlation between initialmean achievement level and geir -?.s calculated for various assumedlevels of reliability of test ins .:- f for grade-level observations(see Table 49).
Calculations using reliability values of 0.85 or less yielded valuesgreater than 1. 00 or imaginary numbers, suggesting that such anassumption was incompatible with the observed values of variancesor the assumption that x*, g, el, and e 2
are uncorrelated. For re-liability between 0. 9 and 1.0 the correlation between the true gain andtrus! initiall.wel differs only slightly from the -0.43 coefficientbetween the observed measures. It appears that the observed fairlyhigh negative correlations between initial level and gain in meanachievement scores cannot be attributed to an artifact of test un-reliability, but should be accepted as indicating a true negative re-lationship.
-foible 49. Relation between assumed relic; ility values of estimates ofmean test scores and "true" correlation between initiallevel and gain (when measured correlation is -0.43).
Assumed Reliability ofPre and Post Tests
1. 00
0. 98
0.95
0. 90
0.85
"True" Correlation BetweenInitial Achievement Level and Gain
-0.43
-0.42
-0.41
-0.40
Assumption not consistent withobserved values of variances
172
VARIANCE OF WEIGHTED AVERAGES
The formula for the variance of aservation is given equal weight is
V =[x. - x- ]
2
3.
n - 12
67TMP-11:
simple average where each ob-
1
The estimated variance for a weighted average such as
is given by the formula
nV = E [bij2 a 2(xi)
i=1
(18)
(19)
. (20)
In this study the x. is a parameter of a distribution of1m. pupils.
3.
The variance a 2(x.) can be3.
a21
a2m.
1
where m. is the number1
servations on the :A sets7-is only necessary to estirr
,
represented as
scores from
(21)
of pupils and a 2 is a constant for all ob-of pupils. The b. and m. are known and it.ate a2 in orderlto use tormula (20).
We know that each value oi
2m. [x.-- x ]
1 1
is an estimate of a 2. Therefore the best estimate of al is
(22)
i
1--.-AmnerMNIVILANNIIIIIIIINIIMAu..."
APPENDIX C 173
2m.
i=1
where n is the number of observations of sets of pupils.
(23)
The standard errors reported for weighted averages in Tables 6and 9 were computed by substituting the value of (V from Equation23 into Equation 21 and substituting the values from Equation 21 intoEquation 20.
671MP-115
APPENDIX D
SPECIFICATIONS OF ACHIEVEMENT TEST DATA
The following tables describe,. for four school districts (1, 3, 4and 13), the form of the achievement test data obtained and the s.:epsnecessary to convert them into national percentile scores. Thesources used for conversion were usually norms or equivalencetables provided by the publishers of the particular tests. In each con-version process the last step was conversion of percentile scores in-to Standard T scores (see Appendix F). Corresponding informationfor the remaining cities is available but has not been included.
The following is a list of abbreviations and terms used in Tables50 through 53:
Type of Test:
ITBS Iowa Tests of Basic Skills
ITED Iowa Tests of Educational Development
MAT Metropolitan Achievement Tests
SAT Stanford Achievement Test
STEP Sequential Tests of Educational Progress
Reading A subtest of the MAT, ITBS, and STEP
Para. Mng. Paragraph Meaning subtest of the SAT
Composite A summary score for several subtests used o.ithe ITED and ITBS
Ability to interpret literary materials A subtest of the ITED
APPENDIX D ,75
Forms and Batteries of Tests:
Pri. Primary battery
Int, Intermediate battery
Adv. Advanced battery
Numbers and letters refer to specific forms.
Types of Scores
Converted scores A statistically derived score used on theSTEP in order to make scores from dif-ferent forms of the STEP comparable.
Grade score, Grade Equivalence or GE Scores reflectingthe grade placement of pupils for whomthe given score is the average or norm.
Percentile score A score or rank indicating the percentageof pupils in the standardization group atthe given grade placement having scoresless than the given score.
Raw score The number of correct answers in a test.
Standard scores A raw score used on the ITED which wasconstructed to have a median of 15 andstandard deviation of 10. With the passageof time, this standard no longer appliesuniformly.
Standard T-score A score representing the placement of theindividual in a norm population cf normallydistributed scores around a mean of 50and standard deviation of 10.
Tab
le 5
0.P
repa
ratio
n of
Dis
tric
t1
achi
evem
ent t
est d
ata.
Gra
deT
est &
Sub
test
For
m U
sed
Mon
th T
este
dF
orm
of D
ata
Con
vers
ion
Pro
cess
Prio
r to
Obt
aini
ngS
tand
ard
T-S
core
s
Pre
Pos
eP
reP
ost P
reP
ost
Ste
p 1
Ste
p 2
Pro
duct
Sou
rce
Pro
duct
Sou
rce
4S
AT
Int.
Int.
Late
Apr
i 1 2
4G
rade
Gra
de %
i le
Yea
r-en
d
Par
agra
ph1-
XI-
X M
ay-M
ay 5
Sco
res
Sco
res
Sco
res
Nor
ms
Mea
ning
5S
AT
Int.
Int.
Late
Mar
chG
rade
Gra
de %
i le
Yea
r-en
d%
i le
Mid
-yea
r
Par
agra
phII-
X1-
W M
ay6-
17S
core
sS
core
sS
core
sN
oran
Sco
res
Nor
ms
Mea
ning
(Pre
)(P
ost)
6S
AT
!nt.
Int.
Feb
.Ja
n.G
rade
Raw
Gra
deG
rade
%ile
Mi d
-yea
r
Par
agra
ph11
-XII
-YS
core
sS
core
sS
core
sS
core
Sco
res
Nor
ms
Mea
ning
(Pos
t)T
able
7S
AT
Adv
. Adv
. Lat
eA
pril
14 G
rade
Raw
Gra
deG
rade
%i l
eY
ear-
end
Par
agra
phW
XM
ay-M
ay 5
Sco
res
Sco
res
Sco
res
Sco
reS
core
sN
orm
s
Mea
ning
(Pos
t)T
able
8S
AT
Adv
. Adv
. Lat
eD
ec.
Gra
de°
Raw
Gra
deG
rade
%i l
eB
egin
ning
Par
agra
phW
XM
ay5-
17S
core
sS
core
sS
core
sS
core
Sco
res
& Y
ear-
end
Mea
ning
(Pos
t)T
able
Nor
ms
Not
e:
aln
1966
-67
Col
lege
Pre
para
tory
cla
sses
wer
e no
t tes
ted.
To
mak
e gr
oups
com
para
ble,
stud
ents
in th
ese
clas
ses
in 1
965-
66
wer
e el
imin
ated
from
:;*
eigh
th g
rade
data
.
Tab
le 5
1.P
repa
ratio
n of
Dis
tric
t 3 a
chie
vem
ent t
est d
ata.
Gra
des
Con
vers
ion
Pro
cess
Prio
r to
Tes
t &F
orm
Use
dM
onth
Tes
ted
For
m o
f Dat
aS
tand
ard
T-S
core
sS
ubte
stP
r.,
Pos
tP
reP
ost
Pre
Pos
tP
rodu
ctS
ourc
e
4, 6
, 8IT
BS
34
Jan.
Jan.
Gra
de%
Ile19
65-6
6M
i d-y
ear
Com
posi
teeq
uiv.
scor
esda
ta to
tabl
es o
fst
ate
nat'l
.na
t'l. °
Idle
nat'l
. GE
norm
sano
rms
norm
s&
%ile
nor
ms
10, 1
2IT
ED
Y-4
X-4
Sep
t.S
ept.
