11
AGGRESSIVE BEHAVIOR Volume 22, pages 333-343 (1996) Measuring Cognitive Distortion in Antisocial Youth: Development and Preliminary Validation of the “How I Think” Questionnaire Alvaro Q. Barriga and John C. Gibbs Psychology Department, The Ohio State University, Columbus, Ohio .......................................... .......................................... The present study describes the development and preliminary validation of the “HOW I Think” questionnaire (HIT), a new measure of self-serving cognitive distortion. The HIT is based upon Gibbs and Potter’s [Gibbs, 1993; Gibbs et al., 19951 four-category typology of cognitive distortion: Self-centered, Blaming Others, MinimizinglMislabeling, and Assuming the Worst. The preliminary HIT exhibited high test-retest and internal consistency reliability, and generally good construct validity. Correlations between the HIT and self-reported antisocial or externalizing behavior remained significant after controlling for internalizing disorders [Achenbach, 19911. The four typological catego- ries performed comparably well. The HIT was partially successful in discriminat- ing criterion groups. The findings are discussed in terms of theory, research, and appkation. 0 1996 Wiley-Liss, Inc. .......................................... .......................................... Key words: cognitive distortion, antisocial youth, externalizing behavior, internalizing behavior INTRODUCTION Cognitive distortions are nonveridical or inaccurate attitudes, thoughts, and beliefs. The role of self-debasing cognitive distortions in the development of anxiety, depres- sion, and other behaviors related to inhibition and withdrawal has been widely recog- nized by researchers and clinicians for decades [e.g., Beck, 1976; Ellis, 19771. The present research focuses on the measurement of a counterpart phenomenon: cognitive distortions that are self-serving. Self-serving cognitive distortion as a factor in antiso- cial behavior has been identified in theoretical, experimental, and applied literature on aggression and delinquency [e.g., Dodge, 1980;Dodge et al., 1990a;Dodge et al., 199Obl. Gibbs and Potter [1993; Gibbs et al., 1995; see Table I] introduced a distinction be- Received for publication October 10, 1995; accepted January 4, 1996. Address reprint requests to Alvaro Baniga, c/o John C. Gibbs, Psychology Department, The Ohio State University, 142 Townshend Hall, 1885 Neil Avenue, Columbus, OH 43210. 0 1996 Wiley-Liss, Inc.

Measuring cognitive distortion in antisocial youth: Development and preliminary validation of the “how I think” questionnaire

Embed Size (px)

Citation preview

Page 1: Measuring cognitive distortion in antisocial youth: Development and preliminary validation of the “how I think” questionnaire

AGGRESSIVE BEHAVIOR Volume 22, pages 333-343 (1996)

Measuring Cognitive Distortion in Antisocial Youth: Development and Preliminary Validation of the “How I Think” Questionnaire Alvaro Q. Barriga and John C. Gibbs

Psychology Department, The Ohio State University, Columbus, Ohio

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The present study describes the development and preliminary validation of the “HOW I Think” questionnaire (HIT), a new measure of self-serving cognitive distortion. The HIT is based upon Gibbs and Potter’s [Gibbs, 1993; Gibbs et al., 19951 four-category typology of cognitive distortion: Self-centered, Blaming Others, MinimizinglMislabeling, and Assuming the Worst. The preliminary HIT exhibited high test-retest and internal consistency reliability, and generally good construct validity. Correlations between the HIT and self-reported antisocial or externalizing behavior remained significant after controlling for internalizing disorders [Achenbach, 19911. The four typological catego- ries performed comparably well. The HIT was partially successful in discriminat- ing criterion groups. The findings are discussed in terms of theory, research, and appkat ion. 0 1996 Wiley-Liss, Inc.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Key words: cognitive distortion, antisocial youth, externalizing behavior, internalizing behavior

INTRODUCTION

Cognitive distortions are nonveridical or inaccurate attitudes, thoughts, and beliefs. The role of self-debasing cognitive distortions in the development of anxiety, depres- sion, and other behaviors related to inhibition and withdrawal has been widely recog- nized by researchers and clinicians for decades [e.g., Beck, 1976; Ellis, 19771. The present research focuses on the measurement of a counterpart phenomenon: cognitive distortions that are self-serving. Self-serving cognitive distortion as a factor in antiso- cial behavior has been identified in theoretical, experimental, and applied literature on aggression and delinquency [e.g., Dodge, 1980; Dodge et al., 1990a; Dodge et al., 199Obl.

Gibbs and Potter [1993; Gibbs et al., 1995; see Table I] introduced a distinction be-

Received for publication October 10, 1995; accepted January 4, 1996.

Address reprint requests to Alvaro Baniga, c/o John C. Gibbs, Psychology Department, The Ohio State University, 142 Townshend Hall, 1885 Neil Avenue, Columbus, OH 43210.

