Measuring Self-Critical and Interpersonal Preoccupations in an Adolescent Inpatient Sample.

By: Susan J. Frank
Department of Psychology, Michigan State University;
Laurie A. Van Egeren
Department of Psychology, Michigan State University
Jennifer S. Paul
Department of Psychology, Michigan State University
Michele O. Poorman
Department of Psychology, Michigan State University
Keith Sanford
Department of Psychology, Michigan State University
Oliver B. Williams
Santa Barbara, California
Daniel T. Field
Rivendell of Michigan Hospital;
Department of Psychiatry, Michigan State University

Acknowledgement: We are grateful to Rivendell of Michigan Hospital and Vendell Corporation for allowing us access to these data. We also thank the many patients who reported on their subjective experiences.

Growing concern about the prevalence of adolescent depression has spurred efforts to identify and measure variables that make certain youth prone to hopelessness and dysphoria (Petersen et al., 1993). One promising approach, based on research with adults, distinguishes preoccupations with relatedness from preoccupations with individuality and self-definition as a way of identifying two distinct types of depression (e.g., Blatt, Quinlan, & Chevron, 1990; Robins & Luten, 1991). This study examines the validity of the Depressive Experiences Questionnaire—Adolescent version (DEQ–A; Blatt, Schaffer, Bers, & Quinlan, 1992), recently developed from an adult instrument to study these two types of depression in adolescents.

The DEQ–A and its parent measure (the Depressive Experiences Questionnaire [ DEQ], Blatt, D'Afflitti, & Quinlan, 1976) assess two dimensions of presumably depressogenic concerns that Blatt and his colleagues denoted as Interpersonal (defined by fears of abandonment and anxieties over the potential loss of gratifying, protective relationships) and Self-Critical (fueled by difficulties maintaining a positive, effective sense of self, and signaled by fears of failure and excessive needs for autonomy and control). Similar distinctions by theorists with related interests include Bowlby's (1980) distinction between “anxious–ambivalent” and “avoidant” attachments, Arieti and Bemporad's (1980) discussion of “other-dominant” and “goal-dominant” motives, and Beck's (1983) comparisons between “socially dependent” personalities and excessively “autonomous” personalities.

Various self-report measures of adults’ preoccupations with relatedness versus individuality can be found in the depression literature; however, the DEQ is the most widely researched (Blaney & Kutcher, 1991) and the first to spawn an adolescent version. Blatt et al. (1976) developed the DEQ by asking college students to indicate relative agreement or disagreement on 66 items describing statements about the self and the self in relation to others that are often associated with but not symptomatic of depression. Three orthogonal factors, extracted from an exploratory factor analysis, have been replicated in other student samples (Jerdonek, 1980; Zuroff, Quinlan, & Blatt, 1990). The first two DEQ factors describe Interpersonal and Self-Critical dimensions. A third, less theoretically articulated factor describes Self-Efficacy. Recently, Blatt et al. (1992) identified comparable dimensions in high school students using the DEQ–A in which items from the adult form were modified to be more appropriate for a younger population. In comparison to the first two (Interpersonal and Self-Critical) factors, the third (Self-Efficacy) factor was marginal but was included for comparison with the DEQ.

A few published studies with the new instrument suggest that hypotheses based on work with adults may apply to adolescents (Blatt, Hart, Quinlan, Leadbeater, & Auerbach, 1993; Fichman, Koestner, & Zuroff, 1994). However, far more research is needed, including research addressing relevant criticisms of the parent measure. In this study, we sought to validate the DEQ–A Interpersonal and Self-Critical scales in a large sample of adolescent psychiatric inpatients: Specific hypotheses addressed the scales’ (a) distinctiveness, (b) concurrent validity, (c) state versus trait properties, and (d) predictive validity.

Independence of Interpersonal and Self-Critical Preoccupations in Clinical Samples

Questions about the distinctiveness of the (adult) DEQ Interpersonal and Self-Critical dimensions in clinical samples were of obvious relevance. Several studies found higher correlations between the two dimensions in patient than in nonpatient samples (Brown & Silberschatz, 1989; Franche & Dobson, 1992). Moreover, Jerdonek (1980), who replicated Blatt et al.'s (1976) three-factor DEQ solution in a sample of college students, found only two factors in an adult clinical sample: one depicting a global dimension of “dysphoric concerns” made up of both interpersonal and self-critical preoccupations and the other describing self-efficacy. Blatt and Zuroff (1992) concluded that absorption in both types of concerns goes hand in hand with severe emotional disturbance; but others refer to the same results in questioning the DEQ's validity in clinical populations (Viglione, Clemmey, & Camenzuli, 1990).

We used confirmatory factor analysis (CFA) to address this issue in our sample of adolescent inpatients. We predicted in Hypothesis 1 that a three-factor model, composed of separate Interpersonal and Self-Critical factors as well as a Self-Efficacy factor, would better fit the observed data than a two-factor model composed of a single Dysphoric Concerns dimension and a Self-Efficacy dimension.

Development of Briefer Scales

In addition to tackling the “distinctiveness” question, the CFA procedures provided a means of addressing criticisms focused on the complexities of scoring the DEQ and DEQ–A (Klein, 1989; Welkowitz, Lish, & Bond, 1985). Scale scores on both are computed as weighted sums of all 66 items. Raw scores for each item are normalized against means and standard deviations identified in the instrument development sample and then weighted by factor scale coefficients from the same study. This procedure is cumbersome (Fichman et al., 1994), and the use of factor weights, based on a single study with a normative sample, potentially compromises generalizability, especially with clinical samples (Viglione et al., 1990). As Welkowitz et al. (1985) pointed out, factor-weighted scores are no more useful than unit weights unless weight estimates are extremely accurate (cf. Dawes, 1979).

Accordingly, we used the CFA results to guide the development of simpler, unit-weight scales for measuring self-critical and interpersonal preoccupations. As we report later correlations between the simpler scales and factors scored according to Blatt et al.'s (1992) procedures were quite high; hence, we were able to use the simpler scales to assess the concurrent validity, state-versus-trait properties, and predictive validity of the two DEQ–A dimensions in our inpatient sample.

Tests of Concurrent Validity: Correlations With Personality and Symptom Dimensions

Several theorists (Arieti & Bemporad, 1980; Beck, 1983; Blatt & Zuroff, 1992) generally agree that an excessive investment in either sociality or individuality, evident in the content of depressive preoccupations, can be observed as well in (a) relatively enduring personality attributes and (b) particular patterns of symptom expression during the clinical phase of a psychiatric illness. This assumption, along with relevant clinical and empirical observations reported in the literature, allowed us to formulate specific predictions bearing on the concurrent validity of the Interpersonal and Self-Critical dimensions of the DEQ–A. We did not make predictions for the more marginal Self-Efficacy dimension.