%i I
e%
i le
none
nor
eC
ompo
site
scor
essc
ores
nat'1
.na
t'l.
norm
sno
rms
Not
e:a
Gra
de e
quiv
alen
t sco
res
wer
e lis
ted
in s
corin
g in
terv
als;
mid
poin
ts o
f the
sco
ring
:nte
rval
s w
ere
used
.
Tab
le 5
2.P
repa
ratio
n of
Dis
tric
t 4 a
chie
vem
ent t
est
data
.
Gra
des
Tes
t &F
orm
Use
dM
onth
Tes
ted
Sub
test
Pre
Pos
tP
reP
ost
For
m o
f Dat
aC
onve
rsio
n P
roce
ss P
rior
toO
btai
ning
Sta
ndar
d 1-
Sco
res
Pro
duct
Sou
rce
4B, 6
8,IT
BS
44
Rea
ding
Oct
.O
ct.
Gra
de e
quiv
a-N
atio
nal
Nat
iona
l nor
ms,
%i l
e sc
ores
begi
nnin
g of
year
10B
,S
TE
P?
?O
ct.
Oct
.C
onve
rted
Nat
iona
lN
atio
nal n
orm
s,
128
Rea
ding
scor
es%
i le
scor
esbe
ginn
ing
ofye
ar°
Not
e:
°Mid
poin
ts o
f per
cent
ile in
terv
als
wer
e us
ed.
1Y11
1111
- 03
Oik
iat4
Tab
le 5
3. P
repa
ratio
n of
Dis
tric
t 13
achi
evem
ent t
est d
ata.
Gra
deT
est &
Sub
iest
For
m U
sed
Pre
Pos
t
Mon
th T
este
dP
reP
ost
For
m o
f Dat
a
Con
vers
ion
Pro
cess
Prio
r to
Obt
aini
n S
tand
ard
T-S
core
sS
tep
1S
tep
2S
tep
3P
rodu
ctS
ourc
eP
rodu
ctS
ourc
eP
rodu
ctS
ourc
e
1S
AT
Pri.
Pri.
May
May
Raw
Sco
res
Gra
deS
ubte
stN
otio
nal
May
-P
arag
raph
I -W
I -W
Sco
res
Con
vers
ion
%i l
e S
core
sJu
ne
Mea
ning
Tab
leN
orm
s
2S
AT
Pri.
Pri.
May
May
Raw
Sco
res
Gra
deS
ubte
stN
atio
nal
May
-P
arag
raph
II -W
II -W
Sco
res
Con
vers
ion
%ite
Sco
res
June
Mea
ning
Tab
leN
orm
s
5S
TE
P
Rea
ding
4A4A
Oct
.O
ct.
Num
ber
ofco
rrec
tan
swer
s
Con
vert
edS
core
sS
ubte
stC
onve
rsio
nT
able
Nat
iona
l%
ile S
core
sF
all N
orm
s
6S
AT
Int.
Int.
Jan.
Oct
.R
aw S
core
sG
rade
Sub
test
1964
Edi
tion
Edi
tion-
Nat
iona
lJa
n-A
pril
Par
agra
phM
eani
ngK
MII-
WS
core
sC
onve
rsio
nT
able
grad
e sc
ore
equi
v. fo
req
uiv.
bT
able
s%
ileS
core
s&
Sep
t. -
Dec
.F
orm
KN
orm
sc(I
nt. I
I)
8S
AT
Adv
.A
dv.
Oct
.-O
ct.
Raw
Sco
res
Gra
d.S
ubte
stA
s ab
ove
Edi
th.s
n-N
atio
nal
Sep
t. -D
ec.
Par
agra
phM
eani
ngJM
WN
ov.
Sco
res
Con
vers
ion
Tab
lefo
r F
orm
Jeq
uiv.
,T
able
s°%
ileS
core
sN
orm
s(A
dv. W
)
11IT
ED
--"A
bilit
y to
inte
rpre
tlit
erar
ym
ater
ials
"
X45
X45
Oct
.O
ct.
Sta
ndar
dS
core
sN
atio
nal
°Ail.
Sco
res
Beg
inni
ng o
fye
ar m
rms
Not
es:
aM
idpo
ints
of p
erce
ntiie
inte
rval
s w
ere
used
.b
Sin
ce d
iffer
ent m
onth
s w
ere
invo
lved
, diff
eren
t nor
m ta
b!es
had
to b
eus
ed.
cFor
ms
KM
and
JM
are
195
3 ed
ition
test
s an
d F
orm
W a
196
4 ed
ition
: equ
ival
ence
bet
wee
n th
e tw
o ed
ition
s re
quire
d us
e of
spe
cial
tabl
es.
.4
fillemal1010114.01.1104.mve.--......
180 67TMP-115
APPENDIX E
CORRECTIONS FOR DIFFERENCES IN PRE-POSTTCCTI Mr n ATCCI I.4 I Isv ...7 v.-% I a...,
In a ntber of instances, achievement tests were administeredat somewhat different times in the pre and post years. These differ-ences generate systematic errors in observations of change in achieve-ment. Achievement test publishers provide norms for only a fewdiP'erent times of year; for example, the Stanford Achievement Test(SAT) has beginning-of-year norms (for September through December),middle-of-year norms (January through April), and end-of-year norms(May and June). These norms are usually obtained by interpolatingbetween the results from pupils in different grades, rather than act-ually determining separate norms for various times of year. No dis-tinction is made between tests administered at different times withinthe interval covered by the norming period. This is consistent withthe recognized crudeness of achievement test scores for individuals.Indeed, publishers carefully point out that the scores for.individualsare quite imprecise. However, for group comparisons which accum-ulate errors from individual scores, more precision is needed. Asan example, if pre-year testing is done in January and post-yeartesting in April, the same norms (middle-of-year) would be used toscore both sets of results. However, the post year results wouldthen contain a positive bias and the pre-year results a negative bias.To determine the extent of errors introduced by differences in timeof year, all cases where there were three or more weeks discrepancybetween pre and post testing dates were examined. These instancesand the factors required for correction are shown in Table 54. Theamount of correction needed was obtained by applying the appropriate"change per month" (Shown in the last column of Tables 55 through 58and calculated by interpola;ing in the appropriate norm tables) to thediscrepancy in test dates 'shown in the "Time Difference" columnsin Table 54). For example, the first entry in Table 54 shows a cor-rection factor of +0. 5. Notice that the SAT In. I was given at the endof the year to this fourth grad,!. The amount of change per monthwas calculated for both the fortieth and the twentieth percentile levels.These two points in the distribution were used to represent the rel-atively low scores in our sample and to diminish the effect of rounding
APPENDIX E 181
errors. Table 56 shows the appropriate "change per month" as 0. 5at the fortieth percentile and 0.8 at the twentieth percentile; aver-aging these two numbers gives 0. 65. From Table 54, the discrepancyin testing dates was three weeks or, approxin.ately, 0.75 month.Therefor..., the c^rr...ctirsn recrir.,1 is to. 66) x (0. 751 . n. c =c chewmin the last column of the first line in Table 54.
Tables 57 arid 58 are similar to Tables 55 and 56 but apply to theMAT Reading Test.
Table 59 shows average changes in achievement level before andafter adjustment for differences in the pre-post testing dates for thefour districts with differences of three or more weeks in test dates.
