social desirability effects in male prisoners

14

Click here to load reader

Upload: carolyn-j-adams

Post on 28-Aug-2016

214 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Social desirability effects in male prisoners

JOURNAL OF RESEARCH IN PERSONALITY 23, 421-434 (1989)

Social Desirability Effects in Male Prisoners

CAROLYN J. ADAMS AND ALAN G. KRASNOFF

University of Missouri-St. Louis

Validity of the Jacobson-Kellogg Self-Description Inventory-Revised (SDI- II) as a multidimensional measure of social desirability (SD) was examined in a sample of male prisoners. Results indicated that the SDI-II may be considered as a unitary construct in this sample and the total score serves as a highly reliable and valid measure of SD influence. High SD condition subjects scored signifi- cantly higher on each SDI-II subscale than low SD condition subjects suggesting that self-presentation in socially desirable ways is latent or masked without a situation to elicit the response. Blacks scored significantly higher than whites on each SDI-II subscale. Additional findings are presented. Q 1989 Academic PWS.

Inc.

Interpretation of test scores has been a long-standing concern of de- velopers and users of structured personality inventories. The distinction between content and style summarized by Jackson and Messick (1958) highlights the problem of distortion of responses to test items through the influence of personal styles. Messick (1960) described several such styles including individual differences in tendencies to respond in a so- cially desirable manner due to the need for approval and tendencies to respond with acquiescence.

An extensive literature summarized by Bentler, Jackson, and Messick (1971) attests to the efforts of researchers to identify and control various response styles in order to isolate a subject’s uncontaminated responses to the content of interest. One of these response styles, social desirability (SD) (Crowne & Marlowe, 1960), has been conceptualized as a multi- dimensional construct. Several studies have reported results which sup- port this view (Berger, Levin, Jacobson, & Millham, 1977; Jackson & Messick, 1961; Jacobson, Kellogg, Cauce, & Slavin, 1977; Johnson, 1981; Millham & Kellogg, 1980; Paulhus, 1984; Ramanaiah, Schill, & Leung, 1977).

We gratefully acknowledge James T. Walker for his comments about the manuscript. Requests for reprints should be sent to Alan Cl. Krasnoff, University of Missouri-St. Louis, Department of Psychology, 8001 Natural Bridge Road, St. Louis, MO 63121-4499.

421

0092-6.566189 $3.00 Copyright Q 1989 by Academic press. Inc. All rights of reproduction in any form reserved

Page 2: Social desirability effects in male prisoners

422 ADAMS AND KRASNOFF

The basic purpose of this study was to assess the characteristics of a newly developed multidimensional measure of SD, the Jacobson-Kellogg Self-Description Inventory-Revised (SDI-II) (Jacobson, Brown, & Ar- iza, 1983). More specifically, this study was undertaken for the following reasons: (1) to determine the effects of circumstances designed to elicit differences in the level of SD responding on the SDI-II; (2) to determine SD differences between blacks and whites in these circumstances; (3) to compare the SD responding of male prisoners, the subjects of this study, with the normative sample of college students (Jacobson et al., 1983); (4) to ascertain how SDI-II scores were related to age, education, IQ, and criminal charges.

Description of the SDI-II

The SDI-II is a new measure of the need for social approval, the content of interest in this study, and is designed to control for the possible operation of nested acquiescence response sets (Jacobson et al., 1983). The SDI-II consists of four subscales. Each subscale with a sample item and the scoring for an SD response is as follows. Subscale 1, attribution of positive traits: I always keep my promises (T = 1). Subscale 2, attribution of negative traits: I sometimes try to dodge responsibilities (F = 1). Subscale 3, denial of positive traits: 1 am not always a good listener (F = 1). Subscale 4. denial of negative traits: I never blame others for my own mistakes (T = 1).