0 1996 Wiley-Liss, Inc.

Page 2: Measuring cognitive distortion in antisocial youth: Development and preliminary validation of the “how I think” questionnaire

334 Barriga and Gibbs

TABLE I. Cognitive Distortions*

1 . Self-centered: according status to one’s own views, expectations, needs, rights, immediate feelings and desires to such a degree that the legitimate views, etc. of others (or even one’s own long-term best interest) are scarcely considered or are disregarded altogether.

2. Blaming Others: misattributing blame to outside sources, especially: another person, a group, or a momentary aberration (one was drunk, high, in a bad mood, etc.); or misattributing blame for one’s victimization or other misfortune to innocent others.

3. Minimizing/Mislabeling: depicting antisocial behavior as causing no real harm, or as being acceptable or even admirable; or referring to others with a belittling or dehumanizing label.

4. Assuming the Worst: gratuitously attributing hostile intentions to others; considering a worst-case scenario for a social situation as if it were inevitable; or assuming that improvement is impossible in one’s own or others’ behavior.

*Gibbs et al. [1995]. Reprinted by permission.

tween primary and secondary self-serving cognitive distortions. Primary cognitive dis- tortions are self-centered attitudes, thoughts, and beliefs, often illustrated in clinical writings on offenders. For example, Samenow [ 19841 quoted a 14-year-old delinquent: “I was born with the idea that I’d do what I wanted. I always felt that rules and regula- tions were not for me” (p. 160). Yochelson and Samenow [1976, 19771 termed such attitudes “ownership,” defined as a sense of entitlement to whatever one desires [cf. Redl and Wineman, 1957, pp. 154-1551.

Primary cognitive distortions comprise the first category (“Self-Centered”) in Gibbs and Potter’s typological model: the other three categories (“Blaming Others,” “Mini- mizing/Mislabeling,” and “Assuming the Worst”) constitute secondary cognitive dis- tortions, which serve to support the primary distortions (seeTable I). Secondary cognitive distortions have been characterized as pre- or post-transgression rationalizations that serve to “neutralize” conscience or guilt and thereby to prevent damage to the self- image following antisocial behavior [Sykes and Matza, 1957; cf. Bandura, 1991; Redl and Wineman, 1957, p. 1461. Gibbs [1991] suggested that secondary distortions reduce the stresses from the consequences of the primary distortions. Two such stresses that can stem from one’s harm to others are: empathic distress (and possibly empathy-based guilt), and cognitive dissonance between harmful actions and a self-definition as one who does not unjustifiably harm others.

Several researchers have attempted to develop measures of self-serving cognitive distortion. Some measures have specifically addressed secondary cognitive distortion in terms of Sykes and Matza’s [ 19571 neutralizing rationalization construct. The Pro- pensity to Neutralize Measure [Lanza-Kaduce et al., 19831 consisted of three vignettes (e.g., a narrative depicting a boy who commits arson through negligence), on the basis of which subjects make moral evaluations. Neutralizing tendency is inferred from the extent to which respondents are influenced by excuses (e.g., “the fire caused only mi- nor damage”; cf. Minimizing/Mislabeling) in evaluating the wrongness of the behavior.

Using the Propensity to Neutralize measure, Radosevich and Krohn [1981] found that neutralization was inversely related to Kohlbergian moral judgment stage, although this finding was not replicated by Lanza-Kaduce et al. [1983] or Schnell119861. Lanza- Kaduce et al. concluded that the measure is “only a rough indicator” and furthermore had psychometric problems such as poor reliability and limited external validity @. 449).

Instead of Sykes and Matza’s neutralization theory, Yochelson and Samenow’s [ 1976,

Page 3: Measuring cognitive distortion in antisocial youth: Development and preliminary validation of the “how I think” questionnaire

Cognitive Distortion in Antisocial Youth 33.5

19771 clinical description of the “thinking errors” of severe recidivist male offenders was used by Garvin [1990] to construct a measure of cognitive distortion. Garvin de- veloped a 64-item pencil-and-paper questionnaire, the Measure of Automatic Thinking Errors” (MATE), that directly elicited self-report responses. Subjects indicated how often, along a 5-point scale from never to always, a statement “applies to you.” Surpris- ingly, a sample of incarcerated delinquents scored lower overall on the MATE, i.e., evidenced less cognitive distortion, than a control sample of nondelinquents. Garvin speculated that the incarcerated youth anticipated that prosocial responding to posi- tively valenced items (e.g., “I am interested in hearing other people’s points of view”) could yield rewards in the form of better staff treatment or early release from the institution.