The adjustment produced noticeable change in the values for Dis-tricts 1 and 12, where the proportion of observations adjusted washigh. In District 1, all cases were adjusted; in, District 12, 25 per-cent of the cases were adjusted. The adjusting of scores is a likelysource of error. As is shown in Table 59, adjustments for the 50observations affected make the change more negative. However, thisdecrease is fairly small compared to the standard errors of thestatistics.
,
Tab
le 5
4. B
ases
for
adju
stm
ent f
acto
rs r
equi
red
by v
aria
tions
in d
ates
bet
wee
n ad
min
istr
atio
nof
pre
and
pos
t tes
ts.
Dis
tric
tG
rade
Tes
t (S
ubte
st)
Dat
e of
Tes
tin
Aca
d. Y
r.65
-66
66-6
7T
ime
Diff
eren
cea
Sco
reb
Diff
eren
ceA
djus
tmen
tF
acto
rc
14
SA
T (
PM
) In
t ILa
teA
pr. 2
43
wee
ks-0
.5+
0.5
May
May
5
15
SA
T (
PM
) In
t I, I
ILa
teM
arch
3 w
eeks
+0.
4-0
.4M
ay6-
17
16
SA
T (
PM
) In
t II
Feb
.Ja
n.1
mon
th-0
.8+
0.8
17
SA
T (
PM
) A
dv.
Late
Apr
. 24
3 w
eeks
-0.2
+0.
2M
ayM
ay 5
18
SA
T (
PM
) A
dv.
Late
Dec
.1-
1/2
mon
ths
+0.
8-0
.8M
ay5-
17
811
MA
T (
Rdg
) H
.S.
Jan
7M
ay 1
54
mon
rhsd
+0.
9-0
.9
12e
3M
AT
(R
dg)
Sep
t.O
ct. 5
3 w
eeks
+1.
6-1
.6P
ri II
15
136
SA
T (
PM
) In
t I, I
IJa
n.O
ct.
1 m
onth
+0.
4-0
.4
Not
es:
.
aD
iffer
ence
is in
ref
eren
ce to
mid
dle
of n
orm
per
iod.
bA
mou
nt o
f Sta
ndar
d T
-sco
re d
iffer
ence
favo
ring
post
test
.c C
orre
ctio
n to
be
appl
ied
to o
bser
vatio
ns (
post
min
us p
re).
dT
hese
dat
a w
ere
base
d on
a s
ingl
e no
rm fo
r th
e 2
test
ings
, rat
her
than
diff
eren
t tim
e of
yea
r no
rms.
eO
nly
one
scho
ol (
Sch
ool 2
) ha
d a
3-w
eek
diffe
renc
e.
Tab
le 5
5. C
hang
e fr
om b
egin
ning
of y
ear
tom
iddl
e of
yea
r no
rms:
SA
T p
arag
raph
mea
ning
.
Gra
deB
atte
ry
Per
cent
i le
Sco
re L
evel
Beg
inni
ngof
Yea
r
Equ
ival
ent
Beg
inni
ng-
of-Y
ear
Gra
de S
core
Equ
ival
ent
Mid
Yea
rP
erce
ntile
Sco
reP
erce
ntile
Sco
reb
Cha
nge
Per
Mon
thS
tand
ard
T S
core
Cha
nge
Per
Mon
th
5In
t I40
4728
30.
8
2039
122
0.8
6In
t 11
4056
292.
80.
8
2047
141.
50.
6
8A
dv.
4075
322
0.5
2062
161
0.4
Not
es:
aAll
are
1964
edi
tion.
bB
ased
on
the
4-m
onth
per
iod
betw
een
mid
-poi
nts
of th
e tw
o no
rm p
erio
ds.
Tab
le 5
6. C
hang
e fr
omm
iddl
e of
yea
r to
end
of y
ear
norm
s:S
AT
par
agra
ph m
eani
ng.
Gra
deB
atte
ry
Per
cent
ileS
core
Lev
elB
egin
ning
of Y
ear
Equ
ival
ent
Mid
-Yea
rG
rade
Sco
re
Equ
ival
ent
End
of Y
ear
Per
cent
ile S
core
Per
cent
ile S
core
Cha
nge
Per
Mon
thb
Sta
ndar
d T
Sco
reC
hang
e F
er M
onth
4In
t I40
4234
20.
5
2034
142
0.8
5In
t II
4052
361.
30.
4
2042
151.
70.
7
6In
t II
4062
380.
70.
2
2050
180.
70.
31
7A
dv.
4070
380.
70.
2
2058
180.
70.
3
8A
dv.
4079
380.
70.
2
2065
180.
70.
3
,
Nor
es:
aA
ll ar
e 19
64 e
ditio
n si
nce
the
6th
grad
e sc
ores
from
the
1953
edi
tion
wer
eco
nver
ted
into
196
4 ed
it4,n
equi
vale
nts.
bB
ased
on
the
3-m
onth
per
iod
betw
een
mid
poin
ts o
f the
two
norm
perio
ds.
t.
Tab
le 5
7. C
hang
e fr
om b
egin
ning
of
year
nor
ms
to m
iddl
e of
yea
r: M
AT
Rea
ding
a.
Gra
deB
atte
ry°
Per
cent
i le
Sco
re L
evel
Beg
inni
ng o
fY
ear
Equ
ival
ent
Equ
ival
ent
Beg
inni
ng-
of-Y
ear
"Sta
ndar
d" S
core
Mid
Yea
rP
erce
nti l
eS
core
Per
cent
ile S
core
Cha
nge
Per
Mon
thb
Sta
ndar
d T
-Sco
reC
hang
e P
er M
onth
31
Prim
ary
II40 20
34 29
13.3
4.4
Not
es:
aThe
calc
ulat
ib
Bas
ed o
n th
e
7.1
4.2
1.8
1.5
ons
for
the
Prim
ary
II ba
ttery
wer
e ba
sed
upon
the
seco
nd g
rade
leve
l.
3-3/
4 m
onth
per
iod
betw
een
mid
poin
ts o
f the
norm
inte
rval
s.
Tab
le 5
8. C
hang
e fr
om e
leve
nth
grad
eto
twel
fth g
rade
nor
ms:
MA
T R
eadi
ng°.
.011
1110
1111
110=
1111
11
Gra
deB
atte
ry
Per
cent
ileS
core
Lev
elM
id-Y
ear
11th
Gra
deE
quiv
alen
t"S
tinda
rd"
Sco
re
I Equ
ival
ent
%ile
Sco
reM
id-Y
ear
12th
Gra
deP
erce
ntile
Sco
reC
hang
e P
er M
onth
Sta
ndar
d T
-Sco
reC
hang
e P
er M
onth
11H
igh
Sch
ool
4052
310.
90.
2620
42.
150.
50.
20--
--...
......
.,N
ote:
aB
ased
on
"age
-con
trol
led
sam
ples
" te
sted
ann
ually
.T
he y
ear
inte
rval
bet
wee
n te
sts
is c
onsi
dere
dto
con
sist
of
ten
acad
emic
mon
ths.
18667TMP-115
Table 59. Adjusted average changes in districts where pre and
post test dates differed by three or more weeks.
I
1
I.
