Assessment OnlineFirst, published on February 18, 2009 as doi:10.1177/1073191108328890

Assessment Volume XX Number X Month XXXX xx-xx © 2009 Sage Publications Development and Initial Validation of 10.1177/1073191108328890 http://asmnt.sagepub.com hosted at the Disinhibition Inventory http://online.sagepub.com A Multifaceted Measure of Disinhibition

Lilian Dindo Elizabeth McDade-Montez Leigh Sharma David Watson Lee Anna Clark University of Iowa

The broad personality trait of disinhibition reflects the tendency to behave in an underconstrained versus overconstrained manner and is associated with externalizing psychopathology and risk-taking behaviors. This article describes the development and initial validation of the Disinhibition Inventory (DIS-I), a multifaceted measure of disinhibition that helps explicate the nature of this important higher-order dimension more fully. Factor analyses of an initial item pool resulted in five content-distinct, yet correlated scales measuring both high (Manipulativeness, Distractibility, Risk Taking) and low (Prosociality, Orderliness) levels of disinhibition that cross-validated in an independent sample. Evidence for the construct validity of the DIS-I is presented, including convergent and discriminant relations with Big-Three and Big-Five/five-factor model measures of personality. Results indicate that the DIS-I scales are associated most strongly with other measures of disinhibition, but that the DIS-I additionally contains content absent in extant adult measures of disinhibition that may prove useful in the assessment of externalizing psychopathology.

Keywords: disinhibition; externalizing; personality; trait; psychopathology

he personality trait of disinhibition versus con- motivated by and comply with social conventions and Tstraint is a broad dimension within the Big-Three norms, are straightforward and highly disciplined, tradition of personality, tapping individual differ- lack spontaneity, and plan carefully before acting. ences in the tendency toward underconstrained versus Within the Big-Five or five-factor model (FFM)1 of overconstrained behavior (Watson & Clark, 1993). personality, both low conscientiousness (C; e.g., care- Disinhibited individuals are driven by stimuli of the less, disorganized, and undisciplined ) and low agree- immediate moment rather than by longer term conse- ableness (A; e.g., inconsiderate, noncompliant, and quences of their behavior. Such individuals tend to brash) are empirically and theoretically related to dis- act in accord with their current feelings, often are inhibition, as evidenced in recent analyses of normal careless and disorganized, recklessly pursue thrilling and abnormal personality traits (Markon, Krueger, & and often dangerous experiences, and have little Watson, 2005). regard for the concerns of others. In contrast, con- Evidence for the external validity of disinhibition strained individuals generally are controlled by and related constructs (e.g., sensation seeking) comes longer term consequences of their behavior, are more from research on its associations with a broad range of characteristic behaviors. For example, highly dis- inhibited individuals are more likely to use and abuse Authors’ Note: Correspondence may be addressed to Lee Anna drugs and alcohol (Kosson, Steuerwald, Newman, & Clark, Department of Psychology, University of Iowa, E11 SSH, Iowa City, IA 52242; phone: 319-335-3391; e-mail: la-clark@ Widom, 1994; McGue, Slutske, Taylor, & Iacono, uiowa.edu. 1997) and to engage in antisocial and risky behavior

1 2 Assessment

(Frick, Kuper, Silverthorn, & Cotter, 1995), including (1984) four-factor model of temperament also includes promiscuous sexual behavior (Miller & Lynam, 2003) an factor consisting of four subscales: and gambling (McCormick, 1993). They also tend to (a) Inhibitory Control, (b) Decision Time, (c) Sensation perform more poorly in college, even after controlling Seeking, and (d) Persistence. for high school grades and Scholastic Aptitude Test As mentioned earlier, the Big-Three dimension of scores (Watson & Clark, 1993). In contrast, con- disinhibition represents a combination of the FFM’s strained individuals exhibit higher levels of spirituality (low) C and A dimensions: “C” primarily reflects impulse and religious conservatism (Watson & Clark, 1993). control and self-discipline, whereas “A” reflects an Disinhibition has been linked to both alcoholism empathic and sensitive interpersonal style (Watson, and antisocial (ASPD) through Clark, & Harkness, 1994). Clark and Watson (1999) such commonalties as deficits in the ability to delay factor analyzed the Disinhibition items from their gratification, tendency to engage in sensation-seeking General Temperament Survey (GTS; Clark & Watson, behavior, and boredom proneness in the absence of 1990) and found evidence of two subfactors: Carefree high levels of stimulation (Sher & Trull, 1994). In Orientation and Antisocial Behavior, broadly corre- addition, risk taking was a consistent predictor of sponding to the FFM traits of (low) C and (low) A, extensive delinquency in a cross-sectional study of respectively. Thus, disinhibition includes weak social male and female criminal offenders (Krueger et al., inhibition. 1994), and Caspi, Moffitt, Newman, and Silva (1996) found that children identified as undercontrolled at Proposed Facets of Disinhibition age 3 were especially likely to have developed alco- hol problems, to be involved in crime, and to have Although there is general agreement regarding the ASPD at age 21. Thus, relevant associations have nature of higher order traits of personality (e.g., the been found in cross-sectional, prospective, and exper- Big Three and Big Five), there currently is no agreed imental studies, leading Krueger and colleagues on conceptualization of the lower order components (Krueger et al., 2002) to propose a model linking dis- that comprise these superfactors. For example, Watson inhibition with externalizing psychopathology (i.e., and Clark (1993) identified nine content areas with ASPD, alcohol and drug dependence, and conduct convergent/discriminant validity relative to this Big- disorder) based on a common genetic vulnerability. Three domain, whereas their Disinhibition Scale The existence of this common vulnerability has received scores only two subfactors (Clark & Watson, 1999). Clark considerable support subsequently from family, psy- (1993) identified three specific traits—Impulsivity, chophysiological, and structural studies. Thus, a Propriety, and Workaholism—that are strongly greater understanding of the nature and structure of this related to the higher order Disinhibition factor in her construct is clearly important to both normal and Schedule for Nonadaptive and Adaptive Personality abnormal personality research. (SNAP), as well as two others (Manipulativeness and Aggression) that frequently have strong secondary Major Models of Personality (and, in some samples, primary) loadings on this and Disinhibition factor. Tellegen’s (in press) higher order constraint factor also has three scales, but only two are the same A number of prominent Big-Three structural mod- as in the SNAP: Control and Traditionalism parallel els of personality have identified disinhibition, or a (low) Impulsivity and Propriety, but the SNAP has no conceptually and empirically related trait, as one of scale paralleling Harm Avoidance. Zuckerman’s three broad higher order factors, including Eysenck’s (1979) Sensation Seeking Scale is composed of four Psychoticism (Eysenck & Eysenck, 1975), Tellegen’s related facets—Disinhibition, Thrill and Adventure (1985) Constraint (i.e., low Disinhibition), Watson Seeking, Experience Seeking, and Boredom and Clark’s (1993) Disinhibition Versus Constraint, Susceptibility, whereas Whiteside and Lynam’s and Gough’s (1987) Norm-Favoring Versus Norm- (2001) UPPS model also includes four traits— Questioning (Gough, 1987; see Clark & Watson, Urgency, Premeditation, Perseverance, and Sensation 1993). Similarly, Rothbart and Ahadi (1994) identi- Seeking—but, again, these only partially overlap fied Effortful Control—consisting of Attention Zuckerman’s four. This may be, at least in part, Focusing, Inhibitory Control, Low-Intensity Pleasure, because the UPPS was developed explicitly to assess and Perceptual Sensitivity—in their three-factor traits associated with “impulsive-like behavior,” so it model of childhood temperament. Buss and Plomin’s does not include much, if any, content related to “A.” Dindo et al. / Development of a Measure of Disinhibition 3

