an investigation of the effects of randomization of the item
TRANSCRIPT
Western Michigan University Western Michigan University
ScholarWorks at WMU ScholarWorks at WMU
Master's Theses Graduate College
4-1974
An Investigation of the Effects of Randomization of the Item Order An Investigation of the Effects of Randomization of the Item Order
in the Mini-Mult in the Mini-Mult
Robert A. Bornstein
Follow this and additional works at: https://scholarworks.wmich.edu/masters_theses
Part of the Psychoanalysis and Psychotherapy Commons
Recommended Citation Recommended Citation Bornstein, Robert A., "An Investigation of the Effects of Randomization of the Item Order in the Mini-Mult" (1974). Master's Theses. 2514. https://scholarworks.wmich.edu/masters_theses/2514
This Masters Thesis-Open Access is brought to you for free and open access by the Graduate College at ScholarWorks at WMU. It has been accepted for inclusion in Master's Theses by an authorized administrator of ScholarWorks at WMU. For more information, please contact [email protected].
AN INVESTIGATION OF THE EFFECTS OF RANDOMIZATION OF THE ITEM ORDER IN THE MINI-MULT
by
Robert A. Bornstein
A Thesis Submitted to the
Faculty of The Graduate College in partial fulfillment
of theDegree of Master of Arts
Western Michigan University Kalamazoo, Michigan
April 1974
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
ACKNOWLEDGEMENTS
I am deeply indebted to Drs. Malcolm Robertson, Paul Fuller
and George Sidney for the advice and constructive criticism from
which I benefited in writing this thesis. My special thanks goes
to Dr. Michael Stoline of the Computer Center, who was of immea
surable assistance in the preparation of this document. The
training and education that I have received from the faculty of
the Department of Psychology have made graduate study at Western
Michigan University a most rewarding experience.
Robert A. Bornstein
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
INFORMATION TO USERS
This material was produced from a microfilm copy of the original document. While the most advanced technological means to photograph and reproduce this document have been used, the quality is heavily dependent upon the quality of the original submitted.
The following explanation of techniques is provided to help you understand markings or patterns which may appear on this reproduction.
1.The sign or "target” for pages apparently lacking from the document photographed is "Missing Page(s)". If it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting thru an image and duplicating adjacent pages to insure you complete continuity.
2. When an image on the film is obliterated with a large round black mark, it is an indication that the photographer suspected that the copy may have moved during exposure and thus cause a blurred image. You will find a good image of the page in the adjacent frame.
3. When a map, drawing or chart, etc., was part of the material being photographed the photographer followed a definite method in "sectioning" the material. It is customary to begin photoing at the upper left hand corner of a large sheet and to continue photoing from left to right in equal sections with a small overlap. If necessary, sectioning is continued again - beginning below the first row and continuing on until complete.
4. The majority of users indicate that the textual content is of greatest value, however, a somewhat higher quality reproduction could be made from "photographs" if essential to the understanding of the dissertation. Silver prints of "photographs" may be ordered at additional charge by writing the Order Department, giving the catalog number, title, author and specific pages you wish reproduced.
5. PLEASE NOTE: Some pages may have indistinct print. Filmed as received.
Xerox University Microfilms300 North Zeeb RoadAnn Arbor, Michigan 48106
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
MASTERS THESIS M-5596BORNSTEIN, Robert Alan AN INVESTIGATION OF THE EFFECTS OF RANDOMIZATION OF THE ITEM ORDER IN THE
MINI-MULT.Western Michigan University, M.A., 1974 Psychology, clinical
U nive rs ity M icro film s, A XEROX Company , A nn A rbor, M ich igan
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
TABLE OF CONTENTS
SECTION PAGE
I INTRODUCTION........................ 1
II METHODS............................ 7
Subjects......................... 8
Procedure........................ 9
III RESULTS............................ 13
IV DISCUSSION.......................... 27
V APPENDICES.......................... 32
Appendix A....................... 33
Appendix B....................... 34
Appendix C....................... 35
Appendix D....................... 36
VI BIBLIOGRAPHY........................ 37
iii
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
LIST OF TABLES
TABLE
1
2
3A
3B
4A
4B
5
6A
6B
6C
Median Age Levels for Groups 1 and 2, Males and Females
Differences Between Like Scales on Two Forms of the Mini-Mult
Effect of the Order of Presentation of the Two Forms of theMini-Mult on All Scales
Effect of the Order of Presentation of Two Forms of the Mini-Mult and the MMPI on All Scales
Mean Scale Values on All Three Forms for Group 1
Mean Scale Values on All Three Forms for Group 2
Differences Between Like Scales on Two Forms of the Mini-Mult and the MMPI
Group 1 _t-Scores and Probability Values for Comparisons Between Mini-Mult-A, Mini-Mult-B, and the MMPI
Group 2 t̂ -scores and Probability Values for Comparisons Between Mini-Mult-A, Mini-Mult-B, and the MMPI
Groups 1 and 2 Combined J:-scores and Probability Values for Comparisons Between Mini-Mult-A, Mini- Mult-B, and the MMPI
iv
PAGE
10
15
16
18
20
21
22
24
25
26
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
INTRODUCTION
Shortly after the publication of the MMPI, users of the test
became aware of the difficulties encountered with an instrument
of the length of the MMPI. These problems included the length
of time required to score the test, and the possible traumatizing
effects on the individuals to whom the test was administered due
to the sheer number of items to which to respond. The first attempt
to reduce the number of items was made by Ferguson (1946) , who
suggested the removal of 200 items that were not, at that time, in
cluded in the scoring of any scales. During the first decade of
the existence of the MMPI, further attempts to shorten the inven
tory consisted of various methods of removing the items not nor
mally scored. These investigations resulted in short forms of the
MMPI with item populations of 372 (MacDonald, 1952A and 1952B)
and 420 (Olson, 1954).