Slaby and Guerra [1988] developed an 18-statement Beliefs Questionnaire. Five be- lief categories supporting aggression were used to generate statements: (a) Legitimacy of Aggression (six items, e.g., “It’s O.K. to hit someone if you just go crazy with an- ger,’’ cf. Blaming Others); (b)Aggression Increases Self-Esteem (three items, e.g., “It’s important to show everyone how tough you are by being a good fighter,” cf. Minimiz- ingmislabeling); (c) Aggression Helps to Avoid a Negative Image (three items, e.g., “If you back down from a fight everyone will think you’re a coward,” cf. Assuming the Worst); (d) Victims Deserve Aggression (three items, e.g., “If someone gets beat up, it’s usually his or her own fault,” cf. Blaming Others); and (e) Victims Don’t Suffer (three items, e.g., “People who get beat up badly probably don’t suffer a lot,” cf. Minimizing/ Mislabeling). Although four of the categories evidenced full or partial discriminant validity, one category did not: incarcerated youth reported significantly less agreement with Victims Deserve Aggression beliefs than did non-incarcerated subjects. Hence, the researchers deleted this category of items from the aggregate score. No correlations were reported between the Beliefs scores and quantitative measures of antisocial behavior.

The present study entailed development, preliminary validation, and subsequent re- finement of a new measure of self-serving cognitive distortion, the “HOW I Think” (HIT) questionnaire. We hypothesized that if self-serving cognitive distortion is indeed criminogenic, then the preliminary HIT should correlate with measures of antisocial behavior or externalizing behavior disorders and should discriminate an antisocial cri- terion group from control groups. Other psychometric properties of the HIT (interrater agreement on item subscale placement, subscale properties, internal consistency, test- retest reliability) were also examined.

METHOD Subjects

The sample included 147 male adolescents aged 14-20 years (M = 16.5, SD = 1.21) who were divided into three criterion groups.

Group 1. Fifty-five male adolescents aged 15-20 years (M = 17.1, SD = .90) incar- cerated at an Ohio Department of Youth Services juvenile correctional facility. These juveniles have typically been adjudicated for severe felonies, e.g., aggravated assault, armed robbery, rape, or murder. The group was 43% African-American, 29% Cauca- sian, 4% Asian-American, and 4% Latino, with 20% not reporting race.

Group 2. Fifty male adolescents aged 14-18 years (M = 16.0, SD = 1.09) in grades 10 through 12 at an urban working class public high school. More than half of the

Page 4: Measuring cognitive distortion in antisocial youth: Development and preliminary validation of the “how I think” questionnaire

336 Barriga and Gibbs

student population are bussed in from other areas of the city, which served to minimize the disparity with the offender group in ethnic status and SES. The group was 38% Caucasian, 32%African-American, 4% Asian-American, and 4% Latino, with 22% not reporting race.

Group 3. Forty-two male adolescents aged 14-18 years (M = 16.1, SD = 1.31) in grades 10 through 12 at a suburban upper middle class public high school. The group was 96% Caucasian and 2% Latino, with 2% not reporting race.

Measures The “How I Think” (HIT) questionnaire [Gibbs et al., 19921. The HIT is de-

signed to measure self-serving cognitive distortion in antisocial youth. Subjects re- spond to the HIT questionnaire items along a 6-point agree/disagree scale (from agree strongly to disagree strongly). The preliminary version of the HIT used in this study was comprised of 60 items, 52 of which state nonveridical attitudes or beliefs, e.g., “If someone is careless enough to lose a wallet, they deserve to have it stolen.” Items were generated by the authors based upon relevant clinical writings and direct field experience.

The 52-item pool provides a representation of at least eight items per cognitive distortion category. These items are also classifiable according to the type of anti- social behavior to which they refer. The items refer to one or another of four cat- egories of antisocial behavior derived from the conduct and oppositional-defiant disorders listed in the Diagnostic and Statistical Manual, Fourth Edition [Ameri- can Psychiatric Association, 19941: Stealing; Lying; Physical Aggression; and dis- respect for rules, laws, or authority (i.e., Opposition/Defiance). These categories are evident in the great preponderance of juvenile offenders’ histories, are fairly broad or generic, and are not overly sensitive (respondents may be reluctant to disclose or respond accurately with respect to excessively “sensitive” crimes, e.g., sexual offenses). For example, the HIT item, “People force me to lie when they ask me too many questions” illustrates the application of a Blaming Others cogni- tive distortion to the “Lying” category.

To ensure comparable content and breadth, the items in each cognitive distortion category collectively apply to all four antisocial behavior categories. Specifically, each cognitive distortion category contains at least two and no more than three items refer- ring to a given antisocial behavior category. Hence, HIT items were generated accord- ing to a Cognitive Distortion by Antisocial Behavior, 4 x 4 design. The four cognitive distortion categories and four behavioral categories define eight HIT subscales.

