This article was downloaded by: [University of Leeds] On: 02 November 2014, At: 07:55 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Journal of Personality Assessment Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/hjpa20
Two Promising Shame and Guilt Scales: A Construct Validity Comparison David H. Harder & Alysa Zalma Published online: 22 Jun 2011.
To cite this article: David H. Harder & Alysa Zalma (1990) Two Promising Shame and Guilt Scales: A Construct Validity Comparison, Journal of Personality Assessment, 55:3-4, 729-745, DOI: 10.1080/00223891.1990.9674108 To link to this article: http://dx.doi.org/10.1080/00223891.1990.9674108
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions
JOURNAL OF PERSONALITY ASSESSMENT, 1990, 55(3&4), 729-745 Copyright o 1990, Lawrence Erlbaum Associates, Inc.
Two Promising Shame and Guilt Scales: A Construct Validity Comparison David H. Harder and Alysa Zalrna Downloaded by [University of Leeds] at 07:55 02 November 2014
Tufts University
This study compared the validity of two promising measures of shame and guilt proneness: revisions of the Harder Personal Feelings Questionnaire (PFQ2;Harder & Lewis, 1987) and the Hoblitzelle Adapted Shame and Guilt Scale (ASGS; Hoblitzelle, 1982). Internal consistency, test-retest stability, factor structure, and construct validity with convergent and discriminant personality dimensions were examined for both scales. In addition to the shame and guilt measures, 63 (37 male, 26 female) mostly freshman college students completed a randomly ordered battery of personality scales theoretically relevant to shame and guilt proneness. Results support the reliability and shame/guilt factor structure of each scale. ASGS Shame correlations appeared marginally more valid with 11 external construct variables than PFQ2 Shame, whereas PFQ2 Guilt was clearly more valid than its correspondingASGS subscale. New, potentially improved scales were constructed from the factor analyses and from item analyses. However, the resulting scales did not show improved validity.
T h e development of valid scales for the measurement of proneness t o shame and guilt is important for clinically relevant assessments and for the eventual understanding of personality differences. These dysphoric affect states are often linked with difficulties in emotional self-regulation and the etiology of psychopathology (Buss, 1980; Freud, 1905/1953, 1966; Lewis, 1971; Mayman, 1974; Wright, O'Leary, & Balkin, 1989). Unfortunately, the task of measuring them has proven t o be quite difficult (Harder & Lewis, 1987). As distinct emotional states that are not always easy to differentiate clinically or experientially (Levin, 1967, 1971; Lewis, 1971; Mayman, 1974) and that are sometimes defended against by conscious selfreports (Harder & Lewis, 1987), shame and guilt have presented formidable assessment problems from the outset.
Downloaded by [University of Leeds] at 07:55 02 November 2014
730
HARDER AND ZALMA
Previous research (Harder & Lewis, 1987)comparing four promising measures for each affect disposition found that the Gottschalk and Gleser (1969; Gottschalk, Winget, & Gleser, 1969) Shame and Guilt Anxiety scales have insufficient test-retest reliability, whereas two other scales for each emotion, the Binder-Harder Early Memories Shame Proneness Scale (Binder, 1970; Harder, 1984), the Mosher Total Guilt Scale (Mosher, 1966, 1968), and the Beall Situational Upset Scale (Smith, 1972) for both shame and guilt failed to demonstrate convergent and discriminant validity. Harder's Personal Feelings Questionnaire (PFQ, Harder & Lewis, 1987) for both shame and guilt proneness were the only scales emerging from this earlier study with satisfactory test-retest stability and substantial indications of construct validity, despite the presence of few items on each scale. In other research, Hoblitzelle (1982) developed a measure called the Adapted Shame and Guilt Scale (ASGS), which was adapted from an earlier adjective endorsement instrument used by Gioella (1981). The ASGS showed high internal consistency and a two-factor structure largely consistent with expectations that the scale items either reflected shame or guilt dispositions; but in preliminary attempts to establish construct validity, the guilt subscale did not ~ e r f o r mwell (Hoblitselle, 1982). In addition, the test-retest reliability of the ASGS was not investigated, leaving the scale's potential validity as a trait measure in doubt. The aim of our study was to investigate comparatively the abilities of the PFQ and the ASGS to assess shame and guilt proneness. To accomplish this goal, each scale was examined for internal consistency, test-retest stability, factor structure, and a pattern of convergent and discriminant correlations with other personality constructs consistent with theoretical expectations for valid measures. The validation constructs employed were depression, self-derogation, social anxiety, shyness, narcissism, public self-consciousness, private self-consciousness, locus of control, intellectualization defense, social desirability, and Machiavellianism. Clinical and empirical literature (Crouppen, 1977; Harder & Lewis, 1987; Lewis, 1971, 1986; Smith, 1972; Wright et al., 1989) has frequently linked depression with shame and guilt, but more strongly with shame. Hence, positive relationships between valid scales for both dysphoric affects and depression were expected; those with shame were stronger. Self-derogation, representing the negative side of the self-esteem coin, was expected to relate to valid shame and guilt measures in a pattern similar to that shown for depression. Because shame experiences ~roducea more global self-devaluation than guilt (Lewis, 1971; Tangney, 1989), shame was expected to show the stronger relationship. Buss (1980) described a theoretical connection among social anxiety, a tendency to avoid interaction with others, and shame proneness. This social avoidance