Relationships With Personality

Interpersonally preoccupied individuals are described as ensuring themselves security and affection by seeking closeness to others; in contrast, self-critical individuals are depicted as protecting their sense of identity and personal control by seeking distance from others (Beck, 1983). These respective patterns have been linked to anxious–ambivalent and compulsively avoidant attachment styles (Blatt & Maroudas, 1992; Zuroff, 1993), but they also can be viewed as general ways of coping that have important implications for a range of personality attributes (Beck, 1983; Blatt & Schichman, 1983).

Previous findings, consistent with presumed differences between the two types in ways of positioning themselves in relation to others, but mostly based on work with adults, led to Hypothesis 2: Interpersonal and self-critical concerns were expected to relate differently to (a) ways of handling hostility and conflict, (b) predilections for deviance versus conformity, (c) degrees of emotional and interpersonal reactivity, and (more tentatively) (d) quality of defenses.

Several studies suggest that, because of fears of rejection and exaggerated needs to please, interpersonally preoccupied types have difficulties openly expressing feelings of anger and hostility, and often act hyperresponsible and overly conforming (Fichman et al., 1994; Klass, 1987; Mongrain, 1993; Zuroff, Moskowitz, Wielgus, Powers, & Franko, 1983). In contrast, self-critical individuals have been shown to be more distressed by signs of failure than rejection (Kutcher & Blaney, 1991) and hence so focused on attaining positive results that they are indifferent to their negative impact on others (Beck, 1983). Researchers find that self-critical persons prefer to avoid or isolate themselves from others; are likely in social interactions to come across as unlikable, irritable, or hostile; and tend to be caught up in overtly conflicted, discordant relationships with family and peers (Blatt, 1991; Hokanson & Butler, 1992; Mongrain, 1993; Robins & Luten, 1991). Alienation from and relative imperviousness to influence by others, excessive investments in autonomy and control, and self-induced pressures for action presumably explain their resentment of authority and predilection for externalizing and deviant behaviors (Beck, 1983; Blatt et al., 1993).

On an emotional plane, interpersonally preoccupied types have been described as labile, with positive or negative feedback from others likely to result in dramatic shifts in mood. Susceptibility to influence and seemingly insatiable needs for support may explain why, as psychiatric patients, they often express optimism about and show little resistance to treatment. In comparison, self-critical types are described as relatively nonreactive to their social environment so that their negative moods are likely to be intense and persistent; moreover, they typically present as pessimistic about and resistant to psychotherapy (Beck, 1983; Blatt & Zuroff, 1992; Mongrain, 1993; Klein, Taylor, Dickstein, & Harding, 1988).

Blatt (1974) hypothesized an additional link between depressive concerns and defensive styles; presumably, interpersonal concerns are associated with the use of less differentiated, more immature defenses (most notably, denial), whereas self-criticism is associated with more differentiated, mature defenses (such as intellectualization and rationalization). However, the evidence for this contention is weak. Beck (1983) has depicted socially dependent persons as more (rather than less) reflective than excessively autonomous patients, and Zuroff et al. (1983) found no support for predicted relationships between the DEQ factors and college students’ use of mature defenses. However, two studies with adult patients relying on objectively scored projective tests (Cramer, Blatt, & Ford, 1988) or clinicians’ judgments (Blatt, Quinlan, Chevron, McDonald, & Zuroff, 1982) tentatively suggest that, when clinically distressed, interpersonally preoccupied individuals may resort to relatively undifferentiated defenses, such as pervasive denial (including denial of their own interpersonal needs), adoption of a caretaker (rather than a cared for) role, hypomanic-like behaviors, and absorption in narcissistic, self-aggrandizing fantasies. We left open the possibility that observed relationships between interpersonal concerns and these sorts of defenses would emerge in our data.

Relationships With Symptom Manifestations of Depression

Both the DEQ and DEQ–A Interpersonal Concerns and Self-Criticism factors generally correlate with global measures of depressed mood (Blatt et al., 1993; Nietzel & Harris, 1990). However, assumptions of continuity between (hypothesized) differences in personality and certain clinical manifestations of depression also find support, with symptom patterns specific to either interpersonal or self-critical concerns seemingly reflecting differences in social distance regulation (Beck, 1983; Blatt et al., 1982; Robins, Block, & Peselow, 1989; Robins & Luten, 1991). Accordingly, Hypothesis 3 proposed that interpersonal preoccupations would be linked to more overt, support-eliciting symptoms of dysphoria and distress such as crying, sadness, anxiety, and somatic complaints, whereas self-criticism would be linked to alienation, anhedonia, and active withdrawal from others.

State Versus Trait Properties

To examine the state-versus-trait properties of constructs measured by the DEQ–A Interpersonal Concerns and Self-Criticism scales, we used longitudinal data collected from 92 of the patients at two or three points during the hospital stay. Research showing a decline in depressive preoccupations as patients recover from depression (Bagby et al., 1994; Klein, Harding, Taylor, & Dickstein, 1988) has led some theorists to question whether the DEQ is measuring state rather than trait dimensions (Klein, Harding, et al., 1988). However, the state and trait properties of these dimensions are likely to be relative rather than absolute.

Depressive preoccupations in normals demonstrate impressive stability over time (Zuroff et al., 1983). Moreover, even after recovery, previously depressed patients report more interpersonal and self-critical preoccupations than never-depressed individuals (Bagby et al., 1994; Klein, Harding, et al., 1988; Franche & Dobson, 1992). Accordingly, in Hypothesis 4 we proposed that adolescent inpatients’ depressive concerns would have state and trait properties, with state properties evident in declines in mean levels from earlier to later points in the hospital stay and trait properties evident in moderate-to-high stability coefficients over the same period.

Predictive Validity

Finally, assumptions about the depressogenic properties of interpersonal and self-critical preoccupations are best tested by longitudinal data, with investigations of their relation to the clinical course of depressed mood offering one way of examining their predictive validity (Klein, 1989). Klein found that both types of preoccupations predicted more depressed mood at 6-month follow-up among adult (especially female) outpatients. This and other studies reporting significant relationships between earlier measures of depressive concerns and later problems with mood (see Neitzel & Harris, 1990) led to Hypothesis 5: We posited that DEQ–A Interpersonal Concerns and Self-Criticism scales, completed close to time of admission, would predict depressed mood measured in the hospital 10 and 25 days later.

