light's retention scale does not predict success in first-grade retainees

5
Pcyrhulogy ,n {he Schools 1982, 10, 310-314 LIGHT’S RETENTION SCALE DOES NOT PREDICT SUCCESS IN FIRST-GRADE RETAINEES’ JONATHAN SANDOVAL Universiry of Caiifornia, Davis Light’s Retention Scale totals, along with measures of reading achievement, mathematics achievement, and self-concept taken at the time first grade nonpromo- tion was being considered, were used to predict nine separate outcome measures of academic, emotional, and social status at the end of the repeated first grade. In none of the stepwise regressions did the Light total add to the prediction of outcome in the children repeating the first grade. This result, added to earlier studies, suggests that Light’s Retention Scale is neither reliable nor valid as a psychometric device. In a previous report (Sandoval, 1980), it was demonstrated that the Light’s Reten- tion Scale Total Score does not meet the traditional standards of reliability and con- current validity. Reliability in the 1980 study was estimated by a measure of internal con- sistency. The stability of the rating was not examined. Although a scale may lack internal consistency, the rating derived still may be stable over time and thus be measuring some construct. The previous report also suggested the scale did not have good internal psy- chometric properties or concurrent validity. It is still possible for a measure to have predictive validity without concurrent validity, although this is not often the case. The present report consists of a one-year follow-up of the children from the first report. The relationship between Light’s Retention Scale and nine outcome measures was assessed via multiple regression. The intent was to determine the predictive validity of Light’s scale for a group of first-grade children not selected on the basis of the scale. METHOD Subjects The same 123 children from the earlier report were followed into the next year. These children were typical of candidates to repeat the first grade, and are the type of children for whom the scale was designed. All were identified in early Spring by teachers and principals as first graders who might need an additional year at this level. Light (1977) states, “Nearly every piece of research on the subject of retention agrees that, in cases where retention has been found beneficial, it is nearly always in a primary grade” (p. 17). Seventy-eight children from the study did repeat first grade and are the subject of this report. The Light score did not enter into the retention decision. Measures Previously, the children had been tested with subtests of the California Achievement Test (CAT; Tiegs & Clark, 1977) and the Woodcock Reading Mastery Tests (Wood- cock, 1973), along with six subtests of the Key Math Diagnostic Test (Connolly, Nacht- man, & Pritchett, 1976). To get a single score for reading, the CAT Reading Com- This study was supported by grant MH 28765-02 from the National Institute of Mental Health Requests for reprints should be sent to Jonathan Sandoval, Dept. of Education, University of California, Davis, Davis, CA 95616. 3 10

Upload: jonathan-sandoval

Post on 06-Jun-2016

301 views

Category:

Documents


6 download

TRANSCRIPT

Page 1: Light's retention scale does not predict success in first-grade retainees

Pcyrhulogy ,n {he Schools 1982, 10, 310-314

LIGHT’S RETENTION SCALE DOES NOT PREDICT SUCCESS IN FIRST-GRADE RETAINEES’

JONATHAN SANDOVAL

Universiry of Caiifornia, Davis

Light’s Retention Scale totals, along with measures of reading achievement, mathematics achievement, and self-concept taken at the time first grade nonpromo- tion was being considered, were used to predict nine separate outcome measures of academic, emotional, and social status at the end of the repeated first grade. In none of the stepwise regressions did the Light total add to the prediction of outcome in the children repeating the first grade. This result, added to earlier studies, suggests that Light’s Retention Scale is neither reliable nor valid as a psychometric device.

In a previous report (Sandoval, 1980), it was demonstrated that the Light’s Reten- tion Scale Total Score does not meet the traditional standards of reliability and con- current validity. Reliability in the 1980 study was estimated by a measure of internal con- sistency. The stability of the rating was not examined. Although a scale may lack internal consistency, the rating derived still may be stable over time and thus be measuring some construct.

The previous report also suggested the scale did not have good internal psy- chometric properties or concurrent validity. It is still possible for a measure to have predictive validity without concurrent validity, although this is not often the case.

The present report consists of a one-year follow-up of the children from the first report. The relationship between Light’s Retention Scale and nine outcome measures was assessed via multiple regression. The intent was to determine the predictive validity of Light’s scale for a group of first-grade children not selected on the basis of the scale.