Downloaded by [University of Leeds] at 07:55 02 November 2014
tendency is typically accompanied by internal experiences of shyness (Pilkonis, 1977), which has been linked often to shame proneness by clinical theoreticians (Levin, 1967, 1971; Lewis, 1971; Mayman, 1974), and by self-consciousness about one's appearance to others. Thus shame was expected to show associaltions with social anxiety, shyness, and public self-consciousness, whereas guilt was not. Because narcissism has been connected clinically and theoretically with shame vulnerability (Kernberg, 1970; Kohut, 1971; Morrison, 1983) and because the type of narcissism that is successful in defending self-esteem seems to be the variant assessed by personality measures (Harder, 1984; Harder &Lewis, 19871, trait narcissism was predicted to correlate inversely with shame proneness and not to relate significantly to guilt. Because guilt experience has been described as having a more internal locus of origin and evaluation than shame experience (Lewis, 1971), private selfconsciousness and internal locus of control were expected to correlate with guilt proclivities but not with shame. In addition, defense styles favored by guiltoriented individuals include a large share of intellecrualization strategies (Lewis, 1971, 1978), so that the prediction was made that guilt, but not shame, would show associations with a measure of that defense. Even though one might theoretically expect shame to correlate inversely with intellectualization, this was not expected for our sample, because most college students are adept at using intellectualizationl as one of many possible psychological defensive strategies. Machiavellianism (Christie & Geis, 1970) was included as a discriminant validity construct for both shame and guilt, because it was not hypothesized to correlate with either variable. These expectations for valid shame and guilt instruments are summarized schematically in Table 1.
TABLE 1 Predicted Constnuct Validity Relationships for Shame and Guilt Scales -
Construct Validity Variable
-
Guilt
-
++ ++ + ++
Beck depression Self-derogation Social Anxiety Shyness Narcissism Public self-consciousness Private self-consciousness (External) Locus of control Intellectualization Social desirability Machiavellianism
+
-
Shame
++
Note. = positive significant correlation. = high positive significant correlation. negative significanr correlation. o = no significance predicted.
4-ICl
C)
-
=
732
HARDER AND ZALMA
METHOD Subjects
Downloaded by [University of Leeds] at 07:55 02 November 2014
Sixty-three college students at a private liberal arts institution, 90% of them freshmen, volunteered as subjects, to fulfill a course requirement in either an introductory or a statistics psychology course. Thirty-seven of the subjects were male, and 26 were female. The age of the sample ranged from 17 to 22, with a mean of 18.46. Subjects were mostly from urban and suburban upper middleclass backgrounds, and most of them maintained slightly better than a B average in school after one term. Families of origin averaged two siblings. Eleven of the 63 subjects reported their parents were divorced.
Procedure The PFQ (Harder & Lewis, 1987)and ASGS (Hoblitzelle, 1982) versions used in our study were slightly modified from their earlier forms. The PFQ was doubled in length by the addition of three new items and the transposition of original items composed of multiple affect descriptors into separate items (one for each descriptor). The entirely new items were: for shame, self-consciousness and feeling disgusting to others, and for guilt, feeling you deserve criticism for what you did. This procedure produced a 10-itemShame subscale and a six-itemGuilt subscale, rated by the subject on a 4-point scale ranging from you never experience the feeling (0) to you experience the feeling continuously or almost continuously (4). (For a list of the items see Table 3.) This scale revision was called the Personal Feelings Questionnaire-2 (PFQZ). The ASGS was modified by eliminating those six adjectives of the original 30 that failed to show loadings of .40 or greater in Hoblitzelle's (1982) factor analyses of the scale with two different samples of college students. The excluded items were: condemned, wrong, chided, exposed, unconscionable, and criminal. The remaining scale items are listed in Table 4. The revised versions of the Shame and Guilt scales were included with other personality instruments in randomly ordered packets given to subjects by a research assistant. The scales were handed out in manila envelopes, with each enclosure marked only by a code number to preserve anonymity of responses. The confidentialityensured in this way was emphasized by the research assistant in order to maximize honesty of reporting. Subjects completed the scales in rooms that could seat up to 20 subjects at one time. The research assistant remained in the room to clarify scale instructions when there were questions. It took approximately 1 hr to complete the ~ a c k eof t materials. The packet included a demographic information form, the lengthened, PFQ2, the shortened ASGS, the Beck Depression Inventory (Beck, 1967), the Rotter Locus of Control Scale (Rotter, 1966),the Marlowe-Crowne Social Desirability Scale (Crowne & Marlowe, 1960, 1964), the Haan Intellectualization Defense
SHAME AND GUILT SCALES
733
Scale (Haan, 1967; Morrissey, 1977), three Fenigstein, Scheier, and Buss (1975) scales-the Social Anxiety Scale, the Private Self-Consciousness Scale, and the Public Self-Consciousness Scale, the Kaplan Self-Derogation Scale (Kaplan, 1975; Kaplan & Pokorny, 1969), the Narcissistic Personality Inventory (Raskin & Hall, 1979, 1981), the Zimbardo Stanford Shyness Inventory (Zimbardo, 1977; Harder & Lewis, 1987), and the Machiavellianism Scale-Revised (Christie & Geis, 1970). The means and standard deviations for these variables are presented in Table 2.