Two types of noncontent response style described by Bentler et al. (1971), agreement-disagreement response acquiescence, and acceptor- rejector acquiescence, are incorporated in the SDI-II. Agreement ac- quiescence is defined as individual differences in the tendency to agree by responding true (T) to test items regardless of content (a yea-sayer) on all subscales and to disagree by responding false (F) (a nay-sayer) on all subscales. Acceptance acquiescence is defined as individual dif- ferences in the tendency to respond T to items that attribute character- istics and F to items that deny such characteristics (an acceptor). Thus, the acceptor responds T to items in Subscales 1 and 2 and F to items in Subscales 3 and 4. Another person may tend to respond F to items that attribute characteristics and T to items that deny such traits (a rejector). In this case the rejector responds F to items in Subscales 1 and 2 and T to items in Subscales 3 and 4.

Visual representations of extreme response style are shown in Fig. 1. A quick estimate of SDI-II validity for a single subject can be made by comparing the subject’s profile of subscale scores with one of the profile extremes. If the highs and lows of an extreme profile are seen in the subject’s profile, it is likely the subject is responding to some degree out of an acquiescence response set rather than to item content.

Page 3: Social desirability effects in male prisoners

PRISONERS AND SOCIAL DESIRABILITY 423

Agreement (responds true to

all Items)

Disagreement (responds false

to ail Items)

Accepter (responds true to

attrlbutlon of traits, false to denial of traits)

Rejector (responds false to

0 1 2 5 4

Social deslrablilty (responds to items

presenting self In Mclally desirable rays)

0 I 2 3 4

FIG. 1. Profiles of extreme agreement, disagreement, accepter, and rejector acquies- cence response styles compared to the extreme social desirability response style.

The validity of SDI-II as a measure of SD is said to be enhanced as control is improved over acquiescence nuisance variables (Jacobson et al., 1983). SD responses for the entire scale of 128 items are keyed T or F equally often-64 items each-across agreement and disagreement acquiescence. Acceptor and rejector acquiescence are controlled in the same way. Thus, the four subscales with 32 items each are constructed with one of the attribution subscales keyed T for SD and the other keyed F. The same arrangement is provided for the denial subscales.

Jacobson et al. (1977) reported a correlation of .82 between the original Jacobson-Kellogg measure (SDI-I) and the Marlowe-Crowne Social

Page 4: Social desirability effects in male prisoners

424 ADAMS AND KRASNOFF

Desirability Scale (MCS). There was no item duplication between the two measures.

Research on SD in Prisoners

A review of SD studies using the MCS with prisoners produced con- flicting findings. Some studies reported that prisoners tended to score higher than college students. These findings suggest that prisoners com- pensate for their imprisonment and regain status by presenting them- selves in socially desirable ways (Crowne & Marlowe, 1964; Fisher & Parsons, I962). In contrast to these findings, Pettijohn (1977) reported no significant differences between prisoner performance on the MCS and college student performance. A confound was evident in Pettijohn’s study-the prisoners were enrolled in college classes. This led Pettijohn to hypothesize that education and/or IQ affected SD scores.

Fisher (1967) had reported earlier that intelligence and educational achievement were significantly negatively correlated with the MCS and, in addition, that criminal charge was unrelated to MCS scores. He also reported that race of subjects significantly influenced scores on the MCS with Mexican-Americans and blacks scoring higher than whites with IQ controlled. Klassen, Hornstra, and Anderson (197.5) also reported blacks scoring higher than whites with both age and education controlled. These researchers also reported a positive correlation of age with MCS scores.

SD Conditions and Hypotheses

In order to achieve the previously noted aims of this study, two samples of male prisoners were identified. A sample of residents from a jail who were eligible for a work release program were classified as high SD condition subjects (WR). Work release ineligible residents (WRI) were designated as low SD condition subjects. Residents had to be sentenced to a jail term to be classified as eligible for work release whereas residents who were still awaiting disposition of their cases were designated as ineligible for work release.