More recently, the attempts to develop short forms have made
a more pronounced reduction in the number of items. Foulds, Caine,
and Creasy (1960), and Jorgenson (1958), made attempts along these
lines; but the forms they developed have failed to gain wide spread
acceptance because they do not predict the full scale MMPI. Kin-
cannon (1968) has developed an abbreviated version, the Mini-Mult,
which reportedly yields only a 9% loss in reliability, and a 14%
loss in profile correspondence when compared with the full MMPI.
1
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
It was felt that this loss in reliability and correspondence was not
sufficient to preclude the use of the Mini-Mult in instances when
the full test could not be administered.
Since Kincannon's introduction of the Mini-Mult, several inves
tigations have been made of the utility of this most recent short
form in a variety of settings, and with a variety of populations.
Generally, the results of these studies lead to two alternative
positions in regard to the utility of the Mini-Mult. One group,
whose work essentially replicated the data reported by Kincannon
(Lacks, 1970; Newton, 1971; Gayton and Wilson, 1971; Gayton, et. al.,
1972) , concluded that the Mini-Mult can be used as an adequate sub
stitute for the full scale MMPI. The other group (Armentrout, 1970;
Armentrout and Rouzer, 1970; Harford, et. al., 1972; and Streiner, et.
al., 1973) disputed the adequacy of the Mini-Mult as an effective sub
stitute for the MMPI. This latter group regarded the Mini-Mult as a
gross indicator of pathology, lacking the precision of the MMPI.
Regardless of whether the authors are supporters or detractors
of the Mini-Mult, certain deficiencies of this instrument are near
ly always reported, that is, the underestimation and overestimation
of various scales on the MMFI. The degree to which this phenomenon
is reported, appears to be related to the specific method of adminis
tering this short form, and the nature of the population under exam
ination.
Two methods of obtaining Mini-Mult data have been utilized in
previous investigations, the "internal" method, and the "external"
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
method. The former involves the completion of a full scale MMPI
followed by scoring with special Mini-Mult keys, while the latter en
tails an independent administration of only the Mini-Mult items.
It has been reported (Kincannon, 1968; Streiner, 1973) that an
internal Mini-Mult yields higher standard scale score correlations
with the MMPI than does an external administration. Furthermore, all
studies utilizing the external form, in comparison with the MMPI,
have failed to yield correlation values of the magnitude reported by
Kincannon (Armentrout, 1970; Armentrout and Rouzer, 1970; Newton,
1971; Gayton, et. al., 1972).
There also seems to be a relationship between the nature of the
subject population and the reported results. The studies which have
employed the independent administration method, also appear to have
studied non-psychiatric populations. Although Newton (1971) used
alcoholic in-patients, these individuals were without psychiatric
diagnoses.
One investigator (Pulvermacher and Bringmann, 1971) used the in
ternal Mini-Mult with a non-psychiatric population (college students),
and obtained results which indicated a poor correlation between the
Mini-Mult and the MMPI.
Hartman and Robertson (1972) employed both internal and external
forms with a psychiatric population and found imprecise scale predic
tions with both forms.
In view of the above data, it is possible that an interaction
exists between the psychiatric nature of the subject population, and
the method of test administration. There is direct experimental evi-
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
dence to support the view that the method of administration of the
Mini-Mult (internal vs. external) affects the degree of correlation
between the long and short forms. (Kincannon, 1968; Streiner, et.
al., 1973)
These data are also in support of the view that performance on
a given item is, in fact, influenced by the context in which it
occurs. (The only difference between the two versions of the Mini-
Mult is that one is embedded within the full MMPI).
Earlier authors have cautioned against altering the context in
which items are presented to subjects. Among those who expressed
skepticism, Gough (1946), questioned the deletion of 200 items from
the individual form of the MMPI. He contended that these items form
ed part of the "question environment" on which the test was original
ly standardized.
The discrepancies between the internal and external forms of the
Mini-Mult in predicting MMPI scores may be a demonstration of the
effects of requiring subjects to respond to a large vs. a small num
ber of items. Or, it may be argued that since the two versions of
the short form are equivalent in terms of the number of scorable items,
the differences arise as a result of the context in which they are
presented.
However, the actual size of the item population is not the only
variable to be considered when assessing the effect of the context in
which certain items occur. The sequence in which these items are pre
sented may also exert an unknown effect.
In the initial formulation of the MMPI as a card-sort test
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
5
(Hathaway and McKinley, 1940), the authors take this sequence effect
into account by suggesting the shuffling of the item cards prior to
administration of the test. They maintain that "in this way no item
has a constant effect on a subsequent item through any series of
individual administrations." Sequence effects cannot be entirely
eliminated because the test itself constitutes a context in which
the items occur.
This sequence effect seems to be ignored in the more recent
group forms of the MMPI, but these forms have the advantage of pre
senting a set of stimuli that is constant for all individuals who
take the test.
The fact that Hathaway and McKinley address themselves to this
question of the effects of sequencing implies that there was reason
to suspect that there would be some effect demonstrated from the
order of presentation. This may contribute to the discrepancies
observed between the two versions of the Mini-Mult due to the elim
ination of intervening items not scored for the Mini-Mult.