Finally, the Revised NEO Personality Inventory behavior (see Whiteside & Lynam, 2001). Second, (NEO PI-R; Costa & McCrae, 1992) includes six the exact number and nature of its lower order com- facet-level traits each for C and A. Thus, the lower order ponents remain unclear, although research consis- facets that comprise the broad Disinhibition factor in tently links certain more specific traits (e.g., risk recent work are varied, range from 2 to 12, and, as taking, aggression vs. responsibility, and conformity) such, are not well established. Therefore, investigation with the broad disinhibition domain. Finally, research of this trait is important for personality models and also indicates that although the factor is broad in research; furthermore, disinhibition’s substantial rela- scope, and is composed of diverse and relatively dis- tion to psychopathology makes it an important trait tinct content areas, it nonetheless clearly has predic- for clinical research as well. tive validity in relation to psychopathology and risky behavior. Thus, improving our understanding and Empirical Findings on the measurement of this trait has significant implications Structure of Disinhibition on practical as well as theoretical levels. In this article, we report on the development and ini- Two studies have examined the underlying structure tial and cross-validation of the Disinhibition Inventory of disinhibition from a hierarchical perspective. (DIS-I), which we undertook to explicate the nature of Zuckerman, Kuhlman, and Camac (1988) factor ana- disinhibition, specifically the structure of its lower lyzed 46 scales from 8 personality self-report question- order components, building on existing conceptualiza- naires and examined three-, five-, and seven-factor tions of the construct. Although several scales within solutions in a sample of 271 students. The three-factor current Big-Three and Big-Five instruments measure solution consisted of Extraversion (E), Neuroticism disinhibition or related constructs, none provide thor- (N), and a Disinhibition factor defined by several strong ough faceted measurement of the full broad domain. markers: sensation seeking, risk taking, aggression, and Our aim was to develop a multifaceted stand-alone impulsivity all had strong positive loadings, whereas measure to allow for further explication of the nature of socialization, responsibility, restraint, and social desir- disinhibition and for targeted comprehensive measure- ability had high negative loadings. Zuckerman et al. ment of this clinically significant personality trait in (1988) labeled this factor “Impulsive-Unsocialized- both research and applied settings. Given the well- Sensation-Seeking.” In the five-factor solution, this supported relations between disinhibition and external- broad factor broke into two subfactors and then, in the izing psychopathology/risk-taking behaviors, further seven-factor solution, into four subfactors: (a) Disinhibited understanding the nature of this construct may prove Aggression versus Responsibility, (b) Impulsivity versus clinically significant in conceptualizing, identifying, Cognitive Structure, (c) Autonomy versus Conformity, and treating such disorders/behaviors. and (d) Anger versus Restraint. Markon et al. (2005, Study 1) conducted a series of meta-analytic factor analyses of four commonly Overview used personality inventories. The three-factor solu- tion yielded the Big Three, with factors that could be Loevinger’s (1957) three basic steps in the scale identified as Neuroticism or Negative Temperament development process—substantive, structural, and (N/NT), Extraversion or Positive Temperament external validation—provided a useful framework. (E/PT), and Disinhibition, with the last marked by Substantive validity—the extent to which an instru- such content as agreeableness, cooperativeness, con- ment’s items accurately reflect the underlying domain— trol, and harm avoidance versus psychoticism was ensured by a systematic sampling of all content that (Eysenck & Eysenck, 1975), stimulus seeking, cal- is potentially relevant to the target construct. lousness, conduct problems, aggression, and novelty Structural validation—the extent to which an seeking. At the four-factor level, this broad factor instrument’s internal structure reflects the empirical broke into factors identifiable as A and C. structure of the target domain—was pursued through A number of conclusions can be drawn from this factor analyses, with the resulting factors providing the research. First, disinhibition is clearly related to the basis for scale creation. Finally, external validation— FFM traits of A and C; note that this distinguishes it establishing that relations between an instrument and from the related construct of impulsivity/impulsive external criteria are consistent with theory—was behavior, which is linked to C, E, and N, depending evaluated by examining the DIS-I scales’ structural on the motivational system underlying the impulsive relations with established measures of personality. 4 Assessment