We hypothesized that WR subjects (high SD condition) would have significantly higher SDI-II scores than WRI subjects (low SD condition) with effects due to race controlled. We reasoned that the condition of competing for a vacancy in the WR program and the assumption that prisoners would perceive the testing as a basis for disqualification from the program would increase their need for approval demonstrated by increased SD responses. The confirmation of this hypothesis represents a test of the construct validity of the SDI-II as a measure of approval motivation. We also predicted that blacks would obtain significantly higher SDI-II scores than whites in accordance with previous research. We hypothesized that charge would be unrelated to SDI-II scores, but

Page 5: Social desirability effects in male prisoners

PRISONERS AND SOCIAL DESIRABILITY 425

that age, education, and IQ would have effects, with older subjects scoring higher than younger subjects and with more educated subjects and subjects with higher IQ obtaining lower SDI-II scores.

Subjects

METHOD

The final jail sample consisted of 90 male subjects housed in small groups of a county correctional institution. The high SD condition consisted of 23 white and 22 black inmates consecutively referred by the social work staff. In order to be referred to the WR group. residents had to be sentenced to a jail term in the county detention facilities with a minimum of 3 months left to serve. An estimated 90% of eligible residents are transferred to the work release program. The contrast low SD condition consisted of 22 white and 23 black comparably housed inmates. These residents were ineligible for the work release program and constituted the WRI group because their cases had not yet been resolved in court. Subjects in the WRI group were arbitrarily referred by the social work staff from the available pool of preadjudicated inmates. These subjects were recruited in order to be comparable in number, type of housing. and racial composition to the WR group. Maximum security inmates were not included as potential candidates for this study.

Approximately IO subjects from the WR group were excluded because of omitted re- sponses, illiteracy, or copying from others. Of the WRI group approximately 10 subjects were excluded for the same reasons. Some 35 additional prospective WRI subjects refused to participate. There were no refusals from the WR group. Although no records were kept characterizing subjects who refused to participate in this study. we have the definite impression that the higher refusal rate for the WRI group reflects less motivation for approval as compared to the WR group. Being asked to participate in a research project such as this one establishes an inherent demand on prospective subjects to comply. To decline requires some degree of noncompliance. In our opinion, the differences to be reported between WR and WRI groups would have been larger had the refusers actually participated.

It should be noted that no independent method of verification was used to evaluate the effort to induce differential social desirability effects in the WR and WRI groups. As noted by an anonymous reviewer, such a “manipulation” check would have strengthened the methodology of this study.

Measures All subjects completed the SDI-II and a demographic questionnaire which requested

each subject’s name, age, race, number of years of education completed, and whether the General Educational Development Program (GED) had been achieved. Subjects’ charges at the time of arrest were obtained from jail files and recoded into three categories: I = property offense; 2 = drug/alcohol offense; 3 = crime against persons. In addition to these measures, the WR sample was administered the Revised Beta Examination-Second Edition (Beta-II) (Kellogg & Morton, 1978) along with a more extensive battery of psy- chological tests.

Procedure Testing was conducted in small groups of from two to nine subjects. After receiving a

description of the study, each subject signed a consent form stating that he understood participation was voluntary and that his responses were confidential and would not in any way influence his legal case. The participants then completed the measures.

Page 6: Social desirability effects in male prisoners

426 ADAMS AND KRASNOFF

RESULTS AND DISCUSSION

Means and standard deviations of each SDI-II subscale and total score for the college sample of 108 subjects (Jacobson et al., 1983), the jail sample of 90 subjects, and eight subgroups of the jail sample are presented in Table 1. Profiles of SDI-II subscale means for WR and WRI, black and white subgroups are shown in Fig. 2. As the figure indicates, the high SD condition WR subjects on the average obtained higher SDI-II scores than did low SD condition WRI subjects; blacks scored higher on the average than did whites.

The effects of these two variables on SDI-II scores were of major interest in this study. A 2 x 2 factorial with a multivariate analysis of variance (MANOVA) was used to test effects of group and race on the four SDI-II subscales. Results showed the high SD WR group scored significantly higher than the low SD WRI group on each subscale at the .02 level or beyond. The analysis also showed that blacks scored sig- nificantly higher than whites on each subscale at the .04 level or beyond. There was no significant interaction between SD group and race, all F ratios cl. A summary of the MANOVA analysis is presented in Table 2 with corresponding univariate F ratios. The effects of other variables on SDI-II subscales were also tested with MANOVAS. For the total jail sample of 90 subjects, no significant effects on SDI-11 scores were found for age, education, or criminal charge.