1
DISTRICT
8 1 '12 13I
Number of Cases 39 55 4 32
Adjusted Cases 39 3 1 7
[Average Change in Mean
Unadjusted 0.27 -0.33 0.83 1.25
Adjusted 0,24 -0.38 0.43 1.16
Standard Error (unadj. ) 0.30 0.27 1.07 0.48
Significance Level (unadj.) 0.35 0.22 0.45 0.01
Average Change in Lowest Decile
Unadjusted 0.08 0.31 2.18 1.40
Adjusted 0.05 0.26 1.78 1.31
Standard Error (unadj. ) 0.28 0.38 2.42 0.71
Significance Level (unadj.) 0.50 0.42 0.42 0.05
187
APPENDIX F
USE OF STANDARD T-SCORES
The Standard T-score was selected for the following reasons: theunits of standard T-scores are nearly equal throughout the range ofachievement levels for a given grade level; results from differenttests can be converted to T-Scores via tables provided by the Testpublisher; test scores from different grades can be converted to thesame T-score levels; test scores obtained at different times of theyear can be converted to the same T-score units. None of the othermeasures has all of these qualities.
It is realized that test scores when converted to the Standard T-score have the above qualities only to a degree. The T-score isobtained by converting the raw score for a specific test to a nationalpercentile rank and then converting the percentile rank to the Stand-ard T-score. For example, a raw score of 11 on the test illustratedin Table 60 is equivalent to a "national" percentile rank of 8. Asshown in Table 61 a national percentile of 8 is equivalent to a stand-ard T-score of 36.0. One source of incompatibility among resultsfrom different tests is the fact that publishers use different popula-tions for developing the conversion table to be used for transformingraw score to "national" percentile. That is, it is really a conversionto a percentile rank for the test population and not the entire nation.
It has been assumed that none of these factors has a serious effecton conclusions in this study, because of the following reasons: (1)there is no attempt to make comparisons of absolute scores since thedifferences from Pre to Post are the important observations; (2) thesame rules were used in the conversion for both Pre and Post years,e.g., interpolation or extrapolation between values in conversion tables.
Tables 60 and 61 provide the reader with examples of the relation-ship between national percentile, Standard T-score, and grade equiva-lence. Any of the statistics on Standard T-scores provided in the sum-mary tables of this report can be interpreted in terms of nationalpercentiles by using Table 61. Table 60 provides a conversion fromnational percentiles to grade equivalence but for only the sixth gradein the latter part of the academic year.
V i
.................
188 67TMP-115
Table 60. Conversion of raw score on Advanced Battery, Form VI, of SATto grade score and national percentile (for tests given in sixthgrades during May & June of academic year) .
No.Right
1
2
3
4
567e9
10
11
12
13
14
15
16
17
18
19
2021
22
232425
2627282930
GradeScoreb
Below 20II
21
232527293235384042'444648505253545658606263646566687072
National i No. r... y1
1/4.71U(s
Percenti lec Right I Score'
5dow 1It
II
II
II
11
II
2
4
68
1011
12
14
16
2021
222428323436384244485254
31 74
32 7633 7834 8035 82..
36 8437 86
38 9039 9440 9941 102
42 104
43 105
44 106
45 10746 108
47 11048 112
49 113
50 114
51 116
52 118
53 12054 121
55 123
56 124
57 126
58 12759 129
60 129+
,..at;onal.......1Percentilec
5862687074767881
838589909"/:
9293949596969798989999999999999999
Notes:
°All data are from publisher's tables accompanying test booklets.
bGrade score is 10 times grade equivalent score. For example, a grade equiva-
lent of 6.0 is a grade score of 60.
cThis national percentile rank is valid only for tests given to the 6th grade inMay or June of the academic year.
APPENDIX F
Table 61. Convernng percentiles to Standard T scores.°
i 89
%b Tc %b Tc %b c
T %b Tc
1 - 26.7 26 - 43.6 51 - 50.3 76 - 57.1n nn r n-f Art et.4 - L7.J L/ - 40.7 52 - 50.5 77 - 5-7.43 - 31.2 28 - 44.2 53 - 50.7 78 - 57.74 - 32.5 29 - 44.5 54 - 51.0 79 - 58.15 - 33.5 30 - 44.7 55 - 51.3 80 - 58.46 - 34.5 31 - 45.0 56 - 51.5 81 - 58.87 - 35.2 32 - 45.4 57 - 51.8 82 - 59.28 - 36.0 33 - 45.6 58 - 52.0 83 - 59.59 - 36.6 34 - 45.9 59 - 52.3 84 - 59.9
10 - 37.2 35 - 46.2 60 - 52.5 85 - 60.411 - 37.7 3.5 - 46.4 61 - 52.8 86 - 60812 - 38.3 37 - 46.7 62 - 53.0 87 - 61.313 - 38.7 38 - 47.0 63 - 53.3 88 - 61.714 - 39.2 39 - 47.2 64 - 53.6 89 - 62.315 - 39.6 40 - 47.5 65 - 53.8 90 - 62.816 - 40.1 41 - 47.7 66 - 54.1 91 - 63.417 - 40.5 42 - 48.0 67 - 54.4 92 - 64.018 - 40.8 43 - 48.3 68 - 54.7 93 -; 64.819 41.2 44 - 48.5 69 - 55.0 94 - 65.620 - 41.6 45 - 48.7 70 - 55.3 95 - 66.521 - 41.9 46 - 49.0 71 - 55.5 96 - 67.522 - 42.3 47 - 49.3 72 - 55.8 97 - 68.823 - 42.6 48 - 49.5 73 - 56.1 90 - 70.324 - 42.9 49 - 49.7 74 - 56.4 99 - 70.525 - 43.3 50 - 50.0 75 - 56.7
Notes:a
Reference 8, p 66.bNet;onal
percentile.
ICEquivalent Standard T-Score.
19067TMP-115
APPENDIX G
DETAILED INFORMATION FOR DISTRICT 10
Tables 62 through 70 provide expenditure information for each of
the 10 schools in District 10. Discussion of these tables and the cor-responding Table for School 1 is in Section 3.
Tables 71 through 79 provide expenditure data by grade and type ofCE program. Discussion of these tables and the corresponding Tablefor School 1 is in Section 3.
Tab
le 6
2.E
xpen
ditu
res
for
regu
lar
and
com
pens
ator
y ed
ucat
ion
prog
ram
s: S
choo
l 2.
Dis
tric
t 10
Sch
ool 0
2R
egul
ar S
choo
l Pro
gram
sC
ompe
nsat
ory
Edu
catio
n P
rogr
ams
Exp
endi
ture
Cat
egor
ies
Tea
cher
Aid
esA
djus
tmen
t Tea
chin
gR
educ
ed C
lass
Siz
eC
entr
aliz
edLi
brar
y
64-6
565
-66
66-6
765
-66
66-6
765
-65
6-67
.
fil$
#$
#$
#$
ri$
#$
#$
Prin
cipa
ls, C
onsu
ltcnt
s, S
upeM
sors
186
891
9362
112
i65
20
Tea
cher
s15
8209
115
8438
915
1077
745
3196
347
3617
4469
10
Libr
ary,
Aud
io-v
isua
l & T
V P
erso
nnel
48
Sec
retw
ial-C
leric
al1
3024
133
601
4070
Tot
al S
alar
ies
9380
497
111
1240
0932
6447
3617
4469
10
Tea
chin
g S
uppl
ies
& O
ther
Exp
ense
s10
2411
4210
6296
714
3827
96
Tot
al In
stru
ctio
nal E
xper
res
9482
898
253
1250
7142
3147
3631
8297
06
Hea
lth S
ervi
ces
Pup
il T
rans
port
atio
n
Pla
nt O
pera
tion
& M
aint
enan
ce17
771
1513
517
490
5335
3
Fix
ed C
harg
es12
620
610
948
4
Foo
d S
ervi
ces
S?u
dent
/Com
mun
ity A
ctiv
ities
Tot
al S
uppo
rt E
xpen
ses
1777
115
135
1749
012
625
910
983
7
Tea
chin
g E
quE
pmen
t33
411
221
4241
217
Tot
al E
xpen
ses
1125
9911
3388
1428
9554
79I4
995
7532
1076
0
Tab
le 6
;.'E
xpen
ditu
res
for
regu
lar
and
com
pens
ator
y,,.