Method
Participants

Participants were 823 patients (414 girls), ages 11 to 17 years (M = 14.4), in a private psychiatric hospital in the Midwest. Subsamples from this larger group were used to test the various hypotheses. Hospital referrals came from the private and public sectors, as well as from television commercials and word of mouth. Participants were mostly White (89%), with 48% receiving Medicaid benefits. Parent educational attainment was similar for mothers and fathers, with both typically reporting a high school degree or some college. Diagnoses based on the revised third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM–III–R; American Psychiatric Association, 1987) obtained from discharge reports were made by hospital psychiatrists. Seventy-two percent of the sample had more than one Axis 1 diagnosis; the most frequent were major depression or dysthymia (71.4%), attention deficit/hyperactivity disorder (ADHD; 26.2%), and oppositional defiant disorder (ODD, 19.9%).

Data for this study were collected over 2 years. Approximately 11% of the adolescent patients admitted during that time were considered ineligible because hospital records indicated mental retardation (Wechsler Intelligence Scale for Children—Third Edition [WISC–III] Full Scale IQ < 70; Wechsler, 1991), psychosis, or organicity. The 823 participants represented approximately 98% of all eligible patients.

Measures

All 823 participants completed the Depressive Experiences Questionnaire—Adolescent Version (DEQ–A; Blatt et al., 1992). The 66 DEQ–A items were rated on a 7-point Likert scale with anchors of 1 = strongly disagree and 7 = strongly agree. The Interpersonal Concerns and Self-Criticism factor scales reportedly are internally consistent, have good test–retest reliability (Blatt et al., 1992), and relate in meaningful ways to normal adolescents’ reports of behavior problems (Blatt et al., 1993) and interpersonal difficulties (Fichman et al., 1994).

Complete data on depressive symptoms were available for 94% of the participants: 357 boys and 381 girls in Grades 7 through 12 completed the Reynold's Adolescent Depression Scale (RADS; Reynolds, 1986), and 25 boys and 12 girls in Grades 4 through 6 completed the Reynold's Child Depression Scale (RCDS; Reynolds, 1991). Both are 30-item self-report questionnaires using a 4-point Likert scale to assess the frequency of depressive symptoms. The RADS has good concurrent validity (Atlas & DiScipio, 1992), assesses domains of depression typically identified in the depression literature (Campbell, Byrne, & Baron, 1994), and is sensitive to gender differences in depressive symptomatology (Campbell, Byrne, & Baron, 1992). Percentile scores, normed for sex and grade, were used in most analyses. Because a few items differ on the RADS and RCDS, only the RADS data were included in an item-level factor analysis used to identify clusters of depressive symptoms.

A subsample of participants completed the Minnesota Multiphasic Personality Inventory—Adolescent Version (MMPI–A; Butcher et al., 1992), a 478-item self-report measure with a true–false response format. In addition to several validity indicators and 10 clinical scales assessing psychopathology and distress, the MMPI–A allows for the scoring of 52 content scales, providing additional information about personality, behavior, and perceptions of the social environment. The MMPI–A became part of the hospital's standard assessment battery shortly after the introduction of the DEQ–A. However, its length made it impractical as a clinical tool in this acute-care facility, and it was discontinued after approximately 8 months. For this reason, MMPI–A data were available for only 231 (28%) of the total sample. Protocols for an additional 26 patients were disregarded because validity scales suggested “faking bad” or careless responding (F > 100) or a defensive test-taking attitude (K > 70; Butcher et al., 1992). Comparisons of participants for whom we did or did not have usable MMPI–A data showed no differences in DEQ–A scores, depression scores, sex, or number of Medicaid recipients.

Procedures

Trained undergraduate research assistants individually administered the DEQ–A and the RADS or RCDS to essentially all consecutive adolescent admissions within 48 hr of admission. Data used to assess changes in depressive preoccupations and depression were collected during the last 8 months of the 24-month study period: Adolescents who were still in the hospital 10 days after the first administration of the DEQ–A and depressive symptoms questionnaire were asked to complete the same measures a second time, and those in this group who continued to be hospitalized 25 days after the first administration were asked to complete them a third time. Patients hospitalized for 21 days or fewer (M = 17, SD = 3.6) were designated the short stay group (23 boys, 30 girls), and those hospitalized for 25 days or more (M = 30, SD = 3.3) were designated the long stay group (22 boys, 17 girls). Data for 12 patients with lengths of stay between 22 and 24 days were disregarded to more clearly delineate group differences in recovery rates. Patients completed the MMPI–A in a group testing situation during the first week in the hospital.

Results
Examination of the Three-Factor Model (Hypothesis 1)

We used confirmatory factor analysis (CFA; LISREL VII,Jöreskog & Sörbom, 1988) to assess whether observed relationships among the DEQ–A items would better fit a three-factor solution that included separate dimensions of Self-Criticism and Interpersonal Concerns as well as a Self-Efficacy dimension (Hypothesis 1) or whether they would be more consistent with a two-factor solution such as that identified in Jerdonek's (1980) study of adult psychiatric patients. The sample comprised the first 670 patients for whom we had complete DEQ–A data.

Blatt et al.'s (1992) scoring procedure weights every item on every factor (or latent variable), with a number of items loading on more than one factor. To test the three-factor model, we first sought to identify subgroups of items primarily associated in Blatt et al.'s solution with one, and only one, factor. The CFA procedure was conducted on a covariance matrix derived from a subsample of 39 items selected from the 66-item questionnaire. Twenty-two selected items were associated with Blatt et al.'s Interpersonal Concerns factor, and eight with their Self-Criticism factor. These 30 items were selected because they simultaneously best defined and discriminated between the Interpersonal Concerns and Self-Criticism dimensions (i.e., in the original solution, they had at least moderate loadings on one factor [≥.40] and relatively low loadings [<.30] on the other). The remaining nine items were associated with the Self-Efficacy factor (with factor loadings ≥.40 on Self-Efficacy and <.40 on the Interpersonal Concerns and Self-Criticism factors).

Because Blatt et al.'s (1992) solution was orthogonal, path coefficients connecting the latent variables were fixed at 0. Coefficients from the latent variables to their respective items were free to be estimated, and item error variances within factors were allowed to correlate if statistically significant.