Downloaded by [University of Leeds] at 07:55 02 November 2014
RESULTS Before testing the construct validity predictions (Table 1) made for relationshi~x between the Shame and Guilt scales and other constructs, the internal consrstency, test-retest stabilxt~(Table 3), and factor structure (Tables 4 and 5) of the PFQ2 and ASGS were investigated. A satisfactory level of internal consistency for both shame and both guilt measures was ascertained, but the ASCiS subscales were superior in this regard. Cronbach's alpha for ASGS Shame was .83; for ASGS Guilt, .89; for PFQ2 Shame, .78; and for PFQ2 Guilt, .72. Two-week test-retest reliability (obtained from a subset of 27 subjects) for ASCS Shame was .93; for ASGS Guilt, -95;for PFQ2 Shame, .91; and for PFQ2 Guilt, .85 -all quite high. If the PFQ2 and ASGS are valid, principal components factor analysis of the two scales was expected to reveal a factor structure consisting predominantly of two factors, one of which would be defined by the shame items and the other by TABLE 2 Means and Standard Deviations of Study Variables - -
Variable PFQ2 shame ASGS shame PFQ2 guilt ASGS guilt Beck depression Self-derogration Social anxiety Shyness Narcissism Public self-consciousness Private self-consciousness (External) Locus of control Intellectualization Social desirability Machiavellianism
M 16.13 30.18 9.76 28.44 27.37 11.03 16.70 29.25 9.18 25.02 31.57 10.37 9.78 12.43 104.45
~
SD
4.51 6.75 3.11 8.17 5.19 6.40 5.92 8.53 4.72 6..81 7..34 3.53 2.48 5.05 8..61
734
HARDER AND ZALMA TABLE 3 Reliability of the ASGS and PFQ2
Affect Subscale
Cronbach's Alpha
Test-Retest Stability
PFQ2 Shame PFQ2 Guilt ASGS Shame ASGS Guilt
.78 .72 .83 .89
.91 35 .93 .95
Downloaded by [University of Leeds] at 07:55 02 November 2014
TABLE 4 Varimax Rotation of the Two-Factor Principal Components Factor Analysis of the PFQ2 (Loadings .40 and up) PFQ2 Item Embarrassed Mild guilt Feeling ridiculous Worry about hurting or injuring someone Self-consciousness Feeling humiliated Intense guilt Feeling "stupid" Regret Feeling "childish" Feeling helpless, paralyzed Feelings of blushing Feeling you deserve criticism for what you did Feeling laughable Feeling disgusting to others Remorse
Original Affect Scale for Item
Factor I -Guilt (29.0% Variance)
Factor 2 -Shame (1 1.4% of Variance)
S G S
.61
.49
-
.62
G S S G S G S S S
.69 .59 .50 .75 .40 .54
-
G S S G
.54
-
-
.76 .58
.47
-
-
-
-
-
-
.68 -
.72 .55 .41
Note. S = shame. G = guilt.
the guilt items. Results indicated that the first two varimax-rotated factors for each scale did conform to this expectation, though more clearly for the ASGS. A two-factor solution for each scale (Tables 4 and 5) produced similar results, a first large factor consisting mostly of guilt-related items (29% of the variance for the PFQ2,34.7% for the ASGS) and a second shame-item factor (11.4% of the variance for the PFQ2, 12.5% for the ASGS). Several unexpected scale-item factor loadings at the .40 level or above appeared; PFQ2 self-consciousness, humiliation, and feeling stupid items on the PFQ2 Guilt factor; and the ASGS disgraceful item on the ASGS Guilt factor, and the ASGS deprecated item on the ASGS Shame factor. However, consistent with expectation, PFQ2 stupid and ASGS disgraceful, did simultaneously load more heavily on their respective
SHAME AND GUILT SCALES
735
TABLE 5 Varimax Rotation of the Two-Factor Principal Components Factor Analysis of the ASGS (Loadings .40 and up)
-
Downloaded by [University of Leeds] at 07:55 02 November 2014
ASGS Item
Bashful Mortified Foolish Liable Humiliated Improper Disgraceful Indecorous Guilty Embarrassed Depressed Reproached Immoral Delinquent Inappropriate Ashamed Wicked Deprecated Indecent Unscrupulous Abashed Unethical Shy Imprudent
Original Affect Scale for ltem
Factor I -Guilt (34.7% Variance)
Factor 2 -Shame (12.5% of Variance)
S S S G S G S G S S S G 6 G G S G G G G S G S G
-
- .41 - .55 - .56 -
-
- .64
Note. S = shame. G = guilt.