-.1u
catio
n pr
ogra
ms:
Sch
ool 3
.
Dis
tric
t 10
Sch
ool 0
3R
egul
ar S
choo
l Pro
gram
sC
ompe
nsat
ory
Edu
catio
n P
rogr
ams
Exp
endi
ture
Cat
egor
ies
.
Tea
cher
Adj
ustm
ent
Red
uced
Aid
esT
each
ing
Cla
ss S
ize
IC
entr
aliz
ed L
ibra
ry
64-6
565
-66
66-6
765
-66
66-6
765
-66
66-6
7
#$
#$
#$
#$
#$
#$
#$
Prin
cipa
ls, C
ons.
, !ta
nts,
Sup
ervi
sors
110
214
193
911
1343
36
Tea
cher
s42
2358
3942
2502
2444
3250
1913
9579
1320
306
1744
6810
Libr
ary,
Aud
io-v
isua
l & T
V P
erso
nnel
150
Sec
reta
rial-C
leric
al2
8098
282
952
9200
Tot
al S
alar
ies
2541
5126
7910
3476
5297
3520
306
1744
6810
Tea
chin
g S
uppl
ies
& O
ther
Exp
ense
s29
8829
8927
2623
9115
4428
20
Tot
al In
stru
ctio
nal E
xpen
ses
2571
39(
2708
9935
0378
1212
620
306
3288
9630
Hea
lth S
ervi
ces
Pup
il T
rans
port
atio
nP
lant
Ope
ratio
n &
t4 a
inte
nanc
e34
812
3349
141
559
353
615
Fix
ed C
harg
es10
947
7
Foo
d S
ervi
ces
Stu
dent
/Com
mun
ity A
ctiv
ities
Tot
al S
uppo
rt E
xpen
ses
3481
233
491
4155
935
310
910
92
Tea
chin
g E
quip
men
t75
819
4645
76
Tot
al E
xpen
ses
2919
5130
4390
3926
9514
464
2154
479
7310
722
Tab
le 6
4. E
xpen
ditu
res
for
regu
lar
and
com
pens
ator
y ed
ucat
ion
prog
ram
s:S
choo
l 4.
Dis
tric
t 10
Sch
ool 0
4R
egul
ar S
choo
l Pro
gram
sC
ompe
nsat
ory
Edu
catio
n P
rogr
ams
.T
each
er A
ides
Exp
endi
ture
Cat
egor
ies
Adj
ustm
ent T
e-ic
hing
Red
uced
Cla
ss S
ize
64-6
565
-66
66-6
765
-66
66-6
765
-66
66-6
7
$I/
$#
$I/
$al
$$
Prin
cipa
ls, C
onsu
ltant
s, S
uper
viso
rs1
7724
183
841
1117
430
Tea
cher
s7
4003
17
4251
88
5562
13
1578
346
58
Libr
ary,
Aud
io-v
isua
l & i'
V P
erso
nnel
83
Sec
reta
rial-C
leric
al1
3536
136
231
140
7016
9146
58
Tot
al S
alar
ies
5129
154
5251
7086
516
9146
58
Tea
chin
g S
uppl
ies
& O
ther
Exp
ense
s49
745
147
359
2
Tot
al in
stru
ctio
nal E
xpen
ses
5178
854
976
71^3
822
8346
58
Hea
lth S
ervi
ces
Pup
il T
rans
port
atio
n
Pla
nt O
pera
tion
& M
aint
enan
ce91
5012
255
1024
4
Fix
ed C
harg
es67
203
Foo
d S
ervi
ces
Stu
dent
/Com
mun
ity A
ctiv
ities
Tot
al S
uppo
rt E
xpen
ses
9150
1225
510
244
6720
3
Tea
chin
g E
quip
men
t14
456
Tot
al E
xpen
ses
6093
867
231
8159
628
0648
61
1"T
Tab
le 6
5. E
xpen
ditu
res
for
regu
lar
and
com
pens
ator
yed
ucat
ion
prog
ram
s: S
choo
l 5.
Dis
tric
t 10
Sch
ool 0
5R
egul
ar S
choo
l Pro
gram
sC
ompe
nsat
ory
Edu
catio
n P
rogr
ams
Exp
endi
ture
Cat
egor
ies
Tea
chin
g A
ides
Adj
ustm
ent T
each
ing
Red
uced
Cla
ss S
ize
Inte
nsiv
eIn
stru
ctio
nal
Impr
ovem
ent
64-6
565
-66
66-6
765
-66
66-6
765
-66
66-6
7
$#
$#
$#
I$
$#
$[
$
Prin
cipa
ls, C
onsu
ltant
s, S
uper
viso
rs1
1021
41
1074
31
1343
3j
2030
6
Tea
cher
s42
2338
1942
; 255
857
4232
6639
1077
5017
2712
446
75
Libr
ary,
Aud
io-v
isua
l&
TV
Per
sonn
el1
121
815
Sec
reta
rial-C
leric
al2
7431
278
642
9090
Tot
al S
alar
ies
2514
6434
9162
7891
2712
457
96
Tea
chin
g S
uppl
ies
& O
ther
Exp
ense
s31
5528
2525
2415
9569
922
Tot
al In
stru
ctio
nal E
xpen
ses
2546
1927
7289
3516
8694
8627
823
5818
Hea
lth S
ervi
ces
Pup
il T
rans
port
atio
nP
lant
Ope
ratio
n &
Mai
nten
cInc
e38
017
3299
240
684
Fix
ed C
harg
es30
211
8237
4
Foo
d S
ervi
ces
Stu
dent
/Com
mun
ity A
ctiv
ities
Tot
al S
uppo
rt E
xpen
ses
3801
732
992
4068
430
211
8237
4
Tea
chin
g E
quip
men
t10
050
119
0113
8929
7
Tot
al E
xpen
ses
2926
3631
0381
3928
7111
690
3039
464
89
Tab
le 6
6. E
xpen
ditu
res
for
regu
lar
and
com
pens
ator
y ed
ucat
ion
prog
ram
s: S
choo
l 6.
Dis
tric
t 10
Sch
ool 0
6R
egul
ar S
choo
l Pro
gram
sC
ompe
nsat
ory
Edu
catio
n P
rogr
ams
Exp
endi
ture
Cat
egor
ies
beac
her
Aid
esA
djus
tmen
t Tea
chin
gR
educ
ed C
lass
Siz
eP
roje
ctQ
ualit
y
64-6
565
-66
66-6
765
-66
6-67
65-6
666
-67
#$
#$
#$
#$
#$
#$
#...