Some debate exists regarding the most appropriate index to assess model fit. We relied on the goodness of fit index (GFI) rather than the chi-square likelihood ratio statistic because the latter is highly likely to be significant and therefore misleading in a sample of this size. The GFI for the proposed three-factor orthogonal model was .83, which approached but did not meet the .90 value generally considered to indicate adequate fit. However, the GFI associated with a two-factor model composed of a global Depressive Concerns factor and a Self-Efficacy factor resulted in an even less adequate fit (GFI = .76).

As a second step, we sought to improve the fit of the predicted three-factor model without sacrificing orthogonality by eliminating five items in the model associated with the Interpersonal Concerns factor, one item associated with the Self-Criticism factor, and six associated with the Self-Efficacy factor; each of these items had modification indices for the lambda-x matrix of greater than a selected cutoff of 25, suggesting that they would “prefer” to load on more than one factor. The resulting 27-item solution had a GFI of .90. However, examination of this solution revealed a high modification index (68.14) for one of only three remaining Self-Efficacy items. As a consequence, we decided to eliminate the Self-Efficacy factor from the model.

An analysis of the final 24-item, two-factor model resulted in a GFI of .92, with 17 items associated with the Interpersonal Concerns factor and seven with the Self-Criticism factor. These items, along with their completely standardized path coefficients, are shown in Table 1. T values for all path coefficients were significant at the .001 level, with the exception of Item 57 (p < .05). Interpersonal Concerns items describe anxieties about abandonment and rejection sensitivity, fears of hurting or offending others, and substantial fluctuations in self-esteem. In contrast, Self-Criticism items focus on what Blatt and Maroudas (1992) described as difficulties maintaining a positive, effective sense of self (4 reverse-scored items are positive self-statements) as well as interpersonal disengagement and an apprehensive, avoidant stance toward relationships with significant others (Bowlby, 1980).

A last question, addressed by an additional CFA analysis, was whether the final model would fit equally well for boys and girls. The GFI resulting from this analysis was .92, indicating that the model was relatively invariant across gender groups.

We computed unit-weight Interpersonal Concerns and Self-Criticism scale scores for the larger sample of 823 inpatients by averaging the raw scores for items associated with the Interpersonal Concerns or Self-Criticism factors in the final LISREL model (see Table 1). Internal consistencies for the unit-weight DEQ–A scales were quite good for Interpersonal Concerns (.87) and lower but adequate (.62) for Self-Criticism concerns. Correlations between the unit- and factor-weighted scales were high (.95 for the Interpersonal Concerns scales and .82 for the Self-Criticism scales, p < .001). In the total sample, the Interpersonal Concerns and Self-Criticism scales correlated more using the unit-weight (r = .23, p < .001) than the factor-weight scales (r = .01); but for girls, these correlations were .33, p < .001, and .14, p < .01 (respectively), whereas for boys, they were .07, ns, and −.15, p < .01. Girls acknowledged more interpersonal preoccupations (M = 4.6, SD = 1.1) than did boys (M = 4.1, SD = 1.0, p < .001), and they also reported more self-criticism (M = 3.9, SD = 1.1) than boys (M = 3.6; SD = 1.1), p < .001. Mean total percentile scores on the RADS (not normalized for sex) also showed girls reporting more depressed mood (M = 76.6; SD = 26.0, p < .001) than boys (M = 63.7; SD = 29.0); controlling for depressed mood did not affect gender differences on the Interpersonal Concerns dimension, p < .001, but gender differences on the Self-Critical dimension virtually disappeared, p > .99.

Concurrent Validity (Hypotheses 2 and 3)

Data reduction

Separate factor analyses of (a) the 52 MMPI–A content scales and (b) the 30 items on the RADS allowed us to simplify tests of concurrent validity. Analysis of the MMPI–A variables, using data from 231 of the patients (125 girls), resulted in eight personality factors. Descriptions in Butcher et al. (1992) and Wrobel (1992) guided interpretations of each dimension. The first and largest factor, defined by a total of 22 scales (with factor loadings ≥ |.50|) accounted for 42.8% of the variance and described General Maladjustment and Alienation: Scales defining this factor described intense negative affect (e.g., hopelessness, hostility) and problems with self-definition (e.g., lack of self-integration, overinvolvement in subjective processes), as well as social and emotional alienation, and tendencies to be “high strung” and “thin-skinned.” A scale assessing negative treatment indicators also loaded on this factor. The remaining seven factors accounted for between 9.3% and 2.0% of the variance. Factor 2, defined by five scales, described Social Discomfort and Dependency (e.g., shyness, oversensitivity to others’ opinions, (low) denial of dependency needs, and feelings of anxiety in social situations). Factor 3, defined by seven scales, described Hypomanic Attributes (e.g., psychomotor acceleration, failure of repressive defenses in controlling emotional expression, denial of needs for affection, and grandiose self-appraisals). The fourth factor (six scales) described problems associated with Deviant Behavior (e.g., conduct and alcohol problems, problems with authorities or at school, and amorality), and the four remaining factors (defined by two to three scales) described Psychoticism, Somatization, Family Discord, and Social Introversion (vs. Social Extraversion).

Factor analysis of the 30 items on the Reynolds (1986) Adolescent Depression Scale using data from 738 of the patients (381 girls) identified four dimensions indicative of different facets of depressive symptom expression. The first of four factors identified by a varimax rotation procedure accounted for 33.9% of the variance and described Alienation/Self-Denigration (e.g., “I feel my parents don't like me;” “I feel I am bad”). The remaining three factors accounted for between 7.7% and 4.1% of the variance, with the second describing Overt Symptoms of Dysphoria (“I feel [worried] [sad] [like crying] [lonely]”); the third, Social Withdrawal/Anhedonia (“I [do not] feel like having fun with other students [−]”, “I [do not] feel happy [−]”); and the fourth, Somatic Complaints (“I get stomachaches”).

Relationships with the MMPI–A personality dimensions

We used hierarchical regression analyses to test predicted relationships between the unit-weight DEQ–A Interpersonal Concerns and Self-Criticism scales and the MMPI–A content factors. In these analyses we controlled for possible confounds with Depressed Mood before assessing effects for depressive concerns. We also tested for interactions, sometimes found in DEQ studies (e.g., Blatt et al., 1982; Riley & McCranie, 1990; Zuroff et al., 1983), between depressive concerns and gender. Significance levels were set at p < .01 to avoid interpreting small and potentially spurious relationships. Each analysis regressed the dependent variable on adolescent sex (Step 1), scores on the RADS (Step 2), and scores on the two DEQ–A scales (Step 3). On the final step, the possibility of Sex × Interpersonal Concerns or Sex × Self-Criticism interactions was tested in a stepwise fashion.