For the total jail sample the effects of SD condition and race on SDI- II total scores were tested using a 2 x 2 factorial analysis of variance (ANOVA). Results showed the high SD group scored significantly higher than the low SD group on SDI-II total scores, F(1, 86) = 16.10, p < .OOOl; and blacks scored significantly higher than whites, F(1, 86) = 10.00, p < .003. Again, there was no interaction between SD condition and race.

Effects of age, education, and criminal charge on SDI-11 total scores for WR and WRI. black and white subgroups were also tested using ANOVAS. Correlations of SDI-II total scores with these variables of interest were also calculated for the total jail sample and these four subgroups.

The mean age of the jail sample of 90 subjects was 25.03 years with a standard deviation of 7.57. No significant effects of age on SDI-II total scores were found for WR and WRI, black and white subgroups. For the 45 whites, age correlated .32 (p < .05) with SDI-11 total scores. There were no other significant correlations between age and SDI-II for other jail subgroups.

The mean number of years of education completed for the total sample was 10.9 with a standard deviation of 1.6. Thirteen percent of the 90 subjects completed high school and 11% completed one or more addi-

Page 7: Social desirability effects in male prisoners

TABL

E 1

MEAN

S AN

D ST

ANDA

RD

DEVI

ATIO

NS

OF

SDI-I

I SU

BSCA

LES

AND

TOTA

L SC

ORE

FOR

COLL

EGE

STUD

ENTS

O AN

D PR

ISON

ERS

Sam

ple

N

Stud

ents

” 10

8 Pr

isone

rs

90

WR

45

W

RI

45

Blac

ks

45

Whi

tes

45

WR

Bla

cks

22

WR

Whi

tes

23

WR

I Bl

acks

23

W

RI

Whi

tes

22

1 At

tribu

tion

of p

ositi

ve

traits

M SD

17.8

5 6.

98

18.1

1 8.

22

20.7

3 7.

67

15.4

9 7.

98

20.4

4 7.

80

15.7

8 8.

04

22.6

8 7.

54

18.8

7 7.

49

18.3

0 7.

59

12.5

5 7.

43

Subs

cale

s

2 3

4 At

tribu

tion

Deni

al

of

Deni

al

of

of n

egat

ive

posi

tive

nega

tive

traits

tra

its

traits

M SD

M

SD

M SD

SDI-I

I to

tal

scor

e

M SD

zi

16.1

1 7.

16

17.2

7 5.

79

13.2

4 7.

06

64.6

7 15

.81

8.00

15

.87

6.97

14

.24

7.39

64

.03

18.8

9 7.

47

17.4

9 7.

17

17.3

1 7.

02

74.4

2 12

.73

7.37

14

.24

6.44

11

.18

6.49

53

.64

17.6

9 7.

72

18.2

9 6.

10

15.7

1 7.

21

72.1

3 13

.93

7.92

13

.44

7.00

12

.78

7.37

55

.93

20.6

8 7.

06

19.5

9 6.

64

19.2

3 7.

19

82.1

8 17

.17

7.60

15

.48

7.22

15

.48

6.49

67

.00

14.8

3 7.

35

17.0

4 5.

40

12.3

5 5.

51

62.5

2 10

.55

6.89

11

.32

6.24

9.

95

7.31

44

.36

24.0

7 g

27.9

9 E

26.6

9 $

25.5

4 25

.84

8

27.9

9 E

25.6

2 & P

26.0

7 22

.60

s

25.5

9

” D

ata

for

colle

ge

stud

ents

fro

m

“A

Rev

ised

M

ultid

imen

siona

l So

cial

Des

irabi

lity

Inve

ntor

y”

by

L.