An Anomalous Responding (AR) scale comprised of eight items was designed to screen for inaccurate or suspect responding. For example, disagreement with “Some- times I gossip about other people,” or “I have covered up things that I have done,” can be interpreted as implausible and therefore inaccurate [cf. Crowne and Marlowe, 19641. Such responding may result from various extraneous motives, e.g., approval- or re- ward-seeking (or punishment-avoiding), carelessness, or cognitive or language incom- petence. In any event, protocols evidencing levels of anomalous responding exceeding a certain threshold (determined by psychometric criteria) were excluded from the sample (see Preliminary Analyses). Efforts were made to make the HIT applicable to a wide range of subjects. The HIT requires only a fourth grade reading level as estimated by the Grammatik program [Wampler, 19881. Delinquents reported understanding each item in one-on-one pilot work.

Page 5: Measuring cognitive distortion in antisocial youth: Development and preliminary validation of the “how I think” questionnaire

Cognitive Distortion in Antisocial Youth 337

The adapted “Nye-Short Self-Reported Delinquency Questionnaire” “ye and Short, 1958; Mitchell and Dodder, 19901. As adapted by Mitchell and colleagues [ 1983; Mitchell and Dodder, 19901, this scale entails 11 items that measure frequency of Minor Delinquency (driving without a license, truancy, disobeying parents, getting drunk, and taking things worth less than $2), Predatory Delinquency (seriously damag- ing property, taking things worth more than $20, and check forgery), and Aggressive Delinquency (assault, strong-arm methods, and car theft), as derived from a factor analy- sis. Psychometric properties of the adapted scale are not reported, but some construct validation is implied by its correlation with neutralization tendency and its factor struc- ture. The scale was supplemented in the present study with eight items: four items adapted from the Elliott et al. [ 19831 Delinquent Behavior Questionnaire, and four items referring to positive acts in order to render the scale less abrasive.

The Child Behavior Checklist Youth Self-Report Form [Achenbach, 19911. The Youth Self-Report (YSR) is a multidimensional scale designed to measure adolescents’ ratings of their personal competencies and problems. Achenbach [ 19911 reports reli- ability and validity for the YSR. The Competence Scales (Activities, Social) are de- signed to assess adolescents’ involvement in activities, their social relationships, and their academic performance. We administered the 120 items that comprise the Problem Scales. The Delinquent Behavior and Aggressive Behavior subscales comprise the Ex- ternalizing Scale; and the Withdrawn, Somatic Complaints, and AnxiousDepressed subscales comprise the Internalizing Scale. These aggregate scales (Externalizing and Internalizing) were used in the present study.

Procedure Subjects were tested in two 50-min sessions approximately 1 week apart. The first

testing session included administration of the HIT and Nye-Short questionnaires. The second session included readministration of the HIT as well as completion of the Child Behavior Checklist Youth Self-Report Form.

Group 1 subjects were tested as one group in the institution’s cafeteria during early evening hours. Groups 2 and 3 were tested in their normal academic classroom settings during regular school hours. Female students were also tested so that they could partici- pate in the same activity but were not included in these or any other data analyses.

RESULTS Preliminary Analyses

Ten graduate students from the Ohio State University Psychology Department were recruited to serve as independent judges for item subscale inclusion. Judges were given descriptions of the eight categories (the four cognitive distortion categories plus the four behavioral categories), and then were asked to classify each item first by cognitive distortion category and then by behavioral category. To be considered for inclusion on a subscale an item had to be accurately classified by six out of the ten judges. By the binomial distribution, there is a 1.97% probability that six or more judges will agree by chance.

Each item was then correlated with all eight subscale totals. Only those items which correlated most highly with their pre-designated cognitive distortion or behavioral cat- egories (as determined by judging) were retained in the final subscales, following a test

Page 6: Measuring cognitive distortion in antisocial youth: Development and preliminary validation of the “how I think” questionnaire

338 Barriga and Gibbs

construction strategy suggested by Jackson [ 19701. Items which were not included in the final subscales were retained and included in the overall HIT score.Al1 of theAnoma- lous Responding (AR) scale items showed adequate and comparable correlations with the overall AR scale (ranging from .49 to .65) and were therefore retained in the AR scale.

Subjects were excluded who displayed high levels of anomalous responding (AR scale scores were inverted). A tentative AR scale cut-off was set at 3.5, the midpoint of the agree-disagree continuum on the Likert scale. Scores above 3.5 reflected outliers and can be theoretically construed as reflecting a suspect level of anomalous respond- ing. This cut-off eliminated one subject from group 1, four subjects from group 2, and one subject from group 3.

Main Analyses In general, the preliminary HIT performed well according to several psychometric

criteria. HIT reliability data and correlations with self-reported antisocial behavior were favorable. The four typological categories performed comparably well, as did the four behavior categories. The HIT achieved partial success in discriminating cri- terion groups.