On the basis of previous discussions of personality differences between the two types of individuals with depression (reviewed in an earlier section), we predicted that Interpersonal Concerns would be associated with social discomfort and dependency, hypomanic attributes, social conformity (i.e., low scores on the Deviant Behavior factor), somatization, and social extraversion. In addition, we predicted that Self-Criticism would be linked to general maladjustment and alienation, deviance, perceptions of family discord, and social introversion. The results shown in Table 2 supported six of the nine predicted relationships. However, Interpersonal Concerns did not relate significantly to Social Discomfort or Somatization, and Self-Criticism did not relate to Social Introversion. Contrary to expectation, interpersonal preoccupations were associated with perceptions (possibly bolstered by denial) of less family discord. Only one interaction with sex was statistically significant (Δ R2 = .06, p < .001), with Self-Criticism related to deviant behavior in girls (b = .50, p < .001) but (surprisingly) not boys (b = −.04).

Relationships with symptom manifestations of depression

In terms of symptom expression, we predicted that (a) both the Interpersonal Concerns and Self-Critical dimensions would be associated with global depression but that (b) Interpersonal Concerns would be linked to scales defined by support-eliciting symptoms of distress (i.e., overt symptoms of dysphoria and somatic complaints), whereas (c) Self-Critical concerns would be associated with symptom scales indicative of social and emotional alienation, and interpersonal withdrawal and anhedonia. Results for the most part were consistent with these predictions (see Table 3). A significant relationship between the Alienation/Self-Denigration factor on the RADS and Interpersonal Concerns was not predicted, but it was much smaller than the relationship between that RADS factor and Self-Criticism. The one significant interaction with sex (Δ R2 = .01, p < .01) was because Interpersonal Concerns were negatively related to social withdrawal/anhedonia for boys (β = −.15, p < .01) but not girls (β = .08)

Comparisons between the unit-weight and factor-weight scales

Regression analyses reassessing the two DEQ–A scales’ concurrent validity using the factor-weight scales in place of the unit-weight scales showed only minimal differences for the two. Twelve of 15 predicted relationships were significant (i.e., p < .01) for the unit-weight scales, as was also true for the factor-weight scales. However, the factor-weight scales were somewhat less discriminating: In four instances, both the Interpersonal Concerns and Self-Criticism scales correlated significantly in the same direction with a given criterion variable, even though only one of the observed relationships had been predicted, and the other observed relationship was theoretically inconsistent. This occurred only once using the unit-weight scales.

State and Trait Properties of the DEQ–A Variables

We used longitudinal data collected at two or three points during the hospital stay to assess state-like changes and trait-like stability in the constructs assessed by the DEQ–A Interpersonal Concerns and Self-Criticism scales. Table 4 shows changes in mean scores for depression and depressive preoccupations in the short stay (n = 53) and long stay (n = 39) groups. We used means and standard deviations from the larger sample (based on data collected within 48 hr of admission) to normalize scores for each variable.

We predicted that state-like qualities of depressogenic schemas would be evident in changes in mean levels over time; however, these changes were only apparent for Self-Criticism. The most immediate and dramatic changes were in the short stay group. Between Time 1 and Time 2, Self-Criticism (as well as depressed mood) decreased significantly in the short stay group but not in the long stay group. Significant changes in Self-Criticism as well as depressive symptoms became evident in the long stay group by Time 3, but mean scores for Interpersonal Concerns did not differ from one time to another in either group.

Additional analyses examined whether the stability of the means for Interpersonal Concerns was due to individuals’ changing in both directions. We calculated individual change (i.e., difference) scores showing changes in depressed mood as well as levels of depressive concerns between the first and last points of measurement (i.e., Time 1 and Time 2 in the short stay group, and Time 1 and Time 3 in the long stay group). Approximately 25% of the patients demonstrated substantial change (defined as greater than 1 standard deviation) in a positive direction on both the Self-Criticism and Depressed Mood dimensions, whereas only 5% showed the same degree of change on these dimensions in a negative direction. Direction of change on the Interpersonal Concerns dimension was more balanced, with 18% showing substantial change in a negative direction and 10% showing substantial change in a positive direction. When we transformed the change scores into absolute values, we found no significant differences in amount of change on the three dimensions. In the short stay group, the mean absolute change score on the Interpersonal Concerns dimension (M = .69, SD = .59) was lower than on the Self-Criticism (M = .84, SD = .69) or Depressed Mood (M = .84, SD = .81) dimensions, but paired comparisons were not statistically significant; moreover, in the long stay group, mean absolute change scores for all three dimensions were essentially the same (approximately .70). In short, these data do not show definitive differences in the state-like properties of the Interpersonal Concerns dimension as compared with the Self-Critical or Depressed Mood dimensions. However, we did find differences in absolute change score variances: These were significantly smaller for the Interpersonal Concerns dimension than for (a) the Depressed Mood dimension in the short stay group (p < .05) or (b) both the Self-Criticism (p < .05) and the Depressed Mood dimensions (p < .01) in the long stay group.

We also predicted that trait-like qualities of depressogenic concerns would be evident in moderate to high stability coefficients over time. Simple correlations suggested at least moderate stability for both depressive preoccupations (r s = .40 to .73) and depressive symptoms (r s = .47 to .54). However, correlations for depressive symptoms (presumably a state variable) potentially were inflated by relationships with depressive concerns (or possibly, vice versa). We therefore used regression analyses to recalculate stability coefficients for Self-Criticism and Interpersonal Concerns between (a) Time 1 and Time 2 after controlling for depression at Time 1, (b) Time 2 and Time 3 after controlling for depression at Time 2, and (c) Time 1 and Time 3 after controlling for depression at Time 1. Likewise, in recalculating stability coefficients for depression, we controlled for Self-Criticism or Interpersonal Concerns at the earlier time point.

As can be seen in Table 5, both depressive concerns and depressive symptoms continued to demonstrate moderate stability in the short stay group. However, in the long stay group, depressive concerns were more stable than depressive symptoms. Lack of stability in depressive symptoms became most apparent when stability was calculated over the 25-day interval and Self-Critical (rather than interpersonal) concerns were controlled for in the regression analysis.