An alternative way to view the item sequence issue is to con
sider the effects of altering the order of presentation of the en
tire item population. As Hathaway and McKinley (1940) contend, this
procedure should have no effect on the scores that result. No re
port of this procedure with the full MMPI or any of the short forms
was found in the literature.
An extension of Hathaway and McKinley's reasoning with the
MMPI would lead one to conclude that an alternation of the order of
presentation of items in the Mini-Mult should likewise have no effect
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
on the scores generated by this abbreviated form.
On the other hand, if the previous work on the Mini-Mult is
taken into consideration, it becomes apparent that there are dif
ferences between various administrative forms of the Mini-Mult
(Newton, 1971; Streiner, et. al. , 1973), as well as between the
Mini-Mult and the full scale MMPI.
This study was designed to determine what effect the order of
presentation of items on the Mini-Mult would have in predicting
scores on the full scale MMPI. To accomplish this, a form of the
Mini-Mult was constructed which consists of a randomization of the
order of presentation of Kincannon's 71 items.
It was hypothesized that the randomized version of the Mini-
Mult would not differ significantly from the standard "external"
Mini-Mult in the prediction of full scale MMPI scores. However, if
significant differences do occur, it is hypothesized that these
differences result from the sequence of items presented on the form
of the external Mini-Mult.
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
METHOD
This study was conducted with the clients of a deferred prosecu
tion agency in Kalamazoo, Michigan. All subjects had either been ar
rested and charged by law enforcement officials with one of a number
of felonies, all involving theft of one form or another (Appendix A);
or had been implicated in these activities by another individual who
had come to the attention of the legal authorities. Most subjects
were never formally changed or incarcerated, but were referred to
this agency as an alternative to the initiation of the conventional
legal processes. The program was designed to give first offenders,
and "non-pattern" offenders an opportunity to escape the stigma of
having a criminal felony record. "Non-pattern" offenders were de-
finded as individuals who may have had previous arrest records, but
not in the area of theft-related offenses.
Subjects' cases were referred to the agency within a maximum of
48 hours following being brought to the attention of the authorities.
Participation in the program was completely voluntary. Several sub
jects terminated prematurely due either to a lack of interest in the
program, or availing themselves of other means of escaping their pre
dicament (such as joining the armed forces). There were no instances
of subjects declining to participate in the study. Those individuals
who demonstrated a lack of interest in the program, or who were not ac
cepted into the program, were processed through traditional legal chan
nels.
Assessment and evaluation of each case (which included the MMPI7
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
and a personal background investigation, as well as interviews with
agency staff) took approximately three weeks, after which the indi
vidual was either accepted or rejected by the agency. If accepted,
the subject would work with probation officers, social workers, and
other personnel to eliminate those personal factors which were re
lated to the current infringement of the law. In addition to this
personal rehabilitation, the subject was also required to make some
form of restitution. This restitution could be made directly to the
victim of the crime, or it could consist of volunteer work in some
community agency. Upon successful completion of the program, the
subject's criminal record (in regard to the specific offense for which
he was referred) would be erased. Due to the brief existence of this
program, it is difficult to state the average length of time required
for successful completion of the program, but it is expected that
this will be 6 to 12 months.
Subjects
The sample consisted of 51 subjects assigned to two treatment
groups. An additional subject (#51) was included inadvertently, due
to overcollection of data. Group 1 consisted of 25 subjects, while
Group 2 contained 26 subjects. All subjects were assigned to the
groups based on the date of their initial contact with the agency.
Assignment was conducted on an alternating basis. That is, the first
male subject referred was assigned to Group 1, while the second male
subject referred was assigned to Group 2. This procedure was utiliz
ed for both males and females until the desired number of subjects
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
9
were obtained. The assignment of males and females was conducted
independently.
The ages of the subjects ranged from 17 to 40 with a median age
of 18.2 years. The median age for all males was 17.78 years, and
the median age for all female subjects was 19.50 years. Only 7 of
the 51 subjects were other than Caucasion, all of these being Negro.
The median grade level for Group 1 was 11.12 years, while the median
grade level for Group 2 was 11.50 years. The educational range was
from the 8th grade to college senior. The median grade level for
all males was 10.85, and the median grade level for all females was
11.83. The total sample included 35 males, and 16 females. Seven
subjects who were included in the sample were ultimately rejected
from the deferred prosecution program.
Subjects participating in the study were told only that they
were assisting the agency in evaluating the intake process. Fol
lowing the completion of the study, all subjects received a thorough
briefing on the nature of the investigation. (The agency was in
fact attempting to evaluate the efficacy of its intake procedure, and
incorporated this research design as a means of accomplishing this
goal.)
Procedure
Two forms of the Mini-Mult were utilized: (1) Kincannon's
original form, and (2) a randomized version of Kincannon*s form.
(Kincannon*s form will herein be referred to as Form A, while the
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
TABLE 1
Median Age Levels for Groups 1 and 2, Males and Females
Group 1___________ Group 2
# of males 18 17# of females 7 9Median Age Males 17.83 17.70Median Age Females 20.50 19.50Median Age Total 18.50 18.07
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
11
randomized version will be referred to as Form B.)