-
.71 .44 .43 -
.45 .77 .83 .78 -
.43 -
.84 .73 .73 .40
-
-.51 - .46 - .67 - .71 - .67 -.41 -
- .73 - .47 - .70 -
-.52 - .52
- .42
-
Shame factors. Also, placement of the ASGS deprecated item, which is actually closer to shame than guilt in definition, may not be so surprising, in that it loaded fairly highly on both factors in the Hoblitzelle (1982) research. One additional oddity, also noted by Hoblitzelle (1982) after a similar analysis, was the appearance of the ASGS guilty item on the ASGS Shame factor. Next, Pearson product-moment correlations between each of the 4 acceptably reliable affect subscales and the 11 construct validity variables were examined (Tables 6 and 7) to determine their correspondence to predictions (Table 1). The PFQ2 Shame scale showed predicted relationships, at the .05 level or better, with depression (r = .41, p < .001), self-derogation (r = .39, p < .01), social anxiety (r = .23, p < .lo), public self-consciousness (r = .30, p < .05, and social desirability (r = - .28, p < .05), as well as an expected nonsignificant relationship with Machiavellianism (r = .01), but it did not meet expectations for locus of control (r = -.02, ns), narcissism (r = .03, ns), private self-
736
HARDER AND ZALMA TABLE 6
Construct Validity Predictions, Correlations, and Relationships Partialled for Social Disirability for PFQ2 and ASGS Shame Scalesa
Downloaded by [University of Leeds] at 07:55 02 November 2014
Construct Validity Variable Depression Self-derogation Social anxiety Shyness Narcissism Public self-consciousness Private self-consciousness Locus of control Intellectualization Social desirability Machiavellianism PFQ2 guilt ASGS guilt
PFQ2 Shame
ASGS Shame
Partialled PFQ2 Shame
-
.41*** .39** .23# -.01 .03
.39** .46*** .39** .25* - .48***
.40** .39** .17 -.lo - .02
+
.30*
.17
.27*
.29* .02 -.I4 - .25* - .01 .52*** .43***
.26* .15 - .04 -.I5 .01 .46*** .39**
Predicted Relationship
++ ++ + ++
o/ -
+
- /o -
o
+ +
.27* .02 -.I6 -
- .02 .35** .39**
Note. ASGS shame and PFQ2 shame correlated .42*** (.40** when partialled for social desirability). "N = 63. #p < .lo. *p < .05. **p < .01. ***p < .OOl.
consciousness (r = .29, p < .05, in the direction opposite to expectation), and shyness (r = - .01). ASGS Shame showed expected relationships with depression ( r = .39, p < .01), self-derogation (r = .46, p < .001), social anxiety (r = .39, p < .01), shyness ( r = .25, p < .05), and narcissism (r = - .48, p < .001), as well as the expected noncorrelation with Machiavellianism ( r = .01), but also did not match predictions for locus of control ( r = .15, ns), public self-consciousness( r = .17, ns), private consciousness (r = -26, p < .05, when no or a negative correlation was expected), and social desirability (r = - .15, ns). Comparison of the results of both shame subscales suggests that they are about equally valid in terms of the number and type of expected relationships that were observed. The ASGS showed the predicted ability to correlate with shyness, narcissism, and more convincingly with social anxiety, whereas the PFQ2 showed the ability to correlate successfully with public self-consciousness and social desirability. Both shame subscales correlated positively with private self-consciousnesswhen no or a negative relationship was predicted. Similarly, neither subscale related in the expected way with locus of control, although ASGS-Shame correlated in the expected direction. Given the higher reliability coefficients for the ASGS Shame scale and the arguable advantage in its not
7387
SHAiVE AND GUILT SCALES TABLE 7
Construct Validity Predictions, Correlations, and Relationships Partialled for Sex and Social Desirability for PFQ2 and ASGS Guilt Scalesa
Downloaded by [University of Leeds] at 07:55 02 November 2014
Construct Validity Variable
Predicted Relatiomhip
Depression Self-derogation Social anxiety Shyness Narcissism Public self-consciousness Private self-consciousness Locus of control Intellectualization Social desirability Machiavellianism PFQ2 shame ASGS shame
+ +
PFQ2 Guilt (Paflials,Sex)
ASGS Guilt (Partials, Sex)
(.37**) (.40**)
o o o
o/ -
.37**
(.29*)
.22#
(.35**)
.16
.46*** .25* -.22 -.08 -.09 .52*** .46***
(.37**) (.18) (-.I21 (-.I41 (.13) (.50***) (.45***)
.07 .03 -.04 - .30* .13 .43*** .39**
(.23#) (.11) (P.16) (- .28*) (.13) (.50***)
.02 - .05 - .06 .12 .39** .35**
+ + + o o
+
+
(.17) (. 10)
(.26*) (.27*) (.11) (-.15)
-
.39** .46*** .07 .lo .02
(.Of3
.20 .16 .lO -.09 .09
(.44***)
-
ASGS Guilt Partials, Social Desirabilit) .15 .14 .02 -.I2 .03
-
Note. PFQ2 guilt and ASGS guilt correlated .30* (.44*** when partialled by sex; .29* when partialled by social desirability).
correlating strongly with social desirability, at this point, the ASGS appears to be the marginally more valid scale. However, the lack of a significant relationship with public self-consciousness is troubling, although r is in the right direction. , The PFQ2 Guilt scale demonstrated much better construct validity than its ASGS counterpart. PFQ2 Guilt showed predicted significant relationships with depression (r = .39, p < .01), self-derogation (r = .46, P < .001), and private self-consciousness(r = .46, P < .001). It also exhibited nonsignificant correlations with social anxiety (r = .07), shyness (r = .lo), narcissism ( r = .02), Machiavellianism (r = .- .09), and social desirability (r = - .08), as expected fcr a valid guilt measure. However, locus of control did not show the expecteld inverse relationship with PFQ2 Guilt. Instead, a marginally significant positive correlation (r .25, p
Journal of Personality Assessment Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/hjpa20
Two Promising Shame and Guilt Scales: A Construct Validity Comparison David H. Harder & Alysa Zalma Published online: 22 Jun 2011.