We collected data on two large undergraduate sam- rejection versus traditionalism, (d) danger seeking ples. We used Sample 1 to develop the DIS-I scales, versus harm avoidance, (e) selfishness versus , and we first describe the substantive and initial struc- (f) goal pursuit, (g) exploitation, (h) playfulness ver- tural validity stage (i.e., the item-level structure of the sus seriousness, (i) lack of self-control/ego mastery ver- instrument). We then report on the DIS-I’s structural sus self- discipline, (j) aggression, (k) disorganization validity in the cross-validation Sample 2, and on the versus organization/orderliness, and (l) inattention external validation stage in both samples. Specifically, versus attention/effortful control. we examined (a) the psychometric properties of the The initial item pool consisted of 154 items, with each scales, including their internal consistency reliabili- of the 12 content domains represented by at least eight ties, gender differences, and scale-level intercorrela- items to ensure adequate coverage. Participants rated tions and factor structure and (b) structural relations themselves on the items using a 5-point Likert-type scale with other measures to investigate convergent and ranging from strongly disagree to strongly agree. discriminant validity. Other Measures Method NEO PI-R—Costa and McCrae (1992). The NEO PI-R, which consists of 240 items rated on a 5-point Participants and Procedures Likert-type scale from strongly disagree to strongly agree, was used to assess the FFM personality traits Development Sample 1. During two semesters, 565 of N, E, Openness (O), A, and C. Each of the five (70% female) undergraduate students at a large mid- higher order traits is composed of six lower order western university were recruited from the ele- facets (see Table 6). Costa and McCrae (1992) mentary psychology course. After giving informed reported strong internal consistency reliabilities for consent, they participated in small group sessions in the five higher order domain scales (coefficient alpha, partial fulfillment of a course research-exposure range = .86 to .92) and strong test–retest reliabilities requirement, completing a battery of personality and during a 6-year period (range = .68 to .83). In the cur- behavioral questionnaires, including the DIS-I item rent sample, alphas were .89 for N, .91 for E, .86 for pool, a subset of which we report on in this article. O, .90 for C, and .88 for A. The mean age of participants was 19.4 and 89% were Caucasian. GTS—Clark and Watson (1990). The GTS is a 90-item true–false format self-report personality Cross-validation Sample 2. During a calendar inventory designed to assess the Big-Three dimen- year, 515 (66% female) undergraduate students sions of NT, PT, and Disinhibition. As mentioned ear- enrolled in various psychology courses at a large mid- lier, the Disinhibition scale contains two subscales, western university participated for extra credit or as Antisocial Behavior and Carefree Orientation, which partial fulfillment of a course requirement. After giv- are strongly negatively correlated with NEO-PI-R A ing informed consent, participants completed a bat- and C, respectively (Clark & Watson, 1999). The PT tery of personality and behavioral questionnaires, scale also contains two subscales, Energy and including the DIS-I, a subset of which we report on in Positive Affect. Watson and Clark (1993) reported this article. See Wensman and Clark (2008) for detailed strong internal consistency reliabilities (coefficient information regarding procedures of the larger alphas = .85 to .90) for the three scales, which are project. Because of an oversight, data on age and eth- largely independent of one another, and strong nicity were not collected, but sample demographics test–retest reliabilities during a 2-month period (.72 are likely to be highly similar to those of development to .80). In the current sample, alphas were .88 for NT, Sample 1. .85 for PT, and .84 for Disinhibition. The subscale Substantive Validity: Item Pool Development alphas ranged from .71 to .79. We addressed Loevinger’s (1957) substantive Schedule for Nonadaptive and Adaptive validity step through a thorough review of the litera- Personality-2 (SNAP-2; Clark et al., in press). The ture, from which we identified 12 potentially relevant SNAP-2 is a 390-item true–false format self-report content domains: (a) impulsivity versus deliberation, questionnaire designed to assess personality charac- (b) irresponsibility versus dependability, (c) norm teristics relevant to both the normal and abnormal Dindo et al. / Development of a Measure of Disinhibition 5 range. The SNAP-2 yields scores on three broad tem- next examined the 4- to 7-factor solutions. The final perament and 12 more specific trait scales. The three one or two factors, respectively, of the 6- and 7-factor temperament scales—NT, PT, and Disinhibition—are solutions lacked clear markers and also had high those of the GTS. Associated with NT are the lower cross-loadings with other factors. The 4- and 5-factor order traits of mistrust, manipulativeness, aggression, solutions, however, contained multiple clear markers self-harm, eccentric perception, and dependency. The for each factor and clear boundaries across factors. lower order traits associated with PT are exhibitionism, The first factor in these solutions was relatively large, entitlement, and (low) detachment. Finally, the lower with content from two and three, respectively, of the order traits associated with Disinhibition are impul- rational content domains. Eigenvalues for Factors 1 sivity, (low) propriety, and (low) workaholism. In through 12 were as follows: 21.7, 11.2, 6.5, 5.8, 3.0, addition, SNAP-2 Disinhibition can be divided into 2.5, 2.1, 1.8, 1.6, 1.6, 1.5, and 1.5. subscales Antisocial Behavior and Carefree Orientation. The five-factor solution was the most differentiated The SNAP-2 scales are internally consistent and tem- solution with clear factors, so it was determined ini- porally reliable: Median internal consistency reliability tially to provide the best fit to the data. Factor 1 in this (Cronbach’s coefficient alpha) was .81 in a community- solution contained content relating to two types of dwelling adult sample and .82 across several patient prosocial behavior: achievement striving and consider- samples. Median short-term reliability in a normal ateness toward others. Factor 2 included items measur- adult sample was .88 for 1 week to 1 month and .85 ing manipulativeness and selfishness. Factor 3 items between 1 month and 4.5 months (Clark et al., in related to attentional control and distractibility. Factor press). In the current Sample 2, scale reliabilities 4 contained risk-taking and sensation-seeking items. ranged from .91 (NT) to .75 (Propriety and both Finally, Factor 5 assessed organization and orderliness Disinhibition subscales); median = .82. Creation of the final DIS-I scales. We first created four scales to capture Factors 2 through 5 by select- Structural Validity Results ing items that were the most distinct markers of the factors (minimum loading of .30) and that had mini- Scale Development mal cross-loadings on other factors (<.10 than the pri- Analytic approach. To examine the general struc- mary loading). We also removed a subset of Factor 2 ture of the DIS-I, we conducted a series of factor items tapping anger and aggression that showed poor analyses with the initial 154-item pool.2 In determin- discriminant validity from N/NT. This process ing how many dimensions to extract, we had four yielded scales we labeled Manipulativeness, Risk main considerations. First, we examined the scree Taking, Distractibility, and Orderliness. plot for a distinct break in its slope. Second, we Given the two distinct types of content in the first looked for psychologically meaningful, clearly inter- large Prosocial factor, we performed additional factor pretable factors. Third, we retained factors that were analyses separately on this factor’s items. Using the well defined and reasonably distinct from one same criteria described above, these analyses resulted another. More specifically, we retained factors that in two clear factors reflecting the distinct content had a number of clear markers, that is, items with domains of Considerateness and Goal Orientation. loadings of |.40| or greater on the target factor and Scales created from these factors were moderately below |.35| on all other factors. Fourth, we retained strongly correlated (r = .55); however, because they factors that were robust and generalizable across rota- measured such clearly separable content, we retained tions (i.e., promax and varimax) and extraction meth- them provisionally as subscales of a broader ods (i.e., principal axis factoring [PAF] and principal Prosocial Behavior scale. components analysis [PCA]). It is noteworthy that the original 12 content domains (i.e., those rationally derived from the liter- Determination of number of factors. As stated ear- ature), although not represented by 12 scales, all are lier, the original item pool was organized rationally adequately modeled within the final five scales. That into 12 content domains with at least eight candidate is, rational scales created from the items of each con- items each, so we first extracted 12 factors. However, tent domain had at least one moderate to strong cor- factors 8 through 12 were poorly defined (i.e., no relation with the final DIS-I scales, with coefficients loadings greater than |.39|) with few markers, so we ranging from .52 to .95. Interestingly, however, the 6 Assessment