Predictive Validity

We used regression analyses to test predicted relationships between the DEQ–A Interpersonal and Self-Critical dimensions, measured soon after admission to the hospital, and depressive symptoms on the RADS, measured 10 or 25 days later in the hospital stay. In both the long stay and short stay groups, we regressed Time 2 Depression scores onto (a) Time 1 Depression (tested in Step 1) and (b) Time 1 Self-Criticism and Time 1 Interpersonal Concerns (tested stepwise in Step 2). In the long stay group, we also regressed Time 3 Depression on (a) Time 1 and Time 2 Depression (tested stepwise on Step 1), and (b) Time 1 and Time 2 Self-Criticism and Interpersonal Concerns (tested stepwise on Step 2). We also assessed whether earlier experiences of depression would predict later self-critical or interpersonal preoccupations.

Both Interpersonal Concerns and Self-Criticism were correlated with Depressed Mood at each time point within each of the two patient groups (average r = .49). Yet none of the predictive pathways from earlier interpersonal concerns to later depressed mood were statistically significant in either group, and earlier self-critical concerns did not predict later depression in the short stay group; in each of these analyses, only earlier depressed mood predicted later depressed mood. In contrast, in the long stay group, both Time 2 Depressed Mood and Time 1 Self-Criticism predicted Time 3 Depression (with both variables in the analysis, β = .37, p < .05, for time 2 Depression and .41 (p < .01) for Time 1 Self-Criticism). None of the pathways from depression to self-criticism were significant.

Discussion

In general, these findings support the validity of the DEQ–A Interpersonal Concerns and Self-Criticism scales as measures of depressive concerns among seriously disturbed adolescents. Concerns about the psychometric limitations of the DEQ–A's parent measure revolve in part around difficulties in identifying the two types of concerns as clearly distinct dimensions in adult patient populations (Viglione et al., 1990). It is difficult to know whether we did not observe the same difficulties because we used the DEQ–A (rather than the DEQ) or because we studied disturbed adolescents (rather than disturbed adults). This question aside, several findings belied the importance of the “distinctiveness” issue in this sample. First, we found no relationship between the factor-weight DEQ–A Interpersonal Concerns and Self-Critical scales. Second, the “first pass” of a confirmatory analysis indicated that observed responses to DEQ–A items identified in Blatt et al.'s (1992) original solution as most definitive of each of the two scales better fit a theoretical model positing separate Interpersonal Concerns and Self-Criticism dimensions than an alternative model positing a single “Dysphoric Preoccupations” dimension. Third, tests of the concurrent validity of the two DEQ–A scales indicated that they either were related to different criterion variables or to the same criterion variable in different ways; and although a total reliance on self-report data may have inflated the magnitude of these relationships, the fact that we found predicted patterns of significant relationships (as well as a sizable number of them) is encouraging.

Comparisons of patterns of change and stability for the DEQ–A Self-Criticism scores and the RADS Depressed Mood scores also were consistent with predictions: Mean levels on both scales exhibited state-like properties (i.e., they declined significantly over time), but Self-Criticism (and Interpersonal Concerns) exhibited more trait-like properties than did Depressed Mood (i.e., stability coefficients over a 25-day period were stronger for the two DEQ–A variables than for the RADS). In contrast, mean levels for the DEQ–A Interpersonal Concerns scales showed little to no change over 10 or 25 days in the hospital. This proved, however, to be largely a function of direction rather than lack of change (i.e., some patients experienced more and others less of these concerns from one measurement point to the next).

Indications that patterns of change for Self-Criticism and Depressed Mood are relatively similar to one another and also quite different from patterns of change associated with Interpersonal Concerns could help to explain why we were able to predict later reports of depressed mood from earlier Self-Criticism scores, but not from earlier Interpersonal Concerns scores. The latter is not an isolated finding: DEQ studies with adults also show stronger and more consistent relationships between antecedent measures of self-criticism and later reports of depressed mood than between antecedent measures of interpersonal concerns and later depressed mood (Neitzel & Harris, 1990). Moreover, different patterns of results in the short stay as compared with the long stay groups bring to mind previous studies suggesting that unique mechanisms are involved in chronic as compared with relatively acute depressions (e.g., Klein, Taylor, et al., 1988). Yet, these same findings serve as a reminder that when, and how often, measurement takes place can affect the nature of observed relationships between dynamic predictor and criterion variables—especially when these variables are changing in different ways or according to different timetables (Cohen, 1991). Accordingly, our results can be cited in support of the predictive validity of the DEQ–A Self-Criticism scale, but they do not necessarily rule out a predictive relationship between DEQ–A Interpersonal Concerns and adolescents’ experiences of depression.

More research is needed to address this and other issues. In particular, it is possible if not probable that our findings were affected by situational and sample characteristics. For example, counterbalanced presses associated with hospitalization could explain why we observed fewer extreme fluctuations in absolute change scores on the Interpersonal Concerns scale as compared with the Self-Criticism and Depressed Mood scales over the hospital stay. Research suggests that interpersonally preoccupied adult patients show relief at being hospitalized, presumably because they see an opportunity to gratify intense needs to be cared for and protected. However, interpersonally preoccupied adolescents, separated by hospitalization from their parents, are also likely to experience intense feelings of abandonment.

Situational factors also may explain the relative absence of gender differences on criterion variables tapping externalizing or internalizing aspects of personality (e.g., deviance or social discomfort). In the general population, boys typically are more externalizing and girls, more internalizing (Achenbach, 1991). However, admission criteria (driven by local mental health politics and insurance criteria) made it difficult to hospitalize adolescents with internalizing problems who did not also engage in disruptive (i.e., externalizing) behaviors, and they also precluded admission of conduct-disordered adolescents who did not have an accompanying “psychiatric” diagnosis. Hence, female patients may have been more externalizing and males patients more internalizing than is typical. Whether more gender differences, or just as important, Gender × Depressive Concerns interactions, would have emerged in a normal or outpatient sample remains to be investigated.

In closing, we want to comment on whether these data justify a preference for either the unit-weight or the factor-weight scales. We have found, and others have commented as well (e.g., Fichman et al., 1994; Klein, 1989), that the length and scoring procedures for the DEQ and DEQ–A can make these instruments prohibitive in certain field settings. However, the two types of scales were highly intercorrelated and generally related to the criterion variables in similar ways, and although the factor-weight scales were somewhat less discriminating, at this point, preferences for one or the other likely will depend on the researchers’ resources and aims.

Footnotes

1  Blatt and his colleagues have also referred to interpersonal preoccupations as “anaclitic” or “dependent” and self-critical preoccupations as “introjective.”

2  In addition to one or the other type of depression, interpersonal concerns presumably can be manifest in panic and other anxiety disorders, and self-critical concerns in obsessive compulsive neurosis, paranoia, and “neurotic” delinquency (Bagby et al., 1992; Beck, 1983; Blatt & Shichman, 1981, 1983).