METHOD Subjects

The same 123 children from the earlier report were followed into the next year. These children were typical of candidates to repeat the first grade, and are the type of children for whom the scale was designed. All were identified in early Spring by teachers and principals as first graders who might need an additional year at this level. Light (1977) states, “Nearly every piece of research on the subject of retention agrees that, in cases where retention has been found beneficial, it is nearly always in a primary grade” (p. 17). Seventy-eight children from the study did repeat first grade and are the subject of this report. The Light score did not enter into the retention decision. Measures

Previously, the children had been tested with subtests of the California Achievement Test (CAT; Tiegs & Clark, 1977) and the Woodcock Reading Mastery Tests (Wood- cock, 1973), along with six subtests of the Key Math Diagnostic Test (Connolly, Nacht- man, & Pritchett, 1976). To get a single score for reading, the CAT Reading Com-

This study was supported by grant MH 28765-02 from the National Institute of Mental Health

Requests for reprints should be sent to Jonathan Sandoval, Dept. of Education, University of California, Davis, Davis, CA 95616.

3 10

Page 2: Light's retention scale does not predict success in first-grade retainees

Light’s Retention Scale 31 1

prehension Score was combined with the Woodcock Word Identification Score (stand- ardized scores were added). The total of the six subtests of the Key Math Test yielded a single score for mathematics. The children also were given the McDaniel-Piers Young Children’s Self-concept Scale (McDaniel, 1978). At the same time, Light’s Retention Scale (Light, 1977) was completed by their teachers. The Total Score was computed us- ing Light’s weights, as was the Short Form Score (with Light’s weights) described in the 1980 report. These measures made up the predictors.

As outcomes, nine separate measures of academic, emotional, and social status were selected: (a) the second year teacher’s rating of academic achievement on a 5-point Likert Scale, (b) the Reading Comprehension score of the CAT, (c) the Total Key Math Score, (d) the teacher’s rating of self-concept on a 5-point scale, (e) the McDaniel-Piers Self-concept score, (f) the teacher’s rating of social skills on a 5-point scale, (g) the Pupil Behavior Rating Survey total score (Lambert, Bower, & Hartsough, 1979), (h) the sec- ond year teacher’s global rating of success of the year on a 5-point scale, and (i) the parent’s rating of success on a 10-point scale. For details of the outcome measures and the predictors, see Sandoval and Hughes (1981). Procedure

The outcome measures were obtained from retesting the children on the same measures one year after the first administration of Light’s Retention Scale, from inter- views and ratings with PBRS with the teacher at this time (Spring of 1980), and from in- terviews with the parents over the following summer. Nine stepwise multiple regressions using forward inclusion criteria were performed, using the three predictors (Reading, Math, and Self-concept) plus Light’s Total Score to account for the variance in each outcome. These regressions were repeated, using the short form total in place of Light’s total.

RESULTS AND DISCUSSION Table 1 contains the summary tables of the nine forward stepwise multiple

regressions. Listed are the steps when the variables enter, the F-ratios, the significance of the F, the multiple correlation, and the univariable correlation. At the bottom of the table is the information about the same analysis with the Short Form Score substituted for the Total Score. Also included are the constant and beta weights of the significant prediction equations.

The first three outcomes are academic measures. They are predicted totally by previous status in academics, and, in the case of the teacher rating, by self-concept. Light scores, Total and Short Form, are not related to academic outcome, a finding consonant with the work of Watson (1979).

Teacher rating of self-concept and McDaniel-Piers self-concept are predicted by reading and previous self-concept, but not by the Light scores. Only McDaniel-Piers self- concept predicts teacher rating of social skills. Light’s Scales are not related to outcomes in the affective and social domain. This finding is not surprising when one considers that few of the items on the Light are related to social or emotional status.

The PBRS rating, an overall measure of mental health adjustment, is predicted by achievement and self-concept, not by the Light scores. The overall w i n g of success by teachers is predicted by self-concept alone. It is only the parent’s rating of success that is predicted by the Short Form of the Light. The regular form does not predict, and, in fact,

Page 3: Light's retention scale does not predict success in first-grade retainees

312 Psychology in the Schools, July, 1982, Vol. 19, No. 3.

TABLE 1 Multiple Regression Summary Tables of Nine Outcomes Predicted from

Achievement, SelJXoncept, and Light's Retention Scale

Step Variable 1 Reading 2 Self-concept 3 Mathematics 4 LightTotaF

(4) (Short Form*) Constant at Step 2: 26.13

Step Variable 1 Reading 2 Mathematics 3 Self-concept 4 LightTotal

(3) (Short Form) Constant at Step 2: -.33

Step Variable 1 Mathematics 2 Reading 3 Self-concept 4 LightTotal

(4) (Short Form) Constant at Step 1 : 15.26

Step Variable 1 Reading 2 Self-concept 3 Mathematics 4 LightTotal

(4) (Short Form) Constant at Step 1 : 29.95

Step Variable 1 Reading 2 Self-concept 3 LightTotal 4 Mathematics

(3) (Short Form) Constant at Step 2: 1902.72

Teacher Rating of Academic Standing* F to Enter Multiple Simple Beta df = 1,73 Significance R r Weight