Table 1 Promax-Rotated Factor Loadings for a Five-Factor Solution of the DIS-I’s Final 65 Items

Item Content 12345

Hard time staying focused .80 .04 −.01 .04 −.04 Easily distracted .74 .07 .03 .06 .03 Lose train of thought easily .70 .08 −.03 .06 −.10 Distracted while working on projects .70 .03 .03 .05 .00 Distractions throw off my concentration .70 .03 −.01 .06 −.05 Short attention span .64 .14 −.08 .05 .07 Can focus attention easily .60 −.21 −.26 −.02 .00 Difficult to distract me while working .56 −.13 .07 −.05 −.12 Can concentrate for long time .53 −.10 −.29 .02 .03 I procrastinate a lot .49 .05 .17 −.18 .09 Play around when I should work .49 .17 .19 −.13 .16 I get bored easily .47 .25 .03 −.03 −.07 Trouble sitting still .40 .07 −.02 .05 .08 Often forget things I need to do .39 .21 −.02 −.15 .01 Can tune out everything around me .38 −.14 −.08 .02 −.08 Can easily control urges .34 −.09 −.20 .07 .14 Easy to take advantage of others −.06 .65 −.13 .04 .02 Lie to get what I want .16 .61 .04 .00 −.01 I manipulate others −.09 .60 −.14 .01 .00 Enjoy pointing out others’ mistakes −.01 .59 .06 .01 −.11 Do what it takes to get my way .06 .59 −.14 .02 .03 Beat the system to get what I want −.02 .58 .22 −.02 .12 Willing to use others −.06 .56 −.11 .03 .00 Telling a joke at others’ expense .02 .56 −.02 −.11 −.07 Get people to do me favors .08 .48 .24 .04 −.01 Take teasing too far .08 .48 .01 −.05 −.02 Make others do my work .22 .47 −.16 −.01 −.08 Often act selfishly .20 .47 −.12 .11 −.06 It’s not cheating if do not get caught −.03 .46 −.12 .05 .14 Ok if my happiness at expense of others .01 .43 −.14 .05 −.07 Take credit for what others did .01 .41 −.14 −.01 .01 Like gambling for high stakes −.12 .36 −.06 −.06 .24 Have high standards of achievement −.20 .08 .60 −.01 −.09 I am a considerate person .11 −.20 .58 .01 .04 Important to succeed in work −.02 .00 .56 .07 −.10 Able to focus when needed −.29 .15 .56 −.02 −.09 Attentive to others’ feelings .13 −.16 .54 .06 .06 Do not give up easily −.27 .04 .53 −.05 .11 When set goal, follow through −.25 .00 .52 .08 .00 Like to be best in everything I do −.16 .23 .52 .04 .04 Try to see others’ perspective .19 −.16 .52 −.11 −.02 Hate to see others’ pain .21 −.17 .51 .02 −.01 Care about others’ problems .10 −.28 .49 −.06 .06 Extra hours on project −.17 −.02 .49 −.02 −.07 Would not hurt another purposely .14 −.22 .49 .04 −.02 Careful about others’ feelings .08 −.21 .48 .05 −.02 I am easygoing .05 −.01 .48 −.05 .15 Always meet deadlines −.14 .02 .47 .13 −.02 Like home neat and tidy .01 .00 .03 .75 −.01 Living space is usually messy −.13 −.17 −.22 .71 .09 Keep things organized −.03 .03 .20 .68 .00 Bothered by clutter .14 .10 .03 .68 −.02 Clean things regularly −.04 .05 .00 .62 .05 Organize things systematically .03 .06 −.07 .55 .05 (continued) Dindo et al. / Development of a Measure of Disinhibition 7