Test-retest reliability for the HIT was high, r (1 35) = .91, P < .0001. Internal consis- tency for the overall HIT was also high, Cronbach’s coefficient alpha = .96. For the HIT cognitive distortion and behavioral subscales, alphas ranged from .78 to .90. The AR scale, with an alpha of .64, demonstrated slightly lower, but adequate internal consistency.

As hypothesized, the HIT correlated highly with self-reported antisocial behavior, both for the total sample and for the three criterion group sub-samples (results are re- ported for the total sample because breakdown by criterion groups revealed highly similar correlation patterns). The HIT correlated with the Externalizing Scale of the Youth Self- Report Form, r (118) = .55, P < .0001, as well as its component subscales, Aggressive Behavior and Delinquent Behavior ( r [ 1181 = .54, P < .001 and r [ 11 81 = .46, respec- tively). The HIT correlated with the Nye-Short Self-Reported Delinquency Question- naire as well, r (126) = .36, P < .0001. A zero-order correlation matrix is reported in Table 11.

The HIT also Correlated with the Internalizing Scale of the Youth Self-Report Form, r ( 1 18) = .30, P < .001. The HIT’s correlation with the Externalizing Scale (r = .55), however, is significantly higher than its correlation with the Internalizing Scale (r = .30) as determined by a t-test for the significance of the difference between dependent r’s (t = 3.57, df = 11 5, P < .01). The Externalizing and Internalizing Scales, themselves, were highly correlated, r (118) = .62, P < .0001. After partialling internalizing scores, the HIT’s correlation with the Externalizing Scale remained highly significant, Y (1 18) = .47, P < .0001, as did the HIT’s correlation with the Nye-Short questionnaire, r (126) = .27, P < .01.

Correlations between the cognitive distortion subscales and self-reported antisocial behavior (i.e., Nye-Short scores and Externalizing Scale scores) were all significant (r’s ranged from .23 to .55; seeTable 11). As expected, the cognitive distortion subscales correlated highly with the overall HIT (r’s ranged from .87 to .92) and also correlated highly with each other (A ranged from .71 to 30). The behavioral subscales of the HIT displayed a similar correlation pattern, but for simplicity and space considerations, they are not reported in the matrix.

Anomalous responding (i.e., the AR scale) was significantly negatively correlated

Page 7: Measuring cognitive distortion in antisocial youth: Development and preliminary validation of the “how I think” questionnaire

Cognitive Distortion in Antisocial Youth 339

TABLE 11. Zero-Order Correlation Matrix* 1 2 3 4 5 6 7 8 N M SD

HIT - 135 2.77 0.83 sc .92d - 135 2.71 0.87 BO .92d .7gd - 135 2.78 0.93

135 2.61 0.88 MM .8gd .8od .78d - 135 3.04 0.97 AW .87d .73d .75d .71d -

135 2.30 0.57 AR -.16 -.I5 -.13 -.lo -.19” - NS .36d .36d .23b .3gd .33d -.32‘ - 126 35.6 12.9 EXT .55d .52d .43d .47d .55d -.33‘ .54d - 118 19.2 10.2 INT .3W .25b .26b .23” .37* -.20” .27b .62d 118 15.9 10.4

*HIT, The “How I Think” Questionnaire; SC, Self-centered; BO, Blaming Others; MM, Minimizing/ Mislabeling; AW, Assuming the Worst; & Anomalous Responding Scale; NS, the adapted Nye-Short Self-Report Delinquency Questionnaire; EXT, Externalizing Scale of the Youth Self-Report; INT, Internal- izing Scale of the Youth Self-Report. “P < .05. bP < .01. “P < .001. dP < .OOol.

with self-reported antisocial behavior, r (1 18) = -.32, P < .001 with Nye-Short scores and r (1 18) = -.33, P < .001 with Externalizing Scale scores, as well as internalizing behavior, r (1 18) = -.20, P < .05 with Internalizing Scale scores. Although the HIT did not correlate negatively with anomalous responding, a significant correlation was ob- tained for one of the cognitive distortion subscales (Assuming the Worst), r (1 18) =

A one-way ANOVA was conducted for each variable across the three criterion groups. A significant main effect was found for every variable with the exception of anomalous responding (i.e., the AR scale). Means and F values are reported in Table 111.

Post-hoc tests by Bonferroni procedure revealed an unexpected pattern of results. As hypothesized, groups 2 and 3 scored lower than group 1 on measures of self-reported antisocial behavior (i.e., Nye-Short scores and Externalizing Scale scores). The same pattern was found for internalizing behavior. However, only group 3 scored signifi- cantly lower than group 1 on the HIT and HIT subscales. Group 2 did not significantly differ from group 1 on these variables, contrary to prediction.