To cite this article: David H. Harder & Alysa Zalma (1990) Two Promising Shame and Guilt Scales: A Construct Validity Comparison, Journal of Personality Assessment, 55:3-4, 729-745, DOI: 10.1080/00223891.1990.9674108 To link to this article: http://dx.doi.org/10.1080/00223891.1990.9674108
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions
JOURNAL OF PERSONALITY ASSESSMENT, 1990, 55(3&4), 729-745 Copyright o 1990, Lawrence Erlbaum Associates, Inc.
Two Promising Shame and Guilt Scales: A Construct Validity Comparison David H. Harder and Alysa Zalrna Downloaded by [University of Leeds] at 07:55 02 November 2014
Tufts University
This study compared the validity of two promising measures of shame and guilt proneness: revisions of the Harder Personal Feelings Questionnaire (PFQ2;Harder & Lewis, 1987) and the Hoblitzelle Adapted Shame and Guilt Scale (ASGS; Hoblitzelle, 1982). Internal consistency, test-retest stability, factor structure, and construct validity with convergent and discriminant personality dimensions were examined for both scales. In addition to the shame and guilt measures, 63 (37 male, 26 female) mostly freshman college students completed a randomly ordered battery of personality scales theoretically relevant to shame and guilt proneness. Results support the reliability and shame/guilt factor structure of each scale. ASGS Shame correlations appeared marginally more valid with 11 external construct variables than PFQ2 Shame, whereas PFQ2 Guilt was clearly more valid than its correspondingASGS subscale. New, potentially improved scales were constructed from the factor analyses and from item analyses. However, the resulting scales did not show improved validity.
T h e development of valid scales for the measurement of proneness t o shame and guilt is important for clinically relevant assessments and for the eventual understanding of personality differences. These dysphoric affect states are often linked with difficulties in emotional self-regulation and the etiology of psychopathology (Buss, 1980; Freud, 1905/1953, 1966; Lewis, 1971; Mayman, 1974; Wright, O'Leary, & Balkin, 1989). Unfortunately, the task of measuring them has proven t o be quite difficult (Harder & Lewis, 1987). As distinct emotional states that are not always easy to differentiate clinically or experientially (Levin, 1967, 1971; Lewis, 1971; Mayman, 1974) and that are sometimes defended against by conscious selfreports (Harder & Lewis, 1987), shame and guilt have presented formidable assessment problems from the outset.
Downloaded by [University of Leeds] at 07:55 02 November 2014
730
HARDER AND ZALMA
Previous research (Harder & Lewis, 1987)comparing four promising measures for each affect disposition found that the Gottschalk and Gleser (1969; Gottschalk, Winget, & Gleser, 1969) Shame and Guilt Anxiety scales have insufficient test-retest reliability, whereas two other scales for each emotion, the Binder-Harder Early Memories Shame Proneness Scale (Binder, 1970; Harder, 1984), the Mosher Total Guilt Scale (Mosher, 1966, 1968), and the Beall Situational Upset Scale (Smith, 1972) for both shame and guilt failed to demonstrate convergent and discriminant validity. Harder's Personal Feelings Questionnaire (PFQ, Harder & Lewis, 1987) for both shame and guilt proneness were the only scales emerging from this earlier study with satisfactory test-retest stability and substantial indications of construct validity, despite the presence of few items on each scale. In other research, Hoblitzelle (1982) developed a measure called the Adapted Shame and Guilt Scale (ASGS), which was adapted from an earlier adjective endorsement instrument used by Gioella (1981). The ASGS showed high internal consistency and a two-factor structure largely consistent with expectations that the scale items either reflected shame or guilt dispositions; but in preliminary attempts to establish construct validity, the guilt subscale did not ~ e r f o r mwell (Hoblitselle, 1982). In addition, the test-retest reliability of the ASGS was not investigated, leaving the scale's potential validity as a trait measure in doubt. The aim of our study was to investigate comparatively the abilities of the PFQ and the ASGS to assess shame and guilt proneness. To accomplish this goal, each scale was examined for internal consistency, test-retest stability, factor structure, and a pattern of convergent and discriminant correlations with other personality constructs consistent with theoretical expectations for valid measures. The validation constructs employed were depression, self-derogation, social anxiety, shyness, narcissism, public self-consciousness, private self-consciousness, locus of control, intellectualization defense, social desirability, and Machiavellianism. Clinical and empirical literature (Crouppen, 1977; Harder & Lewis, 1987; Lewis, 1971, 1986; Smith, 1972; Wright et al., 1989) has frequently linked depression with shame and guilt, but more strongly with shame. Hence, positive relationships between valid scales for both dysphoric affects and depression were expected; those with shame were stronger. Self-derogation, representing the negative side of the self-esteem coin, was expected to relate to valid shame and guilt measures in a pattern similar to that shown for depression. Because shame experiences ~roducea more global self-devaluation than guilt (Lewis, 1971; Tangney, 1989), shame was expected to show the stronger relationship. Buss (1980) described a theoretical connection among social anxiety, a tendency to avoid interaction with others, and shame proneness. This social avoidance
Downloaded by [University of Leeds] at 07:55 02 November 2014