Kincannon (1968) proposed a drastically shortened version of
vhe MMPI which he referred to as the Mini-Mult. This abbreviated
form contained a total of 71 items and generated the standard MMPI
scales with the exception of Mf and Si. Kincannon's results showed
that the use of the Mini-Mult resulted in only a 9% loss in reliabil
ity and a 14% loss in correspondence, in comparison with the standard
MMPI. This loss, according to the author, did not preclude the use
of the Mini-Mult. Despite the authors contentions, there are re
ports of various difficulties with the Mini-Mult such as the over-
astimation and underestimation of scale values (Armentrout, 1970;
Armentrout and Rouzer, 1970).
All subjects received a total of three test administrations,
with an intertest interval of 7 to 14 days. The three test adminis
trations consisted of both forms of the Mini-Mult as well as the full
scale MMPI (booklet form). For all subjects the second test adminis
tration was the MMPI. Group 1 received Form A on the first adminis
tration, while Group 2 received Form B on the first administration.
The third administration consisted of the short form not yet adminis
tered to the subject.
All testing was administered and supervised by various members
of the agency staff, all of whom were instructed in the types of
information that could be provided to the subjects. In every instance,
the test protocols were scored by the present investigator. Immedi
ately after scoring, the data were recorded on a record sheet which
ultimately contained an individual subject's scores on all three
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
12
test administrations (Appendix B). A separate sheet was provided for
each subject, and was coded for identification purposes only. A
record was kept in the subjects' agency file to insure test admin
istration in the proper sequence.
Responses to all three test administrations were recorded on
IBM 805 answer sheets. Scoring templates for hand scoring were
constructed for both short forms, and standard hand scoring keys
were used to score the full scale MMPI.
In instances where obtained raw scores were so low or high
that no t-scores were available (either on the standard MMPI profile
sheet, or in the MMPI manual) it was decided to assign t-scores in
order to incorporate these data in the analysis. In each case, the
most extreme t-score available in the indicated direction was assign
ed. This procedure was necessary only three times; once each on
scales F, Hy, and Sc.
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
RESULTS
The data were analyzed by the use of a two way repeated mea
sures analysis of variance (Winer, 1971). This procedure was per
formed on two occasions, including full scale MMPI data in the second
analysis only. _t-tests were also computed to compare mean values on
the scales of the various forms within each group, and across all
subjects. The data analysis was conducted at the Computer Center,
Western Michigan University utilizing a program entitled RMAOV #1.
9.4.
The various test results were used in the following manner.
The analysis of variance (ANOVA) F test, and the AB interaction test
(from the ANOVA) were both viewed as preliminary tests. In the event
that a significant F value was obtained on any given scale, the t-
tests computed for that scale were referred to, in an attempt to de
termine the source of variation. The AB interaction, if significant
at p<.25, was not interpreted by itself, but rather was used as a
further indication of the necessity of consulting the t-tests. In the
presence of interaction it was felt that interpretation of the main
effects test from the ANOVA was contaminated, and therefore t-tests
were used. (see Bancroft, 1968)
Carmer and Swanson (1973) compared a variety of techniques of
data analysis, and suggested the use of this procedure as being
superior to other methods, e.g. Tukey, Scheffe7! The authors sug-
tested that if the F value in the ANOVA preliminary test was signi-
13
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
14
ficance at the .05 level, then follow-through t-tests should be run
at the sane level of significance. This procedure can be described
as a protected least significant difference method.
The results of the data evaluation revealed the following pat
terns. When the MMPI data were excluded from the analysis, in no
instances were significant differences obtained between like scales
on the two versions of the Mini-Mult. On only one scale (Hs) was
significance at the .05 level approached, (F=3.91, p=0.054, df=l,49).
Only one other scale (Hy) obtained a probability value that was less
than p=.10, (F=2.89, p=0.095, df=l,49); and two scales obtained
values that were less than p=,20, (Pa, F=1.69, p=0.199) and (Sc,
F=2.30, p=0.136). The other scales examined in this first analysis
had obtained F values that were associated with probabilities rang
ing as high as p=0.94 (Scale F). (See Table 2).
Similar results were obtained in reviewing the effects of the
position in which the two forms of the Mini-Mult were presented to
the subjects. As can be seen in Table 3A, in no instance does the
effect of the presentation of the forms approach significance. The
range of probability values in regard to this variable is from
p=0.397 (scale Pd) to p=0.994 (scale D) with a mean probability
value of X=0.730.
The t-tests that were associated with this first analysis re
vealed significant differences between the two forms of the Mini-
Mult on scales L, Hs, Hy, Pt, Sc, and Ma. In all cases, with the
exception of scale L, the differences between forms was observed in
Group 1. Furthermore, on scales L and Ma, the obtained differences
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
TABLE 2
Differences Between Like Scales on Two Forms of the Mini-Mult
AB Interaction Scale F Prob. Prob.
L 1.41 0.24 0.12F 0.01 0.94 0.68K 0.05 0.82 0.70Hs 3.91 0.054 0.70D 0.20 0.66 0.67Hy 2.89 0.095 0.57Pd 0.02 0.90 0.18Pa 1.69 0.199 0.80Pt 0.78 0.38 0.22Sc 2.30 0.136 0.15Ma 0.24 0.63 0.04
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
TABLE 3A
Effect of the Order of Presentation of the Two Forms of the Mini-Mult on All Scales.
Scale F Prob.
L 0.48 0.49F 0.02 0.89K 0.00 0.96Hs 0.05 0.83D 0.00 0.99Hy 0.17 0.68Pd 0.73 0.40Pa 0.12 0.73Pt 0.27 0.61Sc 0.50 0.48Ma 0.00 0.97
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
17
were a result of higher scores on the randomized form of the Mini-
Mult. In the other instances, the differences were related to high
er scores on the original Mini-Mult.