3  DEQ–A Items 2, 4, 9, 10, 16, 19, 20, 22, 23, 26, 28, 32, 36, 37, 40, 41, 45, 46, 50, 51, 52, and 55 were identified as Interpersonal Concerns items, DEQ-A Items 8, 17, 21, 35, 48, 57, 61, and 62 as Self-Criticism items, and DEQ–A Items 1, 14, 15, 24, 33, 42, 59, 60, and 66 as Self-Efficacy items.

4  Others have shown congruency for the DEQ and DEQ–A Interpersonal Concerns and Self-Criticism dimensions across gender groups (Blatt et al., 1976, 1992; Zuroff et al., 1990).

5  Alphas within the various subsamples used in later analyses were quite stable (M = .86 for Interpersonal Concerns and .66 for Self-Criticism).

6  Analyses of the MMPI–A and the Reynolds data resulted in very similar factor solutions for boys, girls, and both sexes combined. Results are available from Susan J. Frank. We expected some overlap between the MMPI–A and RADS factors. However, 71% of 32 correlations between the personality and symptom dimensions were smaller than |.20|, and only 16% were larger |.30|; range = .00 to .45.

7  Archer, Belevich, and Elkins (1994) identified eight MMPI–A factors in the normative sample. Ours were quite similar, even though we had an inpatient sample, used a varimax (rather than an oblique) rotation, and did not include scores for the validity or clinical scales. Means and variances on the MMPI–A factors were very similar for boys and girls (except for Psychoticism). The Discussion section addresses the lack of additional gender differences.

8  Interpretation of the Intraversion/Extraversion factor was based on Butcher et al.'s (1992) description of the two contributing scales: the Cynicism scale, which loaded positively, and the MacAndrews Alcohol scale, which loaded negatively.

9  It was not possible to conduct a structured diagnostic interview, but we did examine relationships between patients’ depressive concerns and psychiatrist diagnosis of a depressive disorder (dysthymia or major depression). Results varied by gender: Boys diagnosed with a depressive disorder (n = 282) were more self-critical (M = 3.66) than boys without this diagnosis (n = 127, M = 3.38) after controlling for interpersonal concerns, F(1, 406) = 6.20, p < .05; girls diagnosed with a depressive disorder (n = 305) had more interpersonal concerns (M = 4.68) than girls without this diagnosis (n = 109, M = 4.37) after controlling for self-criticism, F(1, 411) = 5.97, p < .01. Neither boys’ interpersonal concerns nor girls’ self-critical concerns related to judgments about the presence of a depressive disorder, and depressive concerns, in general, were unrelated to behavioral disorder diagnoses (conduct, oppositional defiant, or attention deficit hyperactivity disorders).

References

Achenbach, T. (1991). Manual for the Child Behavior Checklist/4–18 and the 1991 profile. Burlington: University of Vermont Department of Psychiatry.

American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed., rev.). Washington, DC: Author.

Archer, R. P., Belevich, J. K. S., & Elkins, D. E. (1994). Item-level and scale-level factor structures of the MMPI-A. Journal of Personality Assessment, 62, 332–345.

Arieti, S., & Bemporad, J. R. (1980). The psychological organization of depression. American Journal of Psychiatry, 137, 1360–1365.

Atlas, J., & DiScipio, W. (1992). Correlations of Beck Depression Inventory and Reynolds Adolescent Depression Scale. Psychological Reports, 70, 621–622.

Bagby, R. M., Cox, B. J., Schuller, D. R., Levitt, A. J., Swinson, R. P., & Joffe, R. T. (1992). Diagnostic specificity of the dependent and self-critical personality dimensions in major depression. Journal of Affective Disorders, 26, 59–64.

Bagby, M. R., Schuller, D. R., Parker, J. D. A., Levitt, A., Joffe, R. T., & Shafir, M. S. (1994). Major depression and the self-criticism and dependency personality dimensions. American Journal of Psychiatry, 151, 597–599.

Beck, A. T. (1983). Cognitive therapy of depression: New perspectives. In P. J.Clayton & J. E.Barrett (Eds.), Treatment of depression: Old controversies and new approaches (pp. 265–290). New York: Raven Press.

Blaney, P., & Kutcher, G. S. (1991). Measures of depressive dimensions: Are they interchangeable?Journal of Personality Assessment, 56, 502–512.

Blatt, S. J. (1974). Levels of object representation in anaclitic and introjective depression. Psychoanalytic Study of the Child, 29, 101–157.

Blatt, S. J. (1991). Depression and destructive risk-taking behavior in adolescents. In L. P.Lipsett & L. L.Mitnick (Eds.), Self-regulatory behavior and risk-taking: Causes and consequences (pp. 285–309). Norwood, NJ: Ablex.

Blatt, S. J., D'Afflitti, J. P., & Quinlan, D. M. (1976). The Depressive Experiences Questionnaire. New Haven, CT: Yale University Press.

Blatt, S. J., Hart, B., Quinlan, D. M., Leadbeater, B., & Auerbach, J. (1993). Interpersonal and self-critical dysphoria and behavioral problems in adolescents. Journal of Youth and Adolescence, 22, 253–269.

Blatt, S. J., & Maroudas, C. (1992). Convergences among psychoanalytic and cognitive-behavioral theories of depression. Psychoanalytic Psychology, 9, 157–190.

Blatt, S. J., Quinlan, D. M., & Chevron, E. S. (1990). Empirical investigations of a psychoanalytic theory of depression. In J.Masling (Ed.), Empirical studies of psychoanalytic theories (Vol. 3, pp. 89–147). Hillsdale, NJ: Analytic Press.

Blatt, S. J., Quinlan, D. M., Chevron, E. S., McDonald, C., & Zuroff, D. (1982). Dependency and self-criticism: Psychological dimensions of depression. Journal of Consulting and Clinical Psychology, 50, 113–124.

Blatt, S. J., Schaffer, C. E., Bers, S. A., & Quinlan, D. M. (1992). Psychometric properties of the Depressive Experiences Questionnaire for Adolescents. Journal of Personality Assessment, 59, 82–98.

Blatt, S. J., & Schichman, S. (1981). Antisocial behavior and personality organization. In S.Tuttman, C.Kaye, & M.Zimmerman (Eds.), Object and self: A developmental approach (pp. 325–367). New York: International Universities Press.

Blatt, S. J., & Schichman, S. (1983). Two primary configurations of psychopathology. Psychoanalysis and Contemporary Thought, 6, 187–254.