tendency is typically accompanied by internal experiences of shyness (Pilkonis, 1977), which has been linked often to shame proneness by clinical theoreticians (Levin, 1967, 1971; Lewis, 1971; Mayman, 1974), and by self-consciousness about one's appearance to others. Thus shame was expected to show associaltions with social anxiety, shyness, and public self-consciousness, whereas guilt was not. Because narcissism has been connected clinically and theoretically with shame vulnerability (Kernberg, 1970; Kohut, 1971; Morrison, 1983) and because the type of narcissism that is successful in defending self-esteem seems to be the variant assessed by personality measures (Harder, 1984; Harder &Lewis, 19871, trait narcissism was predicted to correlate inversely with shame proneness and not to relate significantly to guilt. Because guilt experience has been described as having a more internal locus of origin and evaluation than shame experience (Lewis, 1971), private selfconsciousness and internal locus of control were expected to correlate with guilt proclivities but not with shame. In addition, defense styles favored by guiltoriented individuals include a large share of intellecrualization strategies (Lewis, 1971, 1978), so that the prediction was made that guilt, but not shame, would show associations with a measure of that defense. Even though one might theoretically expect shame to correlate inversely with intellectualization, this was not expected for our sample, because most college students are adept at using intellectualizationl as one of many possible psychological defensive strategies. Machiavellianism (Christie & Geis, 1970) was included as a discriminant validity construct for both shame and guilt, because it was not hypothesized to correlate with either variable. These expectations for valid shame and guilt instruments are summarized schematically in Table 1.
TABLE 1 Predicted Constnuct Validity Relationships for Shame and Guilt Scales -
Construct Validity Variable
-
Guilt
-
++ ++ + ++
Beck depression Self-derogation Social Anxiety Shyness Narcissism Public self-consciousness Private self-consciousness (External) Locus of control Intellectualization Social desirability Machiavellianism
+
-
Shame
++
Note. = positive significant correlation. = high positive significant correlation. negative significanr correlation. o = no significance predicted.
4-ICl
C)
-
=
732
HARDER AND ZALMA
METHOD Subjects
Downloaded by [University of Leeds] at 07:55 02 November 2014
Sixty-three college students at a private liberal arts institution, 90% of them freshmen, volunteered as subjects, to fulfill a course requirement in either an introductory or a statistics psychology course. Thirty-seven of the subjects were male, and 26 were female. The age of the sample ranged from 17 to 22, with a mean of 18.46. Subjects were mostly from urban and suburban upper middleclass backgrounds, and most of them maintained slightly better than a B average in school after one term. Families of origin averaged two siblings. Eleven of the 63 subjects reported their parents were divorced.
Procedure The PFQ (Harder & Lewis, 1987)and ASGS (Hoblitzelle, 1982) versions used in our study were slightly modified from their earlier forms. The PFQ was doubled in length by the addition of three new items and the transposition of original items composed of multiple affect descriptors into separate items (one for each descriptor). The entirely new items were: for shame, self-consciousness and feeling disgusting to others, and for guilt, feeling you deserve criticism for what you did. This procedure produced a 10-itemShame subscale and a six-itemGuilt subscale, rated by the subject on a 4-point scale ranging from you never experience the feeling (0) to you experience the feeling continuously or almost continuously (4). (For a list of the items see Table 3.) This scale revision was called the Personal Feelings Questionnaire-2 (PFQZ). The ASGS was modified by eliminating those six adjectives of the original 30 that failed to show loadings of .40 or greater in Hoblitzelle's (1982) factor analyses of the scale with two different samples of college students. The excluded items were: condemned, wrong, chided, exposed, unconscionable, and criminal. The remaining scale items are listed in Table 4. The revised versions of the Shame and Guilt scales were included with other personality instruments in randomly ordered packets given to subjects by a research assistant. The scales were handed out in manila envelopes, with each enclosure marked only by a code number to preserve anonymity of responses. The confidentialityensured in this way was emphasized by the research assistant in order to maximize honesty of reporting. Subjects completed the scales in rooms that could seat up to 20 subjects at one time. The research assistant remained in the room to clarify scale instructions when there were questions. It took approximately 1 hr to complete the ~ a c k eof t materials. The packet included a demographic information form, the lengthened, PFQ2, the shortened ASGS, the Beck Depression Inventory (Beck, 1967), the Rotter Locus of Control Scale (Rotter, 1966),the Marlowe-Crowne Social Desirability Scale (Crowne & Marlowe, 1960, 1964), the Haan Intellectualization Defense
SHAME AND GUILT SCALES
733
Scale (Haan, 1967; Morrissey, 1977), three Fenigstein, Scheier, and Buss (1975) scales-the Social Anxiety Scale, the Private Self-Consciousness Scale, and the Public Self-Consciousness Scale, the Kaplan Self-Derogation Scale (Kaplan, 1975; Kaplan & Pokorny, 1969), the Narcissistic Personality Inventory (Raskin & Hall, 1979, 1981), the Zimbardo Stanford Shyness Inventory (Zimbardo, 1977; Harder & Lewis, 1987), and the Machiavellianism Scale-Revised (Christie & Geis, 1970). The means and standard deviations for these variables are presented in Table 2.