I. Ja

cobs

on,

R.

F.

Brow

n,

and

M.

J Ar

iza,

1983

, Bu

ktin

of

the

Psyc

hono

mic

Socie

ty,

21,

p. 3

92.

Cop

yrig

ht

1983

by

the

Psyc

hono

mic

So

ciet

y,

Inc.

R

eprin

ted

by p

erm

issio

n.

Page 8: Social desirability effects in male prisoners

428 ADAMS AND KRASNOFF

WR 0 4

WRI 0 0

Blacks .- - - se

Whites &- - 4

I 2 3 4

SDI-II Subscales

Note A z= 45 ior each subgroup.

FIG. 2. Profiles of SDI-II subscale means of WR and WRI, black and white jail subgroups.

tional years of education. Whether a subject had passed the GED cri- terion was used as a second measure of educational achievement. In addition to those with a high school education or more, 22% of the sample reached the GED criterion. The remaining 54% completed 10.1 years of schooling on the average. No significant effects on SDI-II total scores of years of education completed or meeting the GED criterion were found for WR, WRI, or white subgroups. Effects of meeting the GED criterion on SDI-II total scores approached significance for blacks, F(1, 43) = 3.99, p < .0521. This result indicates that blacks who had reached the GED criterion tended to obtain lower SDI-II total scores than did blacks who had not met the GED criterion. There were no significant correlations between either measure of education and SDI-II scores for the total jail sample or for any subgroup.

Page 9: Social desirability effects in male prisoners

PRISONERS AND SOCIAL DESIRABILITY 429

TABLE 2 MANOVA ANALYSIS OF EFFECTS OF SOCIAL DESIRABILITY (SD) CONDITION AND RACE ON

SDI-II SCORES WITH CORRESPONDING UNIVARIATE F RATIOS

Source of variance df, d!> F ratio Significance

level

SD condition 1 Attribution of

positive traits 2 Attribution of

negative traits 3 Denial of

positive traits 4 Denial of

negative traits

Race 1 Attribution of

positive traits 2 Attribution of

negative traits 3 Denial of

positive traits 4 Denial of

negative traits

I

I

4

I

I

83

89

89

89

89

83

89

89

89

89

5.90

10.96 .0014

16.29

5.77 .0184

19.15

3.90

9.13

6.52

13.26

4.80 .0312

.OOOl

.0059

.0033

.0125

.0005

Criminal offense categories were property offense, drug or alcohol offense, or crime against persons. Significant effects of criminal charge on SDI-II total scores were found for the white subgroup of 45 subjects only, F(2, 42) = 4.28, p < .02. Differences between means were tested using Scheffe’s procedure controlling for Type I error rate. Results showed that subjects with a criminal charge in violation of drug and/or alcohol statutes scored significantly higher than subjects with a criminal charge against property. Subjects with a criminal charge against persons had intermediate nonsignificant scores. There were no significant cor- relations found between criminal charge and SDI-II total scores for the total jail sample.

The mean Beta-II IQ for the 45 WR subjects who completed the measure was 90.0 with a standard deviation of 12.6. Blacks obtained a mean IQ of 85 and completed 10.9 years of education on the average. Whites obtained a mean IQ of 95 on the Beta-II and averaged 10.8 years of schooling. All but one subject obtained an IQ in the average range or lower. No significant effects of IQ on SDI-II total scores were ob- tained. There were no significant correlations of IQ with SDI-II total scores for the WR, black, or white subgroups. The failure to confirm the inverse relationship between IQ and SD may have been due to the restricted range of intellectual ability found in this sample of prisoners.

Page 10: Social desirability effects in male prisoners

430 ADAMS AND KRASNOFF

Two measures of internal consistency for the SDI-II total score were calculated. To check whether the 128-item SDI-II was too long for sub- jects to attend to in one sitting and therefore possibly subject to random responses, especially on the second half, a correlation between the first half and the second half was made with the Spearman-Brown correction applied. The measure yielded a split-half reliability of .95. The Kuder- Richardson Formula 20 yielded a reliability of .97. These results represent a substantial degree of internal consistency and suggest that a 64-item version would yield results substantially similar to the SDI-II.