Table 1 (continued)

Item Content 12345

I like “to do” lists .14 −.08 .17 .42 −.01 Carefully plan daily activities −.07 .14 .16 .40 −.16 Make a list before shopping .13 −.10 .07 .39 .00 Enjoy taking risks .02 .10 .14 .09 .69 I avoid dangerous situations −.07 −.14 −.18 −.02 .68 Seek thrilling experiences .01 .03 .15 .08 .61 Play it safe rather than take chance −.05 −.11 −.11 −.03 .57 Prefer to do things that are safe −.11 −.06 −.27 .01 .57 Like situations that involve danger .00 .09 .05 .04 .55 Like to be spontaneous .14 −.01 .24 −.01 .50 Make decisions on spur of moment .09 .14 .13 −.09 .48 Note: N = 565. Correlations ≥ .35 are in bold. DIS-I = Disinhibition Inventory. impulsivity content split across multiple factors We then ran a PCA on each random data set, for a rather than forming its own distinct dimension. We total of 50 analyses, recording eigenvalues for each. address this issue further in the Discussion section. The mean and 95th percentile of all eigenvalues were computed, yielding m = 1.7 and the 95th percentile = Item-Level Structure 1.8. Thus, the PA indicates that only factors with Development Sample 1. Factor loadings of the DIS- eigenvalues >1.8 should be retained. As reported I’s 65 items based on a PCA with promax rotation are above, the lowest eigenvalue for the five-factor DIS-I shown in Table 1.3 The 16 marker items of the solution was 2.3, and that for the six-factor solution Distractibility scale, containing content related to dis- was 1.4, indicating that extracting five factors was tractibility, attentional control, and concentration, load appropriate. on Factor 1. Factor 2 includes the 16 items comprising the Manipulativeness scale, with content related to self- Cross-validation Sample 2. We subjected the 65 DIS- ishness, manipulativeness, and taking advantage of I items to a PAF analysis with promax rotation. On the others. Factor 3, Prosociality, contains content related basis of the Sample 1 results, we examined the five- to both achievement striving and kindness toward factor solution, which accounted for a total of 82% of others. As in the initial five-factor solution, these con- the variance. Results revealed that the structure is highly tent sets failed to define distinct factors in the overall reliable. That is, consistent with Study 1, the five factors factor analysis; however, analyses using only the Factor were marked by items tapping distractibility, manipula- 3 items revealed two clear content-specific subfactors. tiveness, a hard-working and considerate interpersonal Therefore, we retained the two subscales, Goal style, orderliness, and thrill-seeking items, respectively. Orientation and Considerateness, each containing eight To assess the precise level of factor-structure sim- markers to examine their convergent and discriminant ilarity across the two data sets, correlations between validity in relation to other variables. Factor 4 includes regression-based factor scores of the two data sets content related to tidiness and need for organization; its were examined. When five factors were extracted, nine items comprise the Orderliness scale. Factor 5 and factor scores were computed in each data set content includes thrill seeking, need for excitement, and using the resulting two sets of factor scoring spontaneity; its eight items comprise the Risk Taking weights, corresponding factor scores correlated .98 scale. Eigenvalues for Factors 1 through 10 are as to .99 in each of the two data sets (see Table 2). follows: 9.7, 6.2, 4.5, 3.9, 2.3, 1.4, 1.4, 1.3, 1.2, and 1.2. Similarly, convergence of a two-factor structure We also ran parallel analysis (PA) based on guide- using just the Prosociality items was also very high lines by Hayton, Allen, and Scarpello (2004) to (rs = .99). In summary, cross-validation factor examine the appropriateness of extracting a analyses of the DIS-I items in an independent sam- five-factor solution.4 We began by generating 50 ran- ple drawn from the same population indicated that dom data sets with the same number of participants the structure of five scales and two subscales was (N = 565) and variables (k = 65) as the actual data set. virtually identical. 8 Assessment

Table 2 Correlations Between Factor Scores of the Two Samples

Study 1 Weights

Study 2 Weights 12345

Applying weights to the Study 1 data (N = 565) 1. Distractibility .99 .08 −.03 .04 .08 2. Manipulativeness .00 .99 −.15 .04 .03 3. Prosociality .01 −.07 .99 .08 .07 4. Orderliness .03 .04 .06 .99 −.07 5. Risk taking .04 .02 .04 −.07 .98 Applying weights to the Study 2 data (N = 515) 1. Distractibility .99 .04 −.03 −.04 −.02 2. Manipulativeness −.02 .99 −.03 −.05 .06 3. Prosociality −.02 .05 .99 .04 .04 4. Orderliness −.06 −.05 .00 .99 .00 5. Risk Taking −.05 .06 .01 −.01 .98 Note: Varimax rotation. Correlations ≥ .35 are in bold.