Downloaded by [University of Leeds] at 07:55 02 November 2014
RESULTS Before testing the construct validity predictions (Table 1) made for relationshi~x between the Shame and Guilt scales and other constructs, the internal consrstency, test-retest stabilxt~(Table 3), and factor structure (Tables 4 and 5) of the PFQ2 and ASGS were investigated. A satisfactory level of internal consistency for both shame and both guilt measures was ascertained, but the ASCiS subscales were superior in this regard. Cronbach's alpha for ASGS Shame was .83; for ASGS Guilt, .89; for PFQ2 Shame, .78; and for PFQ2 Guilt, .72. Two-week test-retest reliability (obtained from a subset of 27 subjects) for ASCS Shame was .93; for ASGS Guilt, -95;for PFQ2 Shame, .91; and for PFQ2 Guilt, .85 -all quite high. If the PFQ2 and ASGS are valid, principal components factor analysis of the two scales was expected to reveal a factor structure consisting predominantly of two factors, one of which would be defined by the shame items and the other by TABLE 2 Means and Standard Deviations of Study Variables - -
Variable PFQ2 shame ASGS shame PFQ2 guilt ASGS guilt Beck depression Self-derogration Social anxiety Shyness Narcissism Public self-consciousness Private self-consciousness (External) Locus of control Intellectualization Social desirability Machiavellianism
M 16.13 30.18 9.76 28.44 27.37 11.03 16.70 29.25 9.18 25.02 31.57 10.37 9.78 12.43 104.45
~
SD
4.51 6.75 3.11 8.17 5.19 6.40 5.92 8.53 4.72 6..81 7..34 3.53 2.48 5.05 8..61
734
HARDER AND ZALMA TABLE 3 Reliability of the ASGS and PFQ2
Affect Subscale
Cronbach's Alpha
Test-Retest Stability
PFQ2 Shame PFQ2 Guilt ASGS Shame ASGS Guilt
.78 .72 .83 .89
.91 35 .93 .95
Downloaded by [University of Leeds] at 07:55 02 November 2014
TABLE 4 Varimax Rotation of the Two-Factor Principal Components Factor Analysis of the PFQ2 (Loadings .40 and up) PFQ2 Item Embarrassed Mild guilt Feeling ridiculous Worry about hurting or injuring someone Self-consciousness Feeling humiliated Intense guilt Feeling "stupid" Regret Feeling "childish" Feeling helpless, paralyzed Feelings of blushing Feeling you deserve criticism for what you did Feeling laughable Feeling disgusting to others Remorse
Original Affect Scale for Item
Factor I -Guilt (29.0% Variance)
Factor 2 -Shame (1 1.4% of Variance)
S G S
.61
.49
-
.62
G S S G S G S S S
.69 .59 .50 .75 .40 .54
-
G S S G
.54
-
-
.76 .58
.47
-
-
-
-
-
-
.68 -
.72 .55 .41
Note. S = shame. G = guilt.
the guilt items. Results indicated that the first two varimax-rotated factors for each scale did conform to this expectation, though more clearly for the ASGS. A two-factor solution for each scale (Tables 4 and 5) produced similar results, a first large factor consisting mostly of guilt-related items (29% of the variance for the PFQ2,34.7% for the ASGS) and a second shame-item factor (11.4% of the variance for the PFQ2, 12.5% for the ASGS). Several unexpected scale-item factor loadings at the .40 level or above appeared; PFQ2 self-consciousness, humiliation, and feeling stupid items on the PFQ2 Guilt factor; and the ASGS disgraceful item on the ASGS Guilt factor, and the ASGS deprecated item on the ASGS Shame factor. However, consistent with expectation, PFQ2 stupid and ASGS disgraceful, did simultaneously load more heavily on their respective
SHAME AND GUILT SCALES
735
TABLE 5 Varimax Rotation of the Two-Factor Principal Components Factor Analysis of the ASGS (Loadings .40 and up)
-
Downloaded by [University of Leeds] at 07:55 02 November 2014
ASGS Item
Bashful Mortified Foolish Liable Humiliated Improper Disgraceful Indecorous Guilty Embarrassed Depressed Reproached Immoral Delinquent Inappropriate Ashamed Wicked Deprecated Indecent Unscrupulous Abashed Unethical Shy Imprudent
Original Affect Scale for ltem
Factor I -Guilt (34.7% Variance)
Factor 2 -Shame (12.5% of Variance)
S S S G S G S G S S S G 6 G G S G G G G S G S G
-
- .41 - .55 - .56 -
-
- .64
Note. S = shame. G = guilt.