14.38 .oo .43 -.43 -.43 3.95 .05 .49 -.34 -.16 .42 .52 .49 -.24 .02 .88 .49 .07

(.W

Reading Comprehension F to Enter Multiple Simple Beta df = 1,71 Significance R r Weight

37.46 .oo .61 .61 .54 11.51 .oo .69 .52 .26

.33 .57 .69 .27

.10 .76 .69 -.13 (.@I

Mathematics F to Enter Multiple Simple Beta df = 1.71 Significance R r Weight

86.53 .oo .76 .76 .81 .58 .45 .77 .34 .ll .74 .77 .19 .05 .83 .77 -.19

(.12)

Teacher Rating of Self-concept F to Enter Multiple Simple Beta df = 1,73 Significance R r Weight

8.42 .oo .35 .35 .92 1.44 .23 .37 .24 .87 .35 .39 .24 .74 .39 .40 .04

( - .07)

Pupil Behavior Rating Scale* F to Enter Multiple Simple Beta df = 1,70 Significance R r Weight

10.29 .oo .38 -.38 -3.72 5.16 .03 .46 -.36 -20.36 .o 1 .91 .46 .05 .02 3 9 .46 -.15

(. 14)

Page 4: Light's retention scale does not predict success in first-grade retainees

Light’s Retent ion Scale

TABLE 1 (continued)

313

Step Variable 1 Self-concept 2 Reading 3 Light Total 4 Mathematics

(3) (Short Form) Constant at Step 2: 21.08

Step Variable 1 Self-concept 2 Mathematics 3 Reading 4 LightTotal

(2) (Short Form) Constant at Step 1 : 10.68

Step Variable 1 Self-concept 2 Reading 3 Mathematics 4 Light Total

(3) (Short Form) Constant at Step 1 : 3.12

Step Variable 1 LightTotal 2 Reading 3 Mathematics 4 Self-concept

(1) (Short Form)

Self-concept F to Enter Multiple Simple df = 1,70 Significance R r

7.03 .01 .32 .32 1.81 .18 .36 .25 1.07 .31 .38 .09 .16 .69 .38 .05

(-.15)

Teacher Rating of Social Skills* F to Enter Multiple Simple df = 1,73 Significance R r

4.93 .03 .27 -.27 .99 .32 .30 .06 .77 .38 .32 -.14 .04 .85 .32 .01

(. 1 3)

Teacher Global Rating of Success* F to Enter Multiple Simple df = 1,73 Significance R r

6.97 .01 .32 -.32 2.63 .I1 .37 -.28 .67 .42 .38 -.22 .32 .58 .39 -.01

(-.ll)

Parental Global Rating of Success F to Enter Multiple Simple df = 1,62 Significance R r

2.61 .ll .20 .20 2.01 .16 .27 .16 .54 .46 .28 - .06 .57 .46 .30 .12

(4.66) ~ 0 3 ) (-.26)

Beta Weight

.30

Beta Weight

-.95

Beta Weight

-.56

Beta Weight

*Lower values equal better adjustment or higher skills.

the sign of the correlation for the long form is opposite from that expected. The other predictors were not related to parental global judgment.

The single significant result with the Short Form is not sufficient to validate Light’s Retention Scale for use with first-grade children, The previous study and the present one together indicate that the psychometric use of Light’s Retention Scale is totally un- tenable.

REFERENCES CONNOLLY, A. J., NACHTMAN, W., & PRITCHETT, E. M. Key Math Diagnostic Arithmetic Test. Circle Pines,

MN: American Guidance Service, 1976.

Page 5: Light's retention scale does not predict success in first-grade retainees

314 Psychology in the Schools, July, 1982, Vol. 19, No. 3.

LAMBERT, N. M., BOWER, E. M., & HARTSOUGH, C. S.

LIGHT, H. W. MCDANIEL., E. The McDaniel-Piers Young Children’s Self-concept Scale. West Lafayette, IN: Purdue

SANDOVAL, J. Reliability and concurrent validity of Light’s Retention Scale. Psychology in the Schools,

SANDOVAI., J., & HUGHES, G. P. Success in non-promoted first grade children, Davis, CA: University of California, Davis, Department of Education, 1981.

T i t ~ s , E. W., & CLARK, W. W. California Achievement Tests. Monterey, CA: California Test Bureau/McGraw Hill, 1977.

WATSON, D. The relative efficiency of the Light Retention Scale in identifying children for retention. Paper presented at the annual convention of the National Association of School Psychologists, San Diego, March, 1979.

Woodcock Reading Mastery Tests. Circle Pines, MN: American Guidance Service, 1973.

The Pupil Behavior Rating Scale. Monterey, CA: Publisher’s Test Service, 1979.

Light’s Retention Scale. San Rafael, CA: Academic Therapy Publication, 1977.

University Department of Education, 1978.

1980, 17, 442-445.

WOODCOCK, R. W.