The means and standard deviations for the college students (Jacobson et al., 1983) and for the jail sample for each subscale and for the SDI- II total score are presented in Table 1. There are no significant mean differences between these groups, p > .lO in each t test. Given that this jail sample reported completing 10.9 years of education on the average, with the group mean age of 25 and a mean Beta-II IQ of 90, these results appear to challenge previous findings that subjects with more education obtain lower SD scores than do subjects with less education (Fisher, 1967); and that prisoners obtain higher SD scores than do college students (Crowne & Marlowe, 1964; Fisher & Parsons, 1962). However. inspec- tion of the mean SDI-II scores for black and white prisoners and for WR and WRI prisoners in comparison to the mean scores for the student sample in Table I reveals cancelling effects associated with differences due to race and to situational conditions.

Table 3 presents the intercorrelations among the SDI-II subscales and total score for college students (Jacobson et al., 1983), all prisoners, WR, WRI, black, and white subgroups. The total prison sample had consistently higher correlations than the students on all 6 subscale com- parisons. For the various inmate subgroups, prisoner correlations ex- ceeded those of students on 21 of 24 subscale comparisons. Tests of the significance of the differences between student and total prisoner subscale correlations yield two significant findings. The prisoners had significantly higher correlations than the students for subscales 1 and 3, t(196) = 2.08, p < .05, and for subscales 2 and 3, t(196) = 4.20, p < .OOl. The more moderate intercorrelations of subscaie scores for college students were interpreted by Jacobson et al. (1983) as support for the conception of SD as a multidimensional construct with attribution and denial aspects. The same logic along with these empirical results suggests that prisoners tend to be more consistent in reaction to SD regardless of the attribution and denial dimensions. Perhaps this is a reflection of their externally controlled environment associated with incarceration.

Using Bartlett’s test (cited in Marascuilo & Levin, 1983) for the identity of independent variance-covariance matrices, which also implicitly tests for the identity of correlation matrices, college and jail matrices were found to be heterogeneous, x*(10, N = 198) = 31.30. p < .OOl.

Page 11: Social desirability effects in male prisoners

PRISONERS AND SOCiAL DESIRABILITY 431

TABLE 3 INTERCORRELATIONS AMONG SD141 SUEXALES AND TOTAL SCORE FOR COLLEGE STUDENTS,

ALL PRISONERS, WR, WRI, BLACK, AND WHITE SUBGROUPS

Subscale” intercorrelations Studentsb All’ WR“ WRId Blackd Whited

1-2 l-3 1-4 2-3 2-4 3-4 I-Total 2-Total 3-Total 4-Total

.67 .77 .72 .75 .72 .78

.57 74 .68 .76 .71 .71

.78 .82 .80 .79 .75 .87

.62 .87 .90 .85 .87 .88

.77 .82 .84 .73 .76 ‘86 A0 .68 .69 .62 .62 .70 .88 .91 .88 .92 .89 .92 90 .95 .95 .92 .93 .95 .74 .89 .90 .89 .88 .89 .91 .91 .91 .87 .88 .93

L? Subscale 1, attribution of positive traits; subscale 2, attribution of negative traits: subscale 3, denial of positive traits; subscale 4, denial of negative traits.

b College students; N = 108; all ps < .OOl; data from “A Revised Multidimensional Social Desirability Inventory” by L. I. Jacobson, R. F. Brown, and M. J. Ariza, 1983, Bulletin of the Psychonomic Society, 21, p. 392. Copyright by the Psychonomic Society, Inc. Reprinted by permission.

’ All prisoners; N = 90; all ps < .OOOl. d WR, WRI, Black, White subgroups; n = 45 for each subgroup; all ps < .OOOl.