Psychometric Properties from .41 and .39 (Risk Taking) to .72 and .71 Internal consistency reliability. As shown in Table (Manipulativeness) in the two samples, respectively, 3, the internal consistency reliabilities—in both the and were highly congruent: All differences were development and cross-validation samples, respec- between .00 and .02, with two exceptions (Total Goal tively—of the total DIS-I score (αs = .90 and .89), Orientation rs were –.65 and –.57, and Total scales (α ranges = .81 to .88 and .82 to .87), and sub- Orderliness rs were –.50 and –.57 in Samples 1 and 2, scales (α ranges = .80 to .83 and .77) were uniformly respectively). Correlations among the five scales ranged high, with average interitem correlations ranging in Samples 1 and 2, respectively, from |.42| and |.37| to from .33 to .38 and .30 for the subscales, .29 to .35 |.04| and |.03| ([low] Prosociality with Manipulativeness and .22 to .36 for the scales, and .12 for the total DIS- and Risk Taking, respectively), averaging |.20| and |.21| I score. Thus, all the scales and subscales fall in the in the two samples, respectively. Using the Prosociality recommended range (.15 to .50; Briggs & Cheek, subscales rather than the overall scale, the range was 1986; Clark & Watson, 1995) for scales that are rea- from .55 to .00 (Goal Orientation with Considerateness sonably homogeneous without containing overly and Risk Taking, respectively) in development Sample redundant items. That for the total score was slightly 1, and in cross-validation Sample 2 from |.46| to |.01| lower than recommended, indicating that the con- (Considerateness with Manipulativeness and Risk struct measured by the DIS-I is quite broad. Taking, respectively), with overall means of |.23| and |.21|, respectively, in the two samples. Descriptive statistics. Also as shown in Table 3, We then used PCAs to examine the hypothesized small-to-medium effect-size gender differences one-factor solution defined by the DIS-I scales (using (Cohen’s d) replicated across Samples 1 and 2, the two subscales of Prosociality rather than the over- respectively, on three scales, one subscale, and the all scale).5 In Sample 1, five of the scales loaded from DIS-I total score: Women scored higher on |.44| to |.80| on the factor (Considerateness = .80, Goal Orderliness (.52 and .60), Prosociality (.43 and .47), Orientation = .73, Manipulativeness = –.67, and Considerateness (.62 and .59), whereas men Orderliness = .52, and Distractibility = –.44), whereas scored higher on Manipulativeness (.68 and .77) and Risk Taking loaded –.24. Similar results were obtained the DIS-I total score (.52 and .67). in Sample 2 where five of the scales loaded from |.48| Scale intercorrelations. We next examined the corre- to |.69| on the factor (Considerateness = .67, Goal lations among the DIS-I total, scale, and subscale Orientation = .64, Manipulativeness = –.69, scores, in each sample (see Table 4). Correlations Orderliness = .61, and Distractibility = –.48), whereas between the DIS-I total score and scale scores ranged Risk Taking loaded –.28.6 Dindo et al. / Development of a Measure of Disinhibition 9

Table 3 Descriptive Statistics for the Disinhibition Inventory (DIS-I) Scales

Study 1 Study 2

DIS-I Scales (Number of Items) α MIIC Scale Mean SD α MIIC Scale Mean SD a a Prosociality (16) .87 .29 2.86* .51 .82 .22 3.06* .35 b b Considerateness (8) .83 .38 2.95* .58 .77 .30 3.09* .41 Goal Orientation (8) .80 .33 2.77 .58 .77 .30 3.03 .42 b b Manipulativeness (16) .87 .29 1.50† .58 .86 .28 2.08† .42 a Distractibility (16) .88 .31 2.15* .62 .87 .29 2.58 .45 a Risk Taking (8) .81 .35 2.23 .63 .82 .36 2.66† .49 b b Orderliness (9) .83 .35 2.25* .70 .83 .35 2.69* .53 b b Total Score (64) .90 .12 1.70† .36 .89 .12 2.27† .27 Note: Study 1: N = 565. Study 2: N = 515. MIIC = mean interitem correlation. *Women or †men scored significantly (p < .01) higher. a. Small (.20 to .49) effect size (Cohen’s d). b. Medium (.50 to .79) effect size (Cohen’s d).

Table 4 Correlations Among the Six Disinhibition Inventory Scales

aStudy 2 M

1 a b 2 3 4 5 Total Prosociality Subscales

1. Prosociality .— .84 .83 −.37 –.20 .03 .34 −.64 .24 .— a. Goal Orientation .88 .— .41 −.16 –.35 .06 .34 −.57 .— .27 b. Considerateness .88 .55 .— −.46 .03 −.01 .22 −.50 .— .23 2. Manipulativeness −.42 −.29 −.45 .— .22 .25 −.17 .71 .25 .26 3. Distractibility −.16 −.33 .05 .24 .— .11 −.20 .65 .18 .18 4. Risk Taking .04 .00 .07 .23 .21 .— −.14 .39 .13 .11 5. Orderliness .32 .33 .24 −.11 –.13 −.15 .— −.57 .21 .22 Total −.66 −.65 −.51 .72 .65 .41 −.50 .__ .__ .__ aStudy 1 M with .24 .— .— .25 .19 .16 .18 .__ .20b/.21c .— Prosociality aStudy 1 M with .— .31 .29 .27 .19 .13 .19 .__ .— .23b /.21c subscales Notes: Study 1 represented below the diagonal; N = 565. Study 2 represented above the diagonal; N = 515. Means computed using Fisher’s r-to-z transformation. Correlations ≥ .35 are in bold. a. Absolute values. b. Study 1. c. Study 2.