-
.71 .44 .43 -
.45 .77 .83 .78 -
.43 -
.84 .73 .73 .40
-
-.51 - .46 - .67 - .71 - .67 -.41 -
- .73 - .47 - .70 -
-.52 - .52
- .42
-
Shame factors. Also, placement of the ASGS deprecated item, which is actually closer to shame than guilt in definition, may not be so surprising, in that it loaded fairly highly on both factors in the Hoblitzelle (1982) research. One additional oddity, also noted by Hoblitzelle (1982) after a similar analysis, was the appearance of the ASGS guilty item on the ASGS Shame factor. Next, Pearson product-moment correlations between each of the 4 acceptably reliable affect subscales and the 11 construct validity variables were examined (Tables 6 and 7) to determine their correspondence to predictions (Table 1). The PFQ2 Shame scale showed predicted relationships, at the .05 level or better, with depression (r = .41, p < .001), self-derogation (r = .39, p < .01), social anxiety (r = .23, p < .lo), public self-consciousness (r = .30, p < .05, and social desirability (r = - .28, p < .05), as well as an expected nonsignificant relationship with Machiavellianism (r = .01), but it did not meet expectations for locus of control (r = -.02, ns), narcissism (r = .03, ns), private self-
736
HARDER AND ZALMA TABLE 6
Construct Validity Predictions, Correlations, and Relationships Partialled for Social Disirability for PFQ2 and ASGS Shame Scalesa
Downloaded by [University of Leeds] at 07:55 02 November 2014
Construct Validity Variable Depression Self-derogation Social anxiety Shyness Narcissism Public self-consciousness Private self-consciousness Locus of control Intellectualization Social desirability Machiavellianism PFQ2 guilt ASGS guilt
PFQ2 Shame
ASGS Shame
Partialled PFQ2 Shame
-
.41*** .39** .23# -.01 .03
.39** .46*** .39** .25* - .48***
.40** .39** .17 -.lo - .02
+
.30*
.17
.27*
.29* .02 -.I4 - .25* - .01 .52*** .43***
.26* .15 - .04 -.I5 .01 .46*** .39**
Predicted Relationship
++ ++ + ++
o/ -
+
- /o -
o
+ +
.27* .02 -.I6 -
- .02 .35** .39**
Note. ASGS shame and PFQ2 shame correlated .42*** (.40** when partialled for social desirability). "N = 63. #p < .lo. *p < .05. **p < .01. ***p < .OOl.
consciousness (r = .29, p < .05, in the direction opposite to expectation), and shyness (r = - .01). ASGS Shame showed expected relationships with depression ( r = .39, p < .01), self-derogation (r = .46, p < .001), social anxiety (r = .39, p < .01), shyness ( r = .25, p < .05), and narcissism (r = - .48, p < .001), as well as the expected noncorrelation with Machiavellianism ( r = .01), but also did not match predictions for locus of control ( r = .15, ns), public self-consciousness( r = .17, ns), private consciousness (r = -26, p < .05, when no or a negative correlation was expected), and social desirability (r = - .15, ns). Comparison of the results of both shame subscales suggests that they are about equally valid in terms of the number and type of expected relationships that were observed. The ASGS showed the predicted ability to correlate with shyness, narcissism, and more convincingly with social anxiety, whereas the PFQ2 showed the ability to correlate successfully with public self-consciousness and social desirability. Both shame subscales correlated positively with private self-consciousnesswhen no or a negative relationship was predicted. Similarly, neither subscale related in the expected way with locus of control, although ASGS-Shame correlated in the expected direction. Given the higher reliability coefficients for the ASGS Shame scale and the arguable advantage in its not
7387
SHAiVE AND GUILT SCALES TABLE 7
Construct Validity Predictions, Correlations, and Relationships Partialled for Sex and Social Desirability for PFQ2 and ASGS Guilt Scalesa
Downloaded by [University of Leeds] at 07:55 02 November 2014
Construct Validity Variable
Predicted Relatiomhip
Depression Self-derogation Social anxiety Shyness Narcissism Public self-consciousness Private self-consciousness Locus of control Intellectualization Social desirability Machiavellianism PFQ2 shame ASGS shame
+ +
PFQ2 Guilt (Paflials,Sex)
ASGS Guilt (Partials, Sex)
(.37**) (.40**)
o o o
o/ -
.37**
(.29*)
.22#
(.35**)
.16
.46*** .25* -.22 -.08 -.09 .52*** .46***
(.37**) (.18) (-.I21 (-.I41 (.13) (.50***) (.45***)
.07 .03 -.04 - .30* .13 .43*** .39**
(.23#) (.11) (P.16) (- .28*) (.13) (.50***)
.02 - .05 - .06 .12 .39** .35**
+ + + o o
+
+
(.17) (. 10)
(.26*) (.27*) (.11) (-.15)
-
.39** .46*** .07 .lo .02
(.Of3
.20 .16 .lO -.09 .09
(.44***)
-
ASGS Guilt Partials, Social Desirabilit) .15 .14 .02 -.I2 .03
-
Note. PFQ2 guilt and ASGS guilt correlated .30* (.44*** when partialled by sex; .29* when partialled by social desirability).
correlating strongly with social desirability, at this point, the ASGS appears to be the marginally more valid scale. However, the lack of a significant relationship with public self-consciousness is troubling, although r is in the right direction. , The PFQ2 Guilt scale demonstrated much better construct validity than its ASGS counterpart. PFQ2 Guilt showed predicted significant relationships with depression (r = .39, p < .01), self-derogation (r = .46, P < .001), and private self-consciousness(r = .46, P < .001). It also exhibited nonsignificant correlations with social anxiety (r = .07), shyness (r = .lo), narcissism ( r = .02), Machiavellianism (r = .- .09), and social desirability (r = - .08), as expected fcr a valid guilt measure. However, locus of control did not show the expecteld inverse relationship with PFQ2 Guilt. Instead, a marginally significant positive correlation (r .25, p