SDI-II subscale intercorrelations for the WR and WRI subgroups, as well as for black and white subgroups, are also found in Table 3. Again using Bartlett’s test, matrices of intercorrelations among the SDI-II subs- tales for WR and WRI subgroups and black and white subgroups were tested for identity and found to be homogeneous, x*(10, N = 90) = 11.31, p > .05 and x2(10, iV = 90) = 9.02, p > .05, respectively. These results suggest that social desirability, as measured by the SDI-II in the jail sample, is a unitary construct. Results from the MANOVA analysis of the effects of group and race also support this interpretation. The uniformly significant effects of group and race on the four outcome measures, Subscales 1 through 4, verify empirically the unidimensionality of the set of dependent variables.

Other evidence from the data supports this interpretation of the uni- dimensionality of the SD construct. There was little to gain and, there- fore, less pull for socially desirable responses from the 45 subjects in the WRI subgroup. Twenty-one of these subjects on at least one subscale obtained a score above or below another subscale of one standard de- viation or more. This high or low in a subject’s profile was considered an indicator of an active agreement or acceptance acquiescence response set. These 21 profiles were characterized by peaks and valleys and, with

Page 12: Social desirability effects in male prisoners

432 ADAMS AND KRASNOFF

this greater interscale variability, produced a group matrix with lower intercorrelations among subscales than did flatter profiles. Only one among these 21 subjects whose profiles showed a response set pattern also obtained an elevation of one standard deviation or more on the SDI- II total score. This total score elevation was considered an indicator of an active SD response set. Two other subjects from this same special group had a SDI-II total score elevation without showing a response style pattern.

The MANOVA analysis testing the effects of SD condition and race showed that the 45 subjects in the higher SD condition gave significantly more SD responses than the lower SD group. This SD activated WR group produced 17 profiles showing response style patterns. There were no significant differences in the number of response style profiles or in SDI-II scores between the special WRI group (n = 21) and the special WR group (n = 17). Only one of the 17 WR subjects whose profiles showed a response set also had a SDI-II total score elevation. Eleven other subjects in this WR special group had a total score elevation of at least one standard deviation without a response set profile. These frequencies indicate that the two variables, response style pattern in- dicating acquiescence response sets and SDI-II total score elevation measuring SD response sets, are nearly mutually exclusive.

Chi-square analysis of the frequencies with a continuity adjustment indicated that the two variables, response style pattern and total score elevation, had a significant inverse relationship, ~‘(1, N = 90) = 4.82, p < .03, and that there were indeed significantly more subjects having SDI-II total score elevations without showing a response style profile than subjects having total score elevations with a response style pattern, ~‘(1, n = 15) = 6.67, p < .Ol.

These results indicated that an SD response set is activated especially in subjects who do not present an acquiescence response style pattern. As the MANOVA showed, the SD response set is more evident in situations which call for SD self-presentation. When such a situation occurs and the SD set is activated, SDI-II scores are higher, the influence of acquiescence response sets decreases, profiles are flatter due to less interscale variability, and group matrices of intercorrelations among sub- scales show higher correlations than those of groups influenced by ac- quiescence response styles.

Although means of SDI-II measures from the jail and college samples were very close, the college sample could possibly have obtained higher means had subjects been SD activated. It seems reasonable to compare the low SD condition WRI subjects with an SDI-II total score mean of 53.64 with the non-SD activated college sample’s mean of 64.67. These data clearly disagree with previous findings that prisoners score higher

Page 13: Social desirability effects in male prisoners

PRISONERS AND SOCIAL DESIRABILITY 433

than college students (Crowne & Marlowe, 1964; Fisher & Parsons, 1962).

Overall, the findings of the study challenge the concept of social de- sirability as a multidimensional construct with attribution and denial aspects (Jacobson et al., 1983) at least for inmates. The evidence suggests that the personality trait of socially desirable self-presentation is latent or masked without a situation pulling for the SD response, a conclusion consistent with the work of Mischel (1968). In a low SD condition the SDI-II subscale profiles identify malingering and/or acquiescence re- sponse styles, but their utility as measures of SD is questionable. In the absence of these response styles, the SDI-II total score is a reliable and valid measure of SD influence, especially in a high SD condition.