These correlational and factor analytic results sug- analyses reveal a similar pattern, with Risk Taking less gest that certain scales (e.g., Manipulativeness vs. Goal centrally related to the DIS-I. Orientation and Considerateness) are more centrally related to the domain defined by the DIS-I, whereas External Validity Results other scales and constructs are less central (e.g., Risk Taking and to a lesser extent, Orderliness). More specif- Sample 1: Convergent and Discriminant ically, these analyses revealed that Manipulativeness is Validity moderately related to each of the other scales except Orderliness; Prosociality and its subscales are moder- Correlations with the GTS and NEO PI-R domains. ately related to all other scales except Prosociality–Risk We next examined relations of the DIS-I Scale and total Taking and Considerateness–Distractibility. Factor scores with well-validated measures of the Big-Three 10 Assessment

Table 5 Correlations of the Disinhibition Inventory With the General Temperament Survey and Revised NEO Personality Inventory Disinhibition Domain Scales

Prosociality Mean

GTS/NEO PI-R Total Pros Con GO Man Distr Risk Taking Ordrly Pros Subscales

Disinhibition domain scales GTS DIS .63 −.31 −.23 −.32 .50a .36 .48 −.25 .38 .36 GTS CO .56 −.28 −.13 −.36 .32 .41 .38 −.29 .34 .32 GTS AB .47 −.22 −.22 −.18 .50 .20 .34 −.13 .28 .27 NEO PI-R A −.54 .50 .58 .30 −.67ab −.10 −.10 .15 .33 .34 A Trust −.31 .33 .35 .24 −.35 −.17 .13 .06 .21 .21 A Straightforward −.49 .34 .38 .22 −.67 −.09 −.21 .10 .28 .28 A Altruism −.44 .57a .62a .38 −.50 −.04 .05 .17 .27 .29 A Compliance −.43 .25 .31 .13 −.45 −.19 −.24 .12 .25 .24 A Modesty −.25 .16 .25 .04 −.47 .03 −.09 .03 .16 .15 A Tender-Minded −.28 .39 .48 .22 −.31 .06 −.06 .15 .19 .21 NEO PI-R C −.72 .52 .31 .60a −.36 −.55 −.24 .47 .43a .43a C Competence −.51 .45 .29 .50 −.29 −.38 −.09 .22 .29 .29 C Order −.45 .22 .11 .29 −.14 −.28 −.16 .63a .29 .27 C Dutifulness −.57 .52 .40 .51 −.38 −.36 −.14 .24 .33 .34 C Achieve Striving −.52 .45 .21 .57 −.23 −.44 −.06 .30 .30 .30 C Self-Discipline −.64 .44 .23 .55 −.29 −.60a −.14 .36 .37 .36 C Deliberation −.54 .25 .15 .28 −.29 −.37 −.50a .32 .34 .32 Other Big-Three and FFM domain scales GTS NT .12 −.12 −.09 −.13 .12 .23a −.16 .14 .15 .15 NEO PI-R N .26 −.14 −.03 −.21 .15 .45a −.13 .04 .19 .17 GTS PT −.25 .33 .20 .38a −.08 −.23 .16 .15 .19 .20 NEO PI-R E −.17 .46 .45a .36 −.16 .01 .38 .14 .24 .26 NEO PI-R O .17 .34 .38a .22 −.28 −.01 .23 −.05 .19 .20 Note: N = 565. Correlations ≥ .35 are in bold. GTS = General Temperament Survey; NEO PI-R = NEO Personality Inventory–Revised; Pros= Prosociality total score; Con = Considerateness subscale of Prosociality; GOL = Goal Orientation subscale of Prosociality; Man = Manipulativeness; Distr = Distractibility; Ordrly = Orderly; DIS = Disinhibition; CO = Carefree Orientation; AB = Antisocial Behavior; A = Agreeableness; C = Conscientiousness; NT = Negative Temperament; N = Neuroticism; PT = Positive Temperament; E = Extraversion; O = Openness; FFM = Five-Factor Model. a. Highest correlation for each GTS or NEO PI-R higher order scale. b. Highest correlation for each Disinhibition Inventory Scale. and Big-Five personality traits: the GTS and the NEO a strong association with at least one facet of NEO PI- PI-R domain scales, respectively. Based on our earlier lit- R A or C. In sharp contrast, its remaining correlations erature review, we expected GTS Disinhibition, and ranged from only |.12| (GTS NT) to |.26| (NEO PI-R N). NEO PI-R A and C, which are associated empirically Thus, as predicted, the overall DIS-I score is broadly with (low) Disinhibition, to be associated moderately to related positively to Disinhibition scales in a measure of strongly with the DIS-I Scale and total scores. In con- the Big Three and negatively to both C and A of the trast, we predicted that NEO PI-R and GTS N/NT and FFM, and only weakly related to N, E, and O. This pat- E/PT would exhibit lower correlations; O was not tern of correlations helps to establish the construct expected to be associated strongly with any DIS-I scales. validity of the overall DIS-I score and demonstrates that Results, shown in Table 5, were largely consistent the DIS-I is broadly and specifically relevant to the dis- with these hypotheses. As expected, the DIS-I total inhibition domain. score correlated most strongly with GTS Disinhibition It is noteworthy, moreover, that the individual DIS- (.63) and its subscales (r = .56 and .47 with Carefree I scales, including the two Prosociality subscales, Orientation and Antisocial Behavior, respectively), as exhibit varying correlational patterns with these other well as with the NEO PI-R C (–.72) and A (–.54) measures. On one hand, Considerateness and domain scales. Additionally, each DIS-I scale exhibited Manipulativeness both showed significantly (p < .001; Dindo et al. / Development of a Measure of Disinhibition 11 zs = 6.97 and –8.72, respectively) stronger correla- Not surprisingly, given the inclusion of the NEO PI- tions with A (rs = .58 and –.67, respectively) than with R facets, the five-factor structure that emerged, shown C (rs = .31 and –.36, respectively). Manipulativeness in Table 6, reflects the FFM. The first factor includes all also correlated significantly (z = 4.82, p < .001) more the NEO PI-R C facets, the GTS Carefree Orientation strongly with GTS Antisocial Behavior (r = .50) than subscale (–), and GTS PT. In addition, three of the DIS- with Carefree Orientation (r = .32). In contrast, Goal I scales, Goal Orientation, Orderliness, and (low) Orientation, Orderliness, and Distractibility correlated Distractibility, loaded .58 or higher on this factor. The significantly more strongly with C (r range = |.47| to core of this factor thus is an orientation toward high |.60|) and with GTS Carefree Orientation (r range= achievement; controlled, orderly behavior; and hard |.29| to |.41|) than with either A (r range= |.10| to |.30|) work. With a loading of .40, PT clearly is not central to or GTS Antisocial Behavior (r range = |.13| to |.20|; z this construct, but the factor is moderately associated range = 3.9 to 5.3, p < .001). In addition, Risk Taking with positive emotionality. The second factor includes correlated more strongly (p < .001; zs = .9.29 and five of the six NEO PI-R E facets and three facets from 10.9, respectively) with GTS Disinhibition (r = .48) other NEO PI-R domains—one each from A, O, and N. than with either A (r = –.10) or C (r = –.24). In addition, DIS-I Considerateness and Risk Taking Finally, it should be noted that Prosociality and its loaded .48 and .47 on this factor, respectively. Thus, the subscales correlated weakly (r = .20) to moderately (r = core of this factor is sociability, warmth, and other pos- .45; M = .37) with markers of E/PT, and Risk Taking itive emotions, plus preference for a stimulus-rich envi- correlated moderately with NEO PI-R E (r = .38) but not ronment. NEO PI-R Impulsiveness (N) also related GTS PT (r = .16). Risk Taking’s stronger association moderately to this dimension. with NEO PI-R E most likely is because of its strong The third factor includes five of the six NEO PI-R A specific correlation with the Excitement-Seeking facet facets and two other NEO PI-R facets (Assertiveness (r = .48; rs with the other five facets ranged from .18 to [E] and [low] Angry Hostility [N]), as well as GTS .30). In contrast, Considerateness and Distractibility Antisocial Behavior. DIS-I Manipulativeness also related moderately to NEO PI-R N (r = .45) and NEO loaded strongly and negatively (–.73) on this factor. The PI-R O (r = .38), respectively. This pattern of correla- core of this factor thus is a prosocial interpersonal style tions again demonstrates that Disinhibition is a multi- at its high end, whereas the low pole reflects manipula- faceted construct containing a broad range of content tiveness and hostility. The fourth factor clearly is N/NT, related primarily to both A and C, with some more spe- as GTS NT and the six NEO PI-R N facets all load .40 cific content related to other FFM domains. Moreover, it or greater on this dimension; consistent with the corre- suggests that there is utility in retaining the subscales of lational findings, DIS-I Distractibility also has a mod- the broad Prosociality factor because of their differential erate (.36) loading on this factor. The final factor is relations to A and C, as well as other FFM domains. Openness; as expected, no disinhibition markers load significantly on this dimension. Thus, consistent with Factor analysis of the DIS-I and GTS scales with expectation, the DIS-I scales loaded substantially (i.e., the NEO PI-R facets. Finally, we conducted a factor |.35| or greater) on every factor except Openness. analysis of the DIS-I scales, GTS, and NEO PI-R facet scales using PCA with promax rotation to deter- Sample 2: Convergent and mine how the DIS-I scales relate to conceptually sim- Discriminant Validity ilar and distinct constructs. For comprehensiveness and to allow for the emergence of all included facets, Correlations with the SNAP. We also examined five factors were extracted. Given the correlational relations between the DIS-I and SNAP scales (Table 7); results shown in Table 5, the DIS-I scales were the results were generally consistent with those found expected to load primarily onto factors reflecting A in Study 1. Consistent with the Study 1 results, the and C, but also to associate at least moderately with DIS-I total score was strongly related to higher order N/NT and E/PT. Because the two subscales of the SNAP Disinhibition (r = .67), its subscales (r = .57 Prosociality scale showed differential correlational and .56 with Carefree Orientation and Antisocial patterns with the GTS and NEO PI-R scales, we used Behavior, respectively), and primary trait Impulsivity them rather than the overall Prosociality scale in (r = .58). Although relatively unrelated to the broad these analyses. domain of SNAP NT (r = .17), DIS-I total correlated 12 Assessment