As suggested by Bancroft (1968), interaction effects may be
taken into consideration, in regard to main effects, when the pro
bability associated with the interaction falls below p=0.25. In
association with the first ANOVA, this phenomenon occurs five times,
on scales L, Pd, Pt, Sc, and Ma. As can be seen from the data
(Table 2), on only one scale (Sc) is a significant interaction value
(p=.15) associated with a scale value that even remotely approaches
significance (p=0.136).
The negligible effects of the order of test presentation are
further demonstrated in the consideration of this variable in regard
to all three test administrations, (Table 3B). As in Table 3A, in
no instance does the effect of order of presentation of the three
forms approach significance. The range of the probability values
associated with this variable in relation to all three forms is from
p=0.15 (scale Pd) to p=0.97 (scale Pt), with a mean probability
value of X=0.592.
A second two way repeated measure ANOVA was performed which
included the full scale MMPI data as well as the data from the two
Mini-Mult forms. The results of this analysis are markedly different
than that obtained from the initial ANOVA.
In regard to the differences between like scales on the various
forms, significance was achieved on 8 of 11 scales that were examined.
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
TABLE 3B
Effect of the Order of Presentation of Two Forms of the Mini-Mult and the MMPI on All Scales
Scale F Prob.
L 1.12 0.30F 0.36 0.55K 0.25 0.88Hs 0.30 0.59D 0.18 0.67Hy 0.41 0.52Pd 2.18 0.15Pa 0.41 0.52Pt 0.00 0.97Sc 0.13 0.72Ma 0.21 0.65
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
19
(See Tables 4A and 4B for mean values on the scales of all three test
forms.) In addition to these scales, another also approached sig
nificance, scale Hs, (F=2.88, p=0.06, df=l,49). The other two scales
included in this analysis received probability values of p=0.54
(scale D), and p=0.35 (scale Pd). As can be seen from the data
(Table 5) scales K, Pt, Sc, and Ma achieved probability values that
were highly significant (p=<.001).
t-tests were also computed in association with this second ANOVA
to compare the three test forms administered to the subjects. For
both groups, comparisons were made between the original and random
ized Mini-Mults; the original Mini-Mult and the MMPI; and the ran
domized Mini-Mult and the MMPI. These three comparisons were also
made on the combined data of both groups. This resulted in a total
of 99 t-test comparisons (9 on each of 11 scales).
A review of the comparisons between the two Mini-Mult forms
reveals that on scales L, Hs, Hy, Pt, and Ma, differences of some
significance were found. On all other scales in comparisons of
this type, substantial agreement was observed. Of the total of 33
comparisons made between the two Mini-Mult forms, only 6 were shown
to be significant.
In comparisons between the original Mini-Mult and the MMPI, it
was found that only on scale Hs were there no significant differenc
es. Of the total of 33 comparisons of this type, 23 obtained signi
ficant values.
In regard to the comparisons between the randomized Mini-Mult
and the MMPI, it was found that scales D and Pd were the only scales
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
20
TABLE 4A
Mean Scale Values on All Three Forms for Group 1.
Scale MM-A MM-B MMPI
L 52.04 52.36 49.32F 56.20 55.84 57.80K 49.80 49.32 53.04Hs 51.28 48.72 50.12D 58.48 57.20 54.64Hy 57.96 56.12 53.76Pd 62.64 60.88 61.48Pa 56.52 55.48 58.00Pt 51.52 48.80 57.44Sc 48.72 45.00 60.12Ma 53.00 55.72 59.32
(Original Mini-Mult is MM-A, Randomized Mini-Mult is MM-B)
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
TABLE 4B
Mean Scale Values on All Three Forms for Group 2
Scale MM-A MM-B MMPI
L 54.69 52.58 53.00F 56.19 56.69 62.77K 49.61 49.73 54.00Hs 51.50 49.77 53.31D 57.88 57.85 58.23Hy 58.46 57.54 56.27Pd 62.88 64.92 67.69Pa 57.69 56.15 61.54Pt 48.19 48.61 61.27Sc 45.23 45.15 66.31Ma 55.27 53.65 63.04
(Original Mini-Mult is MM-A, Randomized Mini-Mult is MM-B)
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
TABLE 5
Differences Between Like Scales on Two Forms of the Mini-Mult and the MMPI.
Prob.Scale F Prob. AB Interaction
L 3.24 0.043 0.127F 4.67 0.012 0.234K 12.79 0.000 0.815Hs 2.88 0.061 0.398D 0.62 0.539 0.383Hy 5 . 46 0.006 0.582Pd 1.07 0.346 0.109Pa 4.10 0.020 0.561Pt 38.55 0.000 0.031Sc 55.00 0.000 0.042Ma 16.98 0.000 0.089
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
on which significant differences were not found. Of the 33
comparisons of this type, 22 reached levels of significance.
Interaction values were also considered in association with
this second ANOVA. It was revealed that levels of significance
for interaction (p=0.25) were reached on six scales (see Table 5).
With the exception of scale Pd, each scale obtaining a significant
interaction value was also observed to have reached a level of
gnificance in regard to scale differences. (See tables 6A, 6B,
and 6C for the t_-tests and associated probability values).