The significant effect of race on SDI-II scores with blacks scoring higher than whites raises the general question of racial differences in inmates’ adaptation in a jail setting. Little information is available in the psychological literature to explain these findings other than that they agree with previous reports (Fisher, 1967; Klassen et al., 1975). One may speculate that black prisoners, a minority group in a restrictive institutional setting which limits inmates’ rights and freedom, are rela- tively more motivated than white prisoners to seek approval from others in the hope of gaining better treatment or avoiding discrimination. Further studies are needed to clarify this matter.

REFERENCES

Bentler, P. M., Jackson, D. N., & Messick, S. (1971). Identification of content and style: A two-dimensional interpretation of acquiescence. Psychological Bulletin, 76, 186- 204.

Berger, S. E., Levin, P., Jacobson, L. I., & Millham, J. (1977). Gain approval or avoid disapproval: Comparison of motive strengths in high need for approval scorers. Journal of Personality, 45, 458-468.

Crowne, D. P., & Marlowe, D. (1960). A new scale of social desirability independent of psychopathology. Journal of Consulting Psychology, 24, 349-354.

Crowne, D. P., & Marlowe, D. (1964). The approval motive. New York: Wiley. Fisher, G. (1967). The performance of male prisoners on the Marlowe-Crowne Social

Desirability Scale: II. Differences as a function of race and crime. Journal of Clinical Psychology, 23, 473-475.

Fisher, G. M., &Parsons, T. H. (1%2). The performance of male prisoners on the Marlowe- Crowne Social Desirability Scale. Journal of Clinical Psychology, 18, 140-141.

Jackson, D. N., & Messick, S. (1958). Content and style in personality assessment. Psy- chological Bulletin, 55, 243-252.

Jackson, D. N., & Messick, S. (1961). Acquiescence and desirability as response deter- minants on the MMPI. Educational and Psychological Measurement, 21, 771-790.

Jacobson, L. I., Brown, R. F., & Ariza, M. J. (1983). A revised multidimensional social desirability inventory. Bulletin of the Psychonomic Society, 21, 391-392.

Jacobson, L. I., Kellogg, R. W., Cauce, A. M., & Slavin, R. S. (1977). A multidimensional social desirability inventory. Bulletin of the Psychonomic Society, 9, 109-l 10.

Johnson, J. A. (1981). The “self-disclosure” and “self-presentation” views of item response

Page 14: Social desirability effects in male prisoners

434 ADAMS AND KRASNOFF

dynamics and personality scale validity. Journal of Personal& and Social Psychology, 40, 761-769.

Kellogg, C. E., & Morton, N. W. (1978). Manual: Revised Beta Examination (Beta-II) (2nd ed.). New York: The Psychological Corp.

Klassen, D., Homstra, R. K., & Anderson, P. B. (1975). Influence of social desirability on symptom and mood reporting in a community survey. Journal of Consulting and Clinical Psychology, 43, 448-452.

Marascuilo, L. A., & Levitt, J. R. (1983). Matrix algebra and multiple regression theory. In Multivariate statistics in the social sciences (pp. 141-176). Monterey. CA: Brooks/Cole Publishing Co.

Messick, S. (1960). Dimensions of social desirability. Journal of Consulting Psychology, 24, 279-287.

Millham, J., & Kellogg, R. W. (1980). Need for social approval: Impression management or self-deception? Journal of Research in Personality, 14, 445-457.

Mischel, W. (1%8). Personality and assessment. New York: Wiley. Paulhus, D. L. (1984). Two-component models of socially desirable responding. Journal

of Personality and Social Psychology, 16, 598-609. Pettijohn, T. F. (1977). Social desirability among male prisoners and college students.

Psychological Reports, 41, 110. Ramanaiah, N. V., Schill, T., & Leung, L. S. (1977). A test of the hypothesis about the

two-dimensional nature of the Marlowe-Crowne Social Desirability Scale. Journal of Research in Personality, 11, 251-259.