...1
$
Prin
cipa
ls, C
onsu
ltant
s, S
uper
viso
rs1
7120
183
911
1117
430
Tea
cher
s9
4761
49
4615
19
6244
44
2556
464
91
Libr
ary,
Aud
io-v
isua
l & T
V P
erso
nnel
47
Sec
reta
rial-C
leric
al1
3280
136
221
4070
Tot
al S
alar
ies
5801
458
164
7768
826
3364
91
Tea
chin
g S
uppl
ies
& O
ther
Exp
ense
s41
510
2764
370
616
740
0
Tot
al In
stru
ctio
nal E
xpen
ses
5842
959
191
7833
133
3966
5840
0
Hea
lth S
ervi
ces
Pup
il T
rans
port
atio
n
Pla
nt O
pera
tion
& M
aint
enan
ce99
2692
2914
357
13
Fix
ed C
harg
es10
333
3
Foo
d S
ervi
ces
Stu
dent
/Com
mun
ity A
ctiv
ities
Tot
al S
uppo
rt E
xpen
ses
9926
9229
1435
710
334
6
Tea
chin
g E
quip
men
t39
669
229
836
4
Tot
al E
xpen
seq
6835
668
420
9308
441
3473
0276
4_.
Tab
le 6
7.E
xpen
ditu
res
for
regu
lar
and
com
pens
ator
yed
ucat
ion
prog
ram
s: S
choo
l 7.
Dis
tric
t 10
Sch
ool 0
7R
egul
ar S
choo
l Pro
gram
sC
ompe
nsat
ory
Edu
catio
n P
rogr
ams
Exp
endi
ture
Cat
egor
ies
Tea
cher
Aid
esA
djus
tmen
t Tea
chin
gR
educ
ed C
lass
Siz
e
Eng
lish
asS
econ
d La
ngua
ge
64-6
565
-66
66-6
765
-66
66-6
765
-66
66-6
7
$#
$#
$#
$#
$#
$I
$
Prin
cipa
ls, C
onsu
ltant
s, S
uper
viso
rs1
8954
193
061
1223
730
4196
7160
Tea
cher
s10
5412
211
6471
411
8609
23
2091
224
9826
2547
80
Libr
ary,
Aud
io-v
isua
l & T
V P
erso
nnel
2434
8
Sec
reta
rial-C
leric
al1
4049
136
231
2539
906
375
Tot
al S
alar
ies
6717
577
643
1008
6821
4524
9880
7512
315
Tea
chin
g S
uppl
ies
& O
ther
Exp
ense
s51
651
064
539
714
2220
3
Tot
al In
stru
ctio
nal E
xpen
ses
6769
178
153
1015
1325
4239
2082
7812
315
Hea
lth S
ervi
ces
Pup
il T
rans
port
atio
n70
5
Pla
nt O
pera
tion
& M
aint
enan
ce17
021
1255
714
879
45
Fix
ed C
harg
es89
109
139
964
Foo
d S
ervi
ces
290
Stu
dent
/Com
mun
ity A
ctiv
ities
Tot
al S
uppo
rt E
xpen
ses
1702
112
557
1487
989
109
544
969
Tea
chin
g E
quip
men
t10
665
325
285
951
8
Tot
al E
xpen
ses
8471
390
710
1164
9832
8442
8196
8113
802
Tab
le 6
8. E
xpen
ditu
res
for
regu
lar
and
com
pens
ator
y ed
ucat
ion
prog
ram
s: S
choo
l 8.
Dis
tric
t 10
Sch
ool 0
8R
egul
ar S
choo
l Pro
gram
sC
ompe
nsat
ory
Edu
catio
n P
rogr
ams
Exp
endi
ture
Cat
egor
ies
Tea
cher
Aid
esA
djus
tmen
t Tea
chin
gR
educ
ed C
lass
Siz
e
Inte
nsiv
eIn
stru
ctio
nal
Impr
ovem
ent
64-6
565
-66
66-6
765
-66
66-6
765
-66
66-6
7
#$
#$
#$
#$
#$
##
$
Prin
cipa
ls, C
onsu
ltant
s, S
uper
viso
rs1
8996
184
401
1132
720
785
Tea
cher
s20
9775
721
1081
5222
1462
855
3858
914
021
2565
Libr
-y, A
udio
-vis
ual &
TV
Per
sonn
el63
Sec
reta
rial-C
leric
al1
4049
129
681
4600
Tot
al S
alar
ies
1108
0211
9561
1622
1239
4114
021
3350
Tea
chin
g S
uppl
ies
& O
ther
Exp
ense
s14
9512
5314
8273
839
8
Tot
al In
stru
ctio
nal E
xpen
ses
1122
9712
0814
1636
9446
7914
419
3350
Hea
lth S
ervi
ces
Pup
il T
rans
port
atio
n
Pla
nt O
pera
tion
& M
aint
enan
ce49
683
1967
822
766
Fix
ed C
harg
es17
661
120
3
P
Foo
d S
ervi
ces
1
Stu
dent
/Com
mun
ity A
ctiv
ities
Tot
al S
uppo
rt E
xpen
ses
4968
319
678
2276
617
661
120
3
Tea
chin
g E
quip
men
t34
110
1079
8
Tot
al E
xpen
ses
1619
8114
0492
1868
0158
6515
828
3553
....-
..-
Iia
Tab
le 6
9.E
xpen
ditu
res
for
regu
lar
and
com
pens
ator
y ed
ucat
ion
prog
ram
s: S
choo
l 9.
Dis
tric
t 10
Sch
ool 0
9R
egul
ar S
choo
l Pro
gram
sC
ompe
nsat
ory
Edu
catio
n P
rogr
ams
Exp
endi
ture
Cat
egor
ies
Tea
cher
Aid
esA
djus
tmen
t Tea
chin
gR
educ
ed C
lass
Siz
e
Inte
nsiv
eIn
stru
ctio
nal
Impr
ovem
ent
64-6
565
-66
66-6
765
-66
66-6
765
-66
66-6
7
#$
#$
#$
#$
#$
$$
Prin
cipa
ls, C
onsu
ltant
s, S
uper
viso
rs1
8996
193
701
1193
720
312
Tea
cher
s16
1027
0416
1052
8218
1489
314
3230
578
5315
65
Libr
ary,
Aud
io-v
isua
l & T
V P
erso
nnel
51
Sec
reta
rial-C
laric
al1
4049
141
481
4600
Tot
al S
alar
ies
1157
4911
8800
1654
6833
0178
5318
77
Tea
chin
g S
uppl
ies
& O
ther
Exp
ense
s12
3410
6799
868
719
9
Tot
al In
stru
ctio
nal E
xpen
ses
1169
8311
9867
1664
6639
8880
5218
77
Hea
lth S
ervi
ces
Pup
il T
rans
port
atio
n
Pla
nt O
pera
tion
& M
aint
enan
ce20
552
1543
817
380
Fix
ed C
harg
es13
634
312
0
Foo
d S
ervi
ces
Stu
dent
/Com
mun
ity A
ctiv
ities
To$
al S
uppo
rt E
xpen
ses
2055
215
438
1738
013
634
312
0
Tea
chin
g E
quiP
men
t10
707
397
Tot
al E
xpen
ses
1375
3513
5305
1838
5648
3187
9219
97
Tab
le 7
0.E
xpen
ditu
res
for
regu
lar
and
com
pens
ator
y ed
ucat
ion
prog
ram
s: S
choo
l 10.