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
24
TABLE 6A
Group 1 jt-scores and Probability Values for Comparisons Between Mini-Mult-A, Mini-Mult-B, and the MMPI
Scale_______ MM-A./MM-B________ MM-A/MMPI________ MM-B/MMPI
L -0.37 3.12* 3.47**F 0.23 -1.02 -1.25K 0.53 -3.59** -4.12**Hs 2.26' 1.02 1.24D 0.82 2.46" 1.64Hy 1.89 4.32** 2.42"Pd 1.23 0.81 -0.42Pa 0.73 -1.03 -1.76Pt 2.01’ -4.38** -6.35**Sc 1.93 -5.91** -7.84**Ma -1.99' -4.63** -2.64*
p=.05 (' ) , P=.02 ("), p=.01 (*), p=<.01 (**)
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
25
TABLE 6B
Group 2 _t-scores and Probability Values for Comparisons Between Mini-Mult-A, Mini-Mult-B, and the MMPI
Scale MM-A/MM-B MM-A /MMPI________ MM-B/MMPI
L 2.42" 1.94 0.48F -0.32 -4.21** -3.89**K -0.13 -4.86** -4.73**Hs 1.53 -1.60 -3.12*D 0.03 -0.22 -0.25Hy 0.95 2.25" 1.30Pd -1.42 -3.36* -1.93Pa 1.07 -2.69* -3.76**Pt -0.31 -9.68** -9.36**Sc 0.04 -10.93** -10.97**Ma 1.18 -5.69** -6.88**
p=.05 ('), p=.02 ("), p=.01 (*), P=<.01(**)
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
26
TABLE 6C
Groups 1 and 2 Combined _t-scores and Probability Values for Comparisons Between Mini-Mult-A, Mini-Mult-B and the MMPI
Scale________ MM-A/MM-B_________ MM-A/MMPI_____ MM-B /MMPI
L 1.50 3.58** 2.08'F -0.71 -3.78** -3.71**K 0.28 -6.05** -6.33**Hs 2.69* -0.45 -3.14*D 0.59 1.56 0.97Hy 2.01' 4.66** 2.65*Pd -0.18 -1.88 -1.70Pa 1.29 -2.68* -3.97**Pt 1.18 -10.11** -11.30**Sc 1.38 -12.10** -13.48**Ma -0.54 -7.39** -6.85**
p=.05 ('), p=.02 ("), p-.Ol (*), p=<.0] (**)
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
DISCUSSION
A review of the results of this investigation indicates sub
stantial agreement with a number of previous investigations of the
utility of the Mini-Mult in predicting full scale MMPI scores.
(Armentrout, 1970; Armentrout and P.ouzer, 1970; Harford, et. al. ,
1972; Streiner, et. al., 1973).
The data in Table 2 fail to indicate a significant difference
between the two forms of the Mini-Mult. When the full scale MMPI
data are introduced, (Table 5), differences are obtained of varying
significance, on 9 of 11 scales. This fact would seem to argue
for the existence of real differences between the MMPI and the Mini-
Mult.
Although neither of the two short forms was particularly
adequate in the prediction of MMPI scores; the null hypothesis is
accepted, due to the lack of any significant differences between the
two versions of the Mini-Mult. In fact, when the _t-test comparisons
between the two short forms are examined (Tables 6A, 6B, and 6C),
it is seen that only 6 comparisons achieve a level of significance.
Of these six, approximately 67% are significant at the .05 level,
with one comparison significant at the .02 level, and one signi
ficant at the .01 level. The combined data from both groups account
for two of these six significant comparisons (one at the .05 level,
and one at the .02 level). In situations where a large number of
comparisons are made, it is expected that some will yield significant
27
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
28
values due to chance alone. It is therefore regarded that in ef
fect, no substantial differences were obtained between the two
forms of the Mini-Mult in their ability to predict full scale MMPI
scores.
Failure to obtain a difference between the original and ran
domized forms of the Mini-Mult is regarded as evidence which sup
ports the hypothesis that the order of presentation of items does
not exert an effect on the ability of the Mini-Mult to predict MMPI
scores. The present data are in support of Hathaway and McKinley
(1940), as it can be seen that randomzation of the order of pre
sentation of items had no apparent effect on the scores that were
generated, as compared to a non-randomized set of items. These data
seem to argue against the notion of a fixed "question environment"
required for an appropriate use of the instrument. An extension of
the implications of these data would lead one to believe that a
similar randomization of the 550 items of the full scale MMPI would
likewise produce no significant differences as compared to the
standard presentation.
The design utilized in this investigation also permitted a
determination to be made as to whether the order of presentation of
test forms had any bearing on the scores that were obtained. As
can be seen from Tables 3A and 3B, in no instance does the order
of presentation of test forms reasonably approach a significant
level. This lack of significance in regard to test form presentation
is as would be expected, and no interpretive value is associated vrith
this fact.
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
29
The most consistent and highly significant finding of this in
vestigation was the overestimation and underestimation of the scale
scores of the MMPI by both versions of the Mini-Mult. As can be
seen from the data (Tables 4A and 4B), the range of these imprecise
predictions is from an overestimation of 4.20 for scale Hy (Group
I) to an underestimation of 21.16 on scale Sc (Group II). The
most accurate prediction of any scale score occurred on scale D for
Group II where both forms of the Mini-Mult underestimated the full
scale MMPI score by less than .40. In spite of the fact that these
inaccurate estimations occurred on every scale, for both groups of
subjects, there is no observable pattern in the magnitude or direc
tion of the error in estimation.
For some of the scales of the Mini-Mult, the deviations from
the MMPI are of the size that the use of the abbreviated form may
not be precluded. In regard to these scales as well as the others,
the individual user of the Mini-Mult must decide the extent to which
accuracy can be sacrificed in the attempt to save time.