Subsequent Refinement of the “How I Think” (HIT) Questionnaire

Psychometric evaluation of the HIT on the item level guided its subsequent refine- ment. Items were evaluated based on the following criteria: 1) ability to differentiate criterion groups; 2) correlations with self-reported delinquent behavior; 3) lack of confoundance with anomalous responding; and 4) item-total correlations. Several items were eliminated due to their relatively poorer performance according to these criteria. One new item was generated to fill a matrix cell that was left with only one item after the winnowing process (see HIT under Measures for description of matrix). Thus, the current version of the HIT still contains at least two and no more than three items representing each cell of the 4 x 4 design matrix.

The current version of the HIT also entails a modified and expanded Anomalous Responding (AR) scale and several unscored prosocial items or “positive fillers” (e.g.,

-.19, P < .05.

Page 8: Measuring cognitive distortion in antisocial youth: Development and preliminary validation of the “how I think” questionnaire

340 Barriga and Gibbs

TABLE 111. Mean Scores for Criterion Groups and ANOVAAnalyses* Group 1 Group 2 Group 3 F Value

HIT X 2.92 2.92 2.42 5.39b

sc X 2.87 2.86 2.31 6.11b

BO X 2.87 2.98 2.43 4.21“

MM X 2.77 2.73 2.27 4.28”

AW X 3.20 3.14 2.71 3.43a

AR X 2.21 2.36 2.34 1.04(ns)

NS X 47.02 28.46 27.34 70.67’

EXT X 23.43 15.86 ’ 17.51 6.75b

INT X 19.68 13.26 13.92 5.06b

SD 0.82 0.78 0.80

SD 0.90 0.79 0.81

SD 0.94 0.96 0.80

SD 0.82 0.86 0.91

SD 1.02 0.92 0.88

SD 0.59 0.54 0.59

SD 10.45 7.79 7.41

SD 9.59 9.39 10.30

SD 11.03 8.20 10.45

*HIT, The “How I Think” Questionnaire; SC, Self-centered; BO, Blaming Others; MM, Minimizing/ Mislabeling; AW, Assuming the Worst; AR, Anomalous Responding Scale; NS, the adapted Nye-Short Self-Report Delinquency Questionnaire; EXT, Externalizing Scale of the Youth Self-Report; INT, Internal- izing Scale of the Youth Self-Report. “P < .05. b P < .01. “P < .001.

“I am generous with my friends”). The positive fillers were included to counterbalance the negative item content of the cognitive distortion and anomalous responding items and to provide additional camouflage. Cognitive distortion items with relatively low endorsement frequencies were reworded to render them less abrasive. In addition, fre- quently endorsed items were generally slotted toward the beginning of the question- naire. Cognitive distortion, anomalous responding, and prosocial items were evenly distributed throughout the questionnaire to provide maximum variety in item content.

DISCUSSION

A preliminary version of the “How I Think” Questionnaire, a new measure of self- serving cognitive distortion, was found to have generally promising psychometric prop- erties. Reliability (test-retest, internal consistency) results for the HIT and HIT subscales were strong, an important finding because adequate reliability is a prerequisite for va- lidity. Construct validity results were largely favorable. The HIT correlated highly with two self-report measures of antisocial behavior and was partially successful in dis- criminating criterion groups.

The present construct validity results are consistent with theoretical claims that self- serving cognitive distortion is an important factor in accounting for antisocial behavior [Crick and Dodge, 1994; Gibbs, 1993; Gibbs et al., 1995; Slaby and Guerra, 1988; Sykes and Matza, 1957; Vorrath and Brendtro, 1985; Yochelson and Samenow, 1976,

Page 9: Measuring cognitive distortion in antisocial youth: Development and preliminary validation of the “how I think” questionnaire

Cognitive Distortion in Antisocial Youth 341

19771. Cognitive distortion accounted for 30% of the variance in externalizing behav- ior. The comparable results across cognitive distortion categories point to the impor- tance of each of the four categories in Gibbs and Potter’s typology: Self-centered, Blaming Others, Minimizing/Mislabeling, andAssuming the Worst [Gibbs, 1993; Gibbs et al., 19951. Furthermore, the high internal consistency of the HIT may reflect a fairly unitary construct, consistent with clinical descriptions of a distinguishable and coher- ent “criminal mind” [Yochelson and Samenow, 1976, 19771.

The construct validity results of the present study are unlikely to constitute a meth- ods artifact (e.g., because the cognitive and behavioral measures were self-report), be- cause if that were true the HIT would have correlated just as highly with the Internalizing Scale as the Externalizing Scale. The HIT correlated significantly more highly with the Externalizing Scale than the Internalizing Scale, despite the high comorbidity among ex- ternalizing and internalizing behaviors. Furthermore, the correlations between self-serving cognitive distortion and antisocial behavior remained significant after partialling internal- izing scores. Also supportive of construct validity for the HIT as well as its categories was the finding that approximately three-fourths of the HIT items did (as hypothesized) corre- late more highly with their intended subscale than with other subscales.