Dis
tric
t 10
Sch
ool 1
0R
egul
ar S
choo
l Vro
gram
sC
ompe
nsat
ory
Edu
catio
n P
rogr
ams
Exp
endi
ture
Cat
egor
ies
Tea
cher
Aid
esA
djus
tmen
t Tea
chin
gR
educ
ed C
lass
Siz
e
64-6
565
-66
66-6
765
-66
66-6
765
-66
66-6
7
#$
#$
#$
#$
igs
#$
$
Prin
cipa
ls, C
onsu
ltant
s, S
uper
viso
rs1
8996
196
701
1103
720
Tea
cher
s14
7360
614
7976
314
1048
104
3258
346
61
Libr
ary,
Aud
io-v
isua
l & T
V P
erso
mte
l39
Sec
reta
rial-C
leric
al1
3757
141
481
5069
Tot
al S
alar
ies
8735
993
581
1209
1639
3746
61
Tea
chin
g S
uppl
ies
& O
ther
Exp
ense
s93
696
211
0962
0
Tot
al In
stru
ctio
nal E
xpen
ses
8829
594
543
1220
2539
3746
61
Hea
lth S
ervi
ces
Pup
il T
rans
port
atio
n
Pla
nt O
pera
tion
& M
aint
enan
ce86
4988
8712
089
Fix
ed C
harg
es13
820
3
Foo
d S
ervi
ces
Stu
dent
/Com
mun
ity A
ctiv
ities
Tot
al S
uppo
rt E
xpen
ses
8649
8887
1208
913
820
3
Tea
chin
g E
quip
men
t35
977
9
Tot
al E
xpen
ses
9594
410
3430
1344
7348
5448
64.
.
200
Table 71. Program expenditures by grades: School 2.
67TMP-115
1965-66RegularPrograms
Teacher Aides,AdjustmentTeachers,
Reduced ClassSize
CentralLibrary
ClinicalReading
I Total CEPrograms
Grade ADM $ $ $ $
K 56 1025 8 1112 681 1793
1 144 26377 2858 1752 461 0
2 V 76 13922 15 08 925 2433
3 68 12456 827 2604 3431
4 65 11907 791 2604 3395
5 80 14654 973 2604 3577
6 58 10624 706 2604 331 0
Sp. 72 13189 876 876
Totals 619 109781 5478 7 5": 10416 23425
I 1966-67RegularPrognams
$
Teacher AidesAdjustmentTeachers,
Reduced ClassSize
$
.
CentralLibrary
$
ClinicalReading
Total CEPrograms
$Grade ADM
K 64 16936 1137 1275 2412
1 110 29108 1955 2192 4147
2 1 07 28313 1902 2132 4034
3 60 15877 1195 5048 7243
4 62 16406 1235 5048 6283
5 63 16670 1256 5048 6304
6 59 15613 1176 5048 6224
Sp. 15 3970 299 299
Totals 540 1428 93 4994 1 0760 20192 36946
APPENDIX G
Table 72. Program expenditures by grades: School 3.
201
1965-66RegularPrograms
Teacher Aides,Adjustment
Teachers'Reduced Class
Size
CentralLibrary
Total CEPrograms
Grade ADM $ $ $ $
K 116 21825 2534 572 3106
1 3 01 56617 6574 1483 8057
2 245 46085 5352 1207 6559
3 210 39510 1035 1035
4 198 37257 976 976
5 1 69 31778 832 832
6 163 30652 803 803
Sp. 216 40636 1064 1064
Totals 1618 3 04360 14460 7972 22432
1966-67Regular
Teacher Aides,AdjustmentTeachers'
Reduced ClassSize
CentralLibrary
Total CEProgramsProg rams
Grade ADM $ $ $ $
K "42 39623 4990 1082 6072
1 227 63342 7976 1729 9705
2 244 68093 8575 1859 10434
3 177 49401 1349 1349
4 192 53603 1464. 1464
5 146 40762 1113 1113
6 137 38248 1044 1044
Sp. 142 39629 1082 1082
I Totals 1407 392701 21541 1 0722 32263
202
Table 73. Program expenditures by grades: School 4.
67TMP-115
1965-661 Teacher Aides
AdjustmentRegular Teachers,Programs I
!Reduced Class
Total CEPrograms
l
1
I Size
Grade ADM $ $ $
K 40 9638 857 857
1, 48 11566 1028 1 028
2 43 1 0362 92i 921
3 29 6988
4 21 5060
5 21 5060
6 28 6747
Sp. 49 11808
Totals 279 67229 2806 2806
1966-67RegularPrograms
Teacher Aides,AdjustmentTeachers,
Reduced ClassSize
Total CEPrograms
Grade ADM $ $ I$
K 43 14206 1620 1620
1 41 13545 1545 1545
2 45 I 14867 1696 1696
3 35 11562
4 26 8592
5 28 9253
6 19 6275
Sp. 10 3305
I Totals 247 81605 4861 4861
APPENDIX G
Table 74. Program expenditures by grades: School 5.
203
1965-66RegularP rog ra rns
Teacher Aides,AdjustmentTeachers,
Reduced ClassSize
Total CEProarams
Grade ADM $ $ $
K 135 31659 2 716 2716
1 228 53448 4587 45 87
2 218 5112 0 4386 4386
3 2 01 47116
4 1 96 45936
5 193 45254
6 139 32590
Sp. 14 3290
Totals 1324 310413. 11 689 11689'
1966-67RegularPrograms
Teacher Aides'AdjustmentTeachers'
Reduced ClassSize
IntensiveIns.ructionalImprovement
Total CEPrograms
,
Grade ADM $ $ $ $
163 41841 7285 7285
1 279 71660 12469 12469
2 238 61131 1 0637 10637
3 224 57506 1700 1700
4 214 54963 1634 1634
5 198 5 0838 1511 151 1
6 198 5 0838 I 1511 151 1
Sp. 16 4125 I122 122
Totals 1530 3 92902 303 91 6488 36879 I
204
Table 75. Program expenditures by grades: School 6.
67TMP-115
1965-66RegularPrograms
leacher Aides,Adjustment
Teachers,Reduced Class
Size
Language ArtsTeacher
ConsultantsTotal CEPrograms
Grade ADM $ $ $
K 33 8183 798 798
1 80 19835 1934 2140 4074
2 58 14375 1402 1552 2954
3 36 9822 963 963
4 38 9421 1016 1016
5 22 5453 588 588
6 5 1238 134 134
Sp. 4 992
Totals 276 68419 4134 6393 10527
1966-67Regular
Programs
Teacher Aides,AdjustmentTeachers,
Reduced ClasseJI ze
ProjectQuality
Language ArtsTotal CEPrograms
IGrade ADM $ $ $ $
K 47 12464 1778 102 1880
1 81 21481 3065 176 4242 7483
2 65 17239 2459 1.41 3405 6005
,.i 65 17239 141 3405 3546
4 48 12734 105 2514 2619
5 0
6 0
Sp. 45 11933 98 98
Totals 351 93090 7302 763 13566 21631
APPENDIX G
Table 76. Program expenditures by grades: School 7.