Since MMPI scores are not absolute, clinicians allow for cer
tain confidence limits around scores generated from the full scale
MMPI. In view of these confidence intervals, perhaps a Mini-Mult
score deviating from the MMPI by a point or two would be sufficient
ly accurate to allow for its use. On scales L, D, Hs, and Pd, the
deviations from the MMPI are within the range that the Mini-Mult
scores may be regarded as accurate approximations of the full scale
scores.
The most dramatic failure of the Mini-Mult in its ability to
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
30
predict scores on the MMPI with the current population occurred on
the "psychotic" scales of the MMPI. On these scales (Pa, Pt, Sc,
Ma), both forms of the Mini-Mult consistently underestimated the
scores obtained on the full scale MMPI. The poorest scale was Sc,
with an underestimation in Group II of more than two standard de
viations (20 points). While the underestimations on the other
"psychotic" scales were not as pronounced as on scale Sc, the dis
crepancies were severe enough to raise serious questions about the
ability of the Mini-Mult to identify "psychotic-type" pathology
within the current population.
It was felt that averaging the scores on all three forms,
across both groups, might prove productive in delineating any pat
terns in the scores that had not been brought out during the formal
data analysis. With the exception of scales L, D, and Hy, both forms
of the Mini-Mult underestimated the values obtained on the full scale
MMPI. On the three scales that were overestimations of the full
MMPI, in each case, the randomized form of the Mini-Mult produced a
smaller overestimation. Of the eight scales that were underesti
mated, on five scales the original Mini-Mult created a smaller dif
ference; while on the other three scales, the randomized form pro
duced a smaller discrepancy.
In summary, there is no readily observable pattern in regard
to the overestimation and underestimation of MMPI scores by either
the original or randomized forms of the Mini-Mult. While there are
no apparent significant differences between the two forms of the Mini-
Mult, quite a substantial difference appears in the comparison of
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
31
the two versions of the short form to the full scale MMPI. This
difference is most pronounced on the "Psychotic" end of the MMPI
where marked discrepancies were observed between the short and long
forms.
It is felt by the present investigator that at its current
stage of development, the Mini-Mult, as proposed by Kincannon
(1968) is not an adequate substitute for the MMPI. Despite the ab
ility to fairly accurately predict scores on the "neurotic" por
tion of the MMPI, the Mini-Mult is found to be sorely lacking in its
performance on the "psychotic" end of the MMPI. Since the ability
to detect "psychotic-type" ideation is one of the strong points in
the use of the MMPI, this defect of the Mini-Mult is even more
dramatically emphasized.
The current investigation seems to present data which are in
support of data cited previously (Armentrout, 1970; Armentrout and
Rouzer, 1970; Harford, et. al., 1972; and Streiner, et. al., 1973).
That is that the Mini-Mult is an ineffective substitute for the MMPI,
and can only be regarded as a gross indicator of psychopathology,
lacking the precision of the full scale MMPI.
Perhaps one method of increasing the precision and predictive
abilities of the Mini-Mult would be the addition of a gradually
increasing number of items until a form is arrived upon which satis
fies the two major criteria for an abbreviated form of the MMPI:
(1) an accurate approximation of the full MMPI, and (2) a sub
stantial savings in time.
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
LIST OF APPENDICES
Appendix A
Appendix B
Appendix C
Appendix D
Criminal Offenses and Possible Maximum Penalties with which Subjects were Charged.
Sample Data Collection Sheet.
Corresponding Item Numbers for Two Forms of the Mini-Mult and the MMPI.
Conversion Table for the Prediction of Standard Scale Raw Scores from the Mini-Mult Raw Scores.
32
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
APPENDIX A
Criminal Offenses and Possible Maximum Penalties with which Subjects were Charged
Criminal Potential # SubjectsOffense____________ Penalty_____________ Charged
Breaking and Entering 15 years 16
Larceny in a Building 4 years-$2,000 18
Entry without Breaking with Intent to CommitLarceny 5 years-$2,500 1
Larceny from a MotorVehicle 5 years-$2,500 7
Larceny over $100 5 years-$2,500 2
Auto Theft (stealing) 5 years-$2,500 0
Auto Theft (joy riding) 2 years-$l,000 1
Larceny by Conversion 5 years-$2,500 1
Receiving and ConcealingStolen Property 5 years-$2,500 5
33
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
APPENDIX B
Sample Data Collection Sheet
Subject: _______________________ Age: __________ Sex:
Group: ________________ Code: 1
2
MINI-MULT MMPI MINI-MULT
L
f
K
Hs
D
Hy
Pd
Pa
Pt
Sc
34
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
APPENDIX C
Corresponding Item Numbers for Two Forms of the Mini-Mult and the MMPI
MMPI MM-A mm-:
2 1 603 2 98 3 4010 4 .5213 5 3015 6 3118 7 5821 8 522 9 6323 10 4624 11 7130 12 131 13 3332 14 6433 15 3835 16 1638 17 6741 18 5043 19 4348 20 6251 21 5459 22 4467 23 671 24 6175 25 4886 26 22103 27 57106 28 12107 29 68109 30 2121 31 21124 32 59125 33 39129 34 25134 35 53137 36 13
MMPI MM-A mm-:
142 37 56153 38 66156 39 36157 40 4159 41 37160 42 19170 43 55178 44 65180 45 27189 46 20190 47 26192 48 34195 49 23200 50 51201 51 47209 52 17215 53 8217 54 7224 55 18225 56 15228 57 49230 58 11234 59 69238 60 14245 61 24252 62 42271 63 41272 64 28274 65 10281 66 29291 67 45296 68 70305 69 3316 70 32317 71 35
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
APPENDIX D
Conversion Table for the Prediction of Standard Scale Raw Scores from the Mini-Mult Raw Scores
Scale
L F Hs D Hy Pd Pa Pt Sc Ma
0 2 2 4 2 13
1 4 4 5 4 152 6 6 7 6 163 8 9 8 8 184 10 11 10 10 205 12 14 11 12 22
6 16 13 14 247 18 14 16 268 21 15 18 289 23 17 20 2910 25 18 22 31
11 28 20 23 3312 30 21 25 3513 33 23 27 3814 35 24 29 3915 37 26 40
16 27 4217 4418 4619 4820 50
2122232425
26
a. This estimated K scale scale score in the usual manner
10 9 5 2 2 8
11 11 7 4 4 1013 13 8 6 7 1115 14 10 9 10 1316 16 12 11 13 1518 18 14 13 15 17
19 20 15 16 18 1821 21 17 18 21 2022 23 19 20 24 2224 25 21 23 26 2425 26 22 25 29 25
27 28 24 27 32 2729 30 26 30 35 2930 32 28 32 3732 33 29 34 4033 35 37 43
35 37 39 4636 39 4838 40 5140 42 5441 57
4344464749
50
core is added to the appropriate to generate K-corrected profiles.