The HIT had only partial success, however, in discriminating criterion groups. Crite- rion group differences were accurately predicted for group 1 vis a vis group 3. The Blaming Others subscale was fully as discriminative of the incarcerated group as the other HIT subscales, in contrast to the inversion for the similar “Victims Deserve Ag- gression’’ category reported for Slaby and Guerra’s [ 19881 measure. Group 2, however, proved to be problematic. This group reported low levels of delinquency similar to group 3 but high levels of cognitive distortion similar to group 1. We expected that group 2 would score roughly halfway in between the other groups on both measures of antisocial behavior and cognitive distortion.

Several factors may have contributed to this surprising result. It may be that group 2 subjects were under-reporting (minimizing) their actual delinquency frequencies. These urban, ethnically diverse, bussed-in subjects may have had more reason to be suspi- cious of their school system than their suburban, primarily white counterparts and, thus may have been more reluctant to confess to their past transgressions. After all, there are reasons to believe that this group should have reported more delinquency than group 3: 1) a teacher at the group 2 high school commented that many of his students had previ- ously been incarcerated by the Ohio Department of Youth Services, and 2) attendance records revealed unusually high absenteeism rates for this group (almost half of the students had missed more than the tolerated number of absences for that grading period).

With respect to cognitive distortion (HIT scores), group 2 may have responded more accurately. The subjects may have perceived less likelihood of repercussions for report- ing honestly in this area. It seems logical that students would fear repercussions for reporting on what they have done, not how they have thought.

One would still expect, however, more evidence of cognitive distortion in group 1 than group 2. It may be that group 2 subjects were responding fairly honestly to the HIT and that group 1 subjects were downplaying (minimizing) their cognitive distortions. There is reason to suspect that self-presentational bias is a common problem in incar- cerated population [Gamin, 19901. Hence, although there were no significant differ- ences in anomalous responding, the actual level of cognitive distortion for group 1 may have been higher than these incarcerated subjects were willing to report.

Page 10: Measuring cognitive distortion in antisocial youth: Development and preliminary validation of the “how I think” questionnaire

342 Barriga and Gibbs

Future research with the current version of the HIT may help to establish further the empirical status of theoretical claims regarding cognitive distortion. In addition to its usefulness as a research tool, the HIT has the potential for clinical application. It may serve as an assessment device to determine the suitability of cognitive therapy or as a pre-tesvpost-test measure of therapeutic progress. More generally, the theoretical frame- work of the HIT provides a working model of the cognitive distortions which are likely to arise within a cognitive therapy context. Addressing these distortions can help youth experience appropriate affective responses of remorse or guilt. Correcting cognitive distortions should be an important element of any multi-component treatment program [e.g., the EQUIP program, Gibbs et al., 19951.

The present study has provided a generally favorable evaluation of a preliminary version of the HIT. These findings have guided the subsequent refinement of the mea- sure. Further research on a variety of samples is needed to investigate the current HIT’S discriminant validity and to determine whether the new, refined HIT is less susceptible to response biases in the assessment of cognitive distortion in antisocial youth.

ACKNOWLEDGMENTS

We thank David Berenson, Bud Potter, and Larry Weiss for their assistance with this project.

REFERENCES AchenbachTM (1991): “Manual for theyouth Self-

Report and 1991 Profile.” Burlington, V T Uni- versity of Vermont Department of Psychiatry.

American Psychiatric Association (1994): “Diagnos- tic and Statistical Manual of Mental Disorders.” Fourth Edition. Washington, DC: American Psy- chiatric Association.

Ball RA (1968): An empirical exploration of neu- tralization theory. In Lefton M, Skipper JK, McCaghy CH (eds): “Approaches to Deviance: Theories, Concepts, and Research Findings.” New York Appleton-Century-Crofts, pp 255- 265.

Bandura A (1991): Social cognitive theory of moral thought and action. In Kurtines WM, Gewirtz JL (eds): “Handbook of Moral Behavior and Devel- opment: Vol. 1. Theory.” Hillsdale, NJ: Erlbaum,

Beck AT (1976): “Cognitive Therapy and the Emo- tional Disorders.” New York: International Uni- versities Press.

Crick NR, Dodge KA (1994): A review and refor- mulation of social information-processing mechanisms in children’s social adjustment. Psy- chological Bulletin 11574-101.

Crowne D, Marlowe D (1964): “The Approval Mo- tive: Studies in Evaluative Dependence.” New York: John Wiley & Sons.

Dodge KA (1980): Social cognition and children’s

pp 45-103.

aggressive behavior. Child Development 5 1 :

Dodge KA, Bates JE, Pettit GS (1990a): Mechanisms in the cycle of violence. Science 250: 1675-1685.