205
1965-66RegularPrograms
Teacher Aides,Adjustment
i Teachers'Reduced Class
Size
English as aSecond
LanguageTotal CEPrograms
Grade ADM $ $ $ $
K 33 9153 803 977 1780
1 56 15539 1362 1658 3020
2 46 12763 1119 1362 2481
3 59 16364 1746 1746
4 27 7493 800 800
5 29 8046 859 859
6 23 6377 681 681
Sp. 54 14976 1598 1598
Totals 1 327 90711 3284 9681 112965
1966-67Regular
Pr°grams
Teacher Aides'Adjustment
Teachers,Reduced Class
Size
English as aSecond
LanguageProjectQuality
Total CEPrograms
Grade ADM $ $ $ $ $
K 29 10660 839 1263 71 2173
1 67 24628 1938 2918 163 5019
2 52 19106 1504 2264 127 3895
3 50 18372 2177 122 2299
4 45 16543 1960 110 2070
5 27 9926 1173 62 1235
6 25 9192 1089 61 1150
Sp. 22 8C85 958 54 1012
Totals 317 116512 4281 13802 770 18853
206
Table 77. Program expenditures by grades: School 8.
1965-66RegularPrograms
Teacher Aides,Adjustment
,Teachers,Reauced Class
Size
$
=
1
Total CEPrograms
$Grade ADM $
K 60 11015 1252 1252
1 120 22043 25 04 2504
2 101 18545 2108 21 08
3 102 18728
4 118 21664
5 115 21116
6 89 10339
Sp. 6 0 11015
Totals 765 134465 5864 5864
1966-67RegularPrograms
Teacher Aides,AdjustmentTeachers,
Reduced ClassSize
IntensiveInstructionalImprovement
Total CEPrograms
Grade ADM $ $ $
K 71 18493 3 999 3 999
1 102 26582 5 745 5 745
2 108 28132 6083 6083
3 84 21893 684 684
4 86 22397 701 701
5 98 25536 798 798
6 97 25274 790 790
Sp. 71 18493 578 578
Totals 71 7 186800 15827 3551 1 9378
APPENDIX G
Table 78. Program expenditures by grades: School 9.
207
1965-66RegularPrograms
Teacher Aides,Adjustment
T - 1...s..,..Ivacimi,Reduced Class
Size
Total CEPrograms
Grade ADM $ $ $
K 37 9025 916 916
1 73 17793 1808 1808
2 85 20729 2106 2106
3 69 16818
4 73 17793
5 75 18280
6 84 20485
Sp. 59 14383
Totals 555 135306f
4830 4830
1966-.67Regular
Programs
Teacher Aides,Adjustment
Teachers,Reduced Class
Size
IntensiveInstructionalImprovement
JTotal CEPrograms
Grade ADM $ $ $
K 39 15977 2103 2103
1 69 28259 3721 3721
2 55 22522 2966 2966
3 75 30704 524 524
4 55 22522 384 384
5 70 28663 489 489
6 62 25391 433 433
Sp. 24 9818 168 168
Totals 449 183856 8792 19981
10788
208
Table 79. Program expenditures by grades: School 10.
67TMP-115
1965-66Regular
Prru,rnme-,u -
Teacher Aides,Adjustment
Teachers,Reduced Class
Size
Total CE0-e oogrOnis
Grade ADM $$
K 54 9743 1125 1125
1 103 18586 2146 2146
2 76 13715 1583 1583
3 83 14977
4 80 14439
5 70 12639
6 50 9029
Sp. 57 10291
Totals 574 103419 4854 4854
1966-67Regular
Programs
Teacher Aides,AdjustmentTeachers,
Reduced ClassSize
Total CEPrograms
Grade ADM $ $
K 41 11430 982 982
1 88 24541 2109 2109
2 74 20642 1773 1:773
3 59 16459
4 83 23143
5 69 19243
6 55 15343
Sp. 13 3631
Totals 483 134432 4864 4864
REFERENCES
1. Grotberg, Edith H. , Learning Disabilities and Remediation inDisadvantaged Children, Rev. Educ. Res. , vol 35 (1965).
2. Raph, Jane B. , Language Development in Socially DisadvantagedChildren, Rev. Educ. Res. , vol 35 (1965).
3. U. S. Office of Education, The States Report: The First Year ofTitle I, Summary Report, U. S. Government Printing Office,Washington, D. C.
4. United States Commission on Civil Rights, Racial Isolation in thePublic Schools U. S. Government Printing Office, Washington,D. C.
5. Manual for Administrators, Supervisors and Counselors: IowaTests cif Basic Skills, Houghton, Mifflin Co. , Boston, Mass. ,1964.
6. Technical Supplement, Stanford Achievement Test, Harcourt,Brace and World, Inc. , New York, New York, 1966.
7. Holt, J. , How Children Fail, Dill, New York, New York, 1964.
8. U. S. Office of Education, Guide of Evaluation of Title I Projects,U. S. Government Printing Office, Washington, D. C. , October1966.
9. Wasserman, William, Education Price and Quantity Indexes,Syracuse University Press, Syracuse, New Yo:rk, 1963.
10. Higgins, M. J. , and J. C. Merwin, Assessing the Progress ofEducation: A Second Report, Phi Delta Kappan, vol 48 (1967).
11. U. S. Office of Education, Pupil Accounting for Local and StateSchool Sy.?tems, U. S. Government Printing Office, Washington,D. C. , 1964.
12. Kendall, M. G. , and A. Stuart, The Advanced Theory of Statistics,vol I, Hafner, New York, New York, 1963.
21067TMP-115
13. Peters, C. C. , and W. R. Van Voorhis, Statistical Proceduresand Their Mathematical Bases, McGraw Hill, 1940.
14. Thomson, G. H. , A Formula to Correct for the Effect of Errorsof Measurement on the Correlation of Initial Values withGains, Journal of Experimental Psychology, vol 7 (1925)
15. Scheffe, Analysis of Variance, Wiley, 1959.
UNCLASSIFIEDSecurity Classification
DOCUMENT CONTROL DATA - R&D(Security classification of title. body of abstract and indexing annotation must be entered sewn Ike overall mien S. classified)
1. ORIGINATIN 0 ACTIVITY (Corporate autho;)
General Electric, TEMPOWashington Operation
2.. REPORT SECURITY C LAssIFICATION
Unclas sified -26. OROuP
3. REPORT TITLE
SURVEY AND ANALYSES OF RESULTS FROM rITLE I FUNDING FOR
COMPENSATORY EDUCATION4. DESCRIPTIVE NOTES (Type et tepee end inlusive data.)
1. AUTH0R(3) (Last no Wet now. WNW)
S. REPORT DATE
I March 196870. TOTAL. NO. OF PASmS
225
........71A No. or
15
S.. CONTRAcy OR 4IRANT NO.
HEW-05-67-55IL PROJECT NO.
4.
d. .
SAL ORIOINATOWS REPORT NUMSERM
67TMP-115
SAL 27:411111111PORY NO(S) (Any sift' woolen OW any be ottelened
10. A VA IL AMLITY/LiMITATION NOTICES
Distribution only through Project Officers, Department of Health, Education, andWelfare.
Is. SUPPLEMENTARY NOTES 12. SPONSOR
Department Health, Education and WelfarWashington, D.C. 20211
13. AMSTRACT
This report analyzes data from a sample of 132 schools which received fundsfrom Title I for compensatory education to aid educationally disadvantaged pupils.Mose of the eleven school districts from which the schools were drawn wereselected because there was reason to believe that successful compensatory educa-tion programs were in progress in at least some of the district schools. Conclu-sions are based on a comparison of achievement scores in 1966-67, after pupilswere exposed to compensatory education from Title funds, with achivementscores in 1965-66.
_
UNCLASSIFIEDSecurity Classification
,..