Blatt, S. J., & Zuroff, D. C. (1992). Interpersonal relatedness and self-definition: Two prototypes for depression. Clinical Psychology Review, 12, 527–562.

Bowlby, J. (1980). Attachment and loss: Vol. 3. Loss, separation and depression. New York: Basic Books.

Brown, J. D., & Silberschatz, G. (1989). Dependency, self-criticism, and depressive attributional style. Journal of Abnormal Psychology, 98, 187–188.

Butcher, J. N., Williams, C. L., Graham, J. R., Archer, R., Tellegen, A. M., Ben-Porath, Y. S., & Kaemmer, B. (1992). MMPI-A: Manual for administration, scoring, and interpretation. Minneapolis: University of Minnesota Press.

Campbell, T. L., Byrne, B., & Baron, P. (1992). Gender differences in the expression of depressive symptoms in early adolescents. Journal of Early Adolescence, 12, 326–338.

Campbell, T. L., Byrne, B., & Baron, P. (1994). The Reynolds Adolescent Depression Scale: An exploratory factor analytic study [Special issue: Facets of social intelligence]. European Review of Applied Psychology, 44, 319–325.

Cohen, P. (1991). A source of bias in longitudinal investigations of change. In L. M.Collins & J. L.Horn (Eds.), Best methods for the analysis of change: Recent advances, unanswered questions, future directions (pp. 18–25). Washington, DC: American Psychological Association.

Cramer, P., Blatt, S. J., & Ford, R. Q. (1988). Defense mechanisms in the anaclitic and introjective personality configuration. Journal of Consulting and Clinical Psychology, 56, 610–616.

Dawes, R. M. (1979). The robust beauty of improper linear models in decision making. American Psychologist, 34, 571–582.

Fichman, L., Koestner, R., & Zuroff, D. C. (1994). Depressive styles in adolescence: Assessment, relation to social functioning, and developmental trends. Journal of Youth and Adolescence, 23, 315–330.

Franche, R., & Dobson, K. (1992). Self-criticism and interpersonal dependency as vulnerability factors to depression. Cognitive Therapy and Research, 16, 419–435.

Hokanson, J. E., & Butler, A. C. (1992). Cluster analysis of depressed college students’ social behaviors. Journal of Personality and Social Psychology, 62, 273–280.

Jerdonek, P. M. (1980). The dimensionality and construct validity of an ego psychological measure of depressive experiences. Doctoral dissertation, Ohio University. Dissertation Abstracts International, 41, 69B.

Jöreskog, K. G., & Sörbom, D. (1988). LISREL VII. A guide to the program and applications. Chicago: SPSS..

Klass, E. T. (1987). Situational approach to assessment of guilt: Development and validation of a self-report measure. Journal of Psychopathology and Behavioral Assessment, 9, 35–48.

Klein, D. N. (1989). The Depressive Experiences Questionnaire: A further evaluation. Journal of Personality Assessment, 53, 703–715.

Klein, D. N., Harding, K., Taylor, E. B., & Dickstein, S. (1988). Dependency and self-criticism in depression: Evaluation in a clinical population. Journal of Abnormal Psychology, 97, 399–404.

Klein, D. N., Taylor, E. B., Dickstein, S., & Harding, K. (1988). Early-onset primary dysthymia: Comparison with nonbipolar, nonchronic primary major depression on demographic, clinical, familial, personality, and socioenvironmental variables and short-term outcome. Journal of Abnormal Psychology, 97, 387–398.

Kutcher, G. S., & Blaney, P. H. (1991). Dependency, self-criticism, rejection, and failure.Unpublished manuscript, University of Miami, Miami, FL.

Mongrain, M. (1993, August). How are dependent and self-critical individuals perceived by their peers?Paper presented at the 101st Annual Convention of the American Psychological Association, Toronto, Ontario, Canada.

Nietzel, M. T., & Harris, M. J. (1990). Relationship of dependency and achievement/autonomy to depression. Clinical Psychology Review, 10, 279–297.

Petersen, A. C., Compas, B. E., Brooks-Gunn, J., Stemmler, M., Ey, S., & Grant, K. (1993). Depression in adolescence. American Psychologist, 48, 155–168.

Reynolds, W. M. (1986). Assessment of depression in adolescents: Manual for the Reynolds Adolescent Depression Scale. Odessa, FL: Psychological Assessment Resources.

Reynolds, W. M. (1991). Assessment of depression in children: Manual for the Reynolds Child Depression Scale. Odessa, FL: Psychological Assessment Resources.

Riley, W. T., & McCranie, E. W. (1990). The Depressive Experiences Questionnaire: Validity and psychological correlates in a clinical sample. Journal of Personality Assessment, 54, 523–533.

Robins, C. J., Block, P., & Peselow, E. D. (1989). Relations of sociotropic and autonomous personality characteristics to specific symptoms in depression patients. Journal of Abnormal Psychology, 98, 86–88.

Robins, C. J., & Luten, A. G. (1991). Sociotropy and autonomy: Differential patterns of clinical presentation in unipolar depression. Journal of Abnormal Psychology, 100, 74–77.

Viglione, D. J., Clemmey, P. A., & Camenzuli, L. (1990). The Depressive Experiences Questionnaire: A critical review. Journal of Personality Assessment, 55, 52–64.

Wechsler, D. (1991). Wechsler Intelligence Scale for Children-III (WISC-III) manual. San Antonio, TX: Harcourt, Brace, Jovanovich.

Welkowitz, J., Lish, J. D., & Bond, R. N. (1985). The Depressive Experiences Questionnaire: Revision and validation. Journal of Personality Assessment, 49, 89–94.

Wrobel, T. A. (1992). Validity of Harris and Lingoes MMPI subscale descriptors in an outpatient sample. Journal of Personality Assessment, 59, 14–21.

Zuroff, D. C. (1993). Adult attachment styles of individuals vulnerable to depression.Unpublished manuscript, McGill University.

Zuroff, D. C., Moskowitz, D. S., Wielgus, M. S., Powers, T. A., & Franko, D. L. (1983). Construct validation of the Dependency and Self-Criticism scales of the Depressive Experiences Questionnaire. Journal of Research in Personality, 17, 226–241.

Zuroff, D. C., Quinlan, D. M., & Blatt, S. J. (1990). Psychometric properties of the Depressive Experiences Questionnaire. Journal of Personality Assessment, 55, 65–72.

Submitted: May 8, 1995 Revised: December 2, 1996 Accepted: December 9, 1996