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
BIBLIOGRAPHY
Armentrout, J.A. Correspondence of the MMPI and Mini-Multin a college population. Journal of Clinical Psychology, 1970, 26, 49 3-495.
Armentrout, J.A., and Rouzer, D.L. Utility of the Mini-Mult with delinquents. Journal of Consulting and Clinical Psychology, 1970, _34, 450.
Bancroft, T.A. Topics in intermediate statistical methods,volume one. Ames, Iowa: Iowa University Press, 1968, p.
Carmer, S.G. and Swanson, M.R. An evaluation of ten pair-wise multiple comparison procedures by monte carlo methods. Journal of the American Statistical Association, 1973, 68, 66-74.
Edwards, A.L. Statistical methods. (2nd ed.) New York: Holt,Rinehart and Winston, Inc., 1967, p. 425.
Ferguson, R.G. A useful adjunct to the Minnesota Personality Inventory scoring and analysis. Journal of Clinical Psychology, 1946, 2_, 248-253.
Foulds, G.A., Caine, T.M., and Creasy, M.A. Aspects of extra- and intro-punitive expression in mental illness. Journal of Mental Science, 1960, 106, 599-610.
Gayton, W.F., Ozmon, K.L., and Wilson, W.T. Investigation of a written form of the Mini-Mult. Psychological Reports, 1972, 30, 275-278.
Gayton, W.F., and Wilson, W.T. Utility of the Mini-Mult in a child guidance clinic setting. Journal of Personality Assessment, 1971, _15, 569-575.
Glass, G.V. and Stanley, J.C. Statistical methods in education and psychology. Englewood Cliffs, N.J.: PrenticeHall, Inc. , 1970, p. 521.
Gough, H.G. Diagnostic patterns on the MMPI. Journal of Clinical Psychology, 1946, 2_, 23-37.
Harford, T., Lubetkin, B., and Alpert, G. Comparison of the standard MMPI and the Mini-Mult in a psychiatric outpatient clinic. Journal of Consulting and Clinical Psychology, 1972, _39_, 243-245.
37
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.
38
Hartman, G,, and Robertson, M. Comparison of the Mini-Mult and the MMPI in a community mental health agency.Proceedings of the Annual Convention of the American Psychological Association, 1972, _7, 33-34
Hathaway, S.R., and McKinley J.C. A multiphasic personality schedule (Minnesota): I. Construction of the schedule. Journal of Psychology, 1940, _10_, 249-254.
Jorgenson, C. A short form of the MMPI. Australian Journal of Psychology, 1958, 10, 341-350.
Kincannon, J.C. Prediction of the standard MMPI scale scores from 71 items: The Mini-Mult. Journal of Consultingand Clinical Psychology, 1968, _32, 319-325.
Lacks, P.B. Further investigation of the Mini-Mult. Journal of Consulting and Clinical Psychology, 1970, _35, 126-127.
MacDonald, G.L. A study of the shortened group and individual forms of the MMPI. Journal of Clinical Psychology,1952A, 8, 309-311.
MacDonald, G.L. Effect of test-retest interval and item arrangement of the shortened forms of the MMPI. Journal of Clinical Psychology, 1952B, _8, 408-410.
Newton, J.R. A comparison of studies of the Mini-Mult. Journal of Clinical Psychology, 1971, 21_, 489-490.
Olson, G.W. The Hastings short form of the group MMPI. Journal of Clinical Psychology, 1954, JL(3, 386-388.
Pulvermacher, G.P., and Bringmann, W.C. The Mini-Mult used with French-Canadian college students. Psychological Reports, 1971, 29_, 134.
Streiner, D.L., Woodward, C.A., and Goodman, J.T. Comparisons of the MMPI and Mini-Mult. Canadian Journal of Behavioral Science, 1973, 5, 76-82.
Winer, B.J. Statistical principles in experimental design.(2nd ec.) New York: McGraw-Hill, 19 71, p. 525-526.
Reproduced with permission o f the copyright owner. Further reproduction prohibited w ithout permission.