Dodge KA, Price JM, Bachorowski JA, Newman JP (1990b): Hostile attributional biases in severely aggressive adolescents. Journal of Abnormal

Elliott DS, Ageton SS, Huizinga D, Knowles BA, Canter RJ (1983): “The Prevalence and Incidence of Delinquent Behavior: 19761980.’’ Boulder, CO: Behavioral Research Institute.

Ellis A (1977): Rational-emotive therapy: Research data that support the clinical and personality hy- potheses of RET and other modes of cognitive- behavioral therapy. Counseling Psychologist 7:2-42.

Garvin LM (1990): “Measure of Automatic Think- ing Errors: The Relationship Between Errors in Thinking and Antisocial Behavior.” Unpublished doctoral dissertation, Syracuse University, New York.

Gibbs IC (1991): Sociornoral developmental delay and cognitive distortion: Implications for the treatment of antisocial youth. In Kurtines WM, Gewirtz JL (eds): “Handbook of Moral Behav- ior and Development: Vol 3. Application.” Hillsdale, NJ: Lawrence Erlbaum Associates, pp

162-170.

Psychology 99:385-392.

95-1 10.

Page 11: Measuring cognitive distortion in antisocial youth: Development and preliminary validation of the “how I think” questionnaire

Cognitive Distortion in Antisocial Youth 343

Gibbs JC (1993): Moral-cognitive interventions. In Goldstein AP, Huff CR (eds): “The Gang Inter- vention Handbook.” Champaign, IL: Research Press, pp 159-185.

Gibbs JC, BarrigaAQ, Potter G (1992): “The How I Think Questionnaire.” Unpublished question- naire, The Ohio State University.

Gibbs JC, Potter G, Goldstein AP (1995): “The EQUIP Program: Teaching Youth to Think and Act Responsibly Through a Peer-Helping Ap- proach.’’ Champaign, IL: Research Press.

Jackson D (1970): A sequential system for person- ality scale development. In Spielberger CD (ed): “Current Topics in Clinical and Community Psy- chology, Vol2.” Academic Press: New York.

Kahn TJ, Chambers HJ (1991): Assessing reoffense risk with juvenile sexual offenders. Child Wel- fare 70:333-345.

Lanza-Kaduce L, Radosevich M, Krohn MD (1983): Cognitive moral development, neutralizing defi- nitions, and delinquency. In Laufer WS, Day JM (eds): “Personality Theory, Moral Development, and Criminal Behavior.” Lexington, MA: Lex- ington Books.

Mannle H (1972): “An Empirical Exploration and Interpretation of NeutralizationTheory Predicted Upon Sexual Differences in the Socialization Process.” Unpublished doctoral dissertation, Florida State University, Tallahassee.

Mitchell J, Dodder RA (1983): Types of neutraliza- tion and types of delinquency. Joumal of Youth and Adolescence 12307-318.

Mitchell J, Dodder RA, Norris TD (1990): Neutral- ization and delinquency: A comparison by sex and ethnicity. Adolescence 25:487497.

Nye FI, Short JF, Jr (1958): Scaling delinquent be-

havior. American Sociological Review 22: 326-33 1.

Radosevich M, Krohn MD (198 1): Cognitive moral development and legal socialization. Criminal Justice and Behavior 8:401424.

Redl F, Wineman D (1951): “Children Who Hate.” Glencoe, IL: Free Press.

Redl F, Wineman D (1957): “The Aggressive Child.” New York: Free Press.

Samenow SE (1984): “Inside the Criminal Mind.” New York: Random House.

Schnell SV (1986): “Delinquents With Mature Moral Reasoning: A Comparison With Delayed Delin- quents and Mature Nondelinquents.” Unpub- lished doctoral dissertation, the Ohio State University, Columbus.

Slaby RG, Guerra NG (1988): Cognitive media- tors of aggression in adolescent offenders: l . assessment. Developmental Psychology 24: 580-588.

Sykes GM, Matza D (1957): Techniques of neutral- ization: A theory of delinquency. American So- ciological Review 22:664670.

Trachtenberg S, Viken RJ (1994): Aggressive boys in the classroom: Biased attributions or shared perceptions? Child Development 65:829-835.

Vorrath HH, Brendtro LK (1985): “Positive Peer Culture” (2nd. Ed). New York: Aldine.

Wampler BE (1988): “Grammatik.” San Francisco, CA: Reference Software.

Yochelson S, Samenow SE (1976): “The Criminal Personality:AProfileForChange”(Vol. 1). New York: Jason Aronson.

Yochelson S, Samenow SE (1977): “The Criminal Personality: The Change Process” (Vol. 2). New York: Jason Aronson.