Table 6 Principal Components Analysis of the Disinhibition Inventory Scales, GTS Scales, and NEO PI-R Facets

CE A NO

NEO PI-R (C) Achievement Striving .80 .15 −.13 .01 .05 NEO PI-R (C) Self-Discipline .79 .04 .04 −.14 −.09 NEO PI-R (C) Order .71 .11 −.09 .18 −.31 GTS Carefree Orientation −.67 .35 −.21 −.07 −.13 DIS-I Goal Orientation .64* .20 −.03 .01 .15 NEO PI-R (C) Deliberation .63 −.29 .29 .07 −.07 DIS-I Orderliness .62* .21 −.09 .34 −.28 NEO PI-R (C) Dutifulness .62 .17 .18 −.01 .06 NEO PI-R (C) Competence .61 .20 .00 −.10 .17 DIS-I Distractibility −.58 .36 −.03 .36 −.14 GTS Positive Temperament .40 .33 −.19 −.23 .11 NEO PI-R (E) Gregariousness −.06 .88 .02 −.12 −.38 NEO PI-R (E) Excitement Seeking −.05 .84 −.13 −.02 −.04 NEO PI-R (E) Warmth .08 .78 .26 −.01 −.02 NEO PI-R (E) Positive Emotions .09 .72 .09 −.11 .06 NEO PI-R (A) Altruism .14 .59 .47 .04 .04 NEO PI-R (O) Feelings .12 .53 .12 .26 .37 NEO PI-R (E) Activity .32 .52 −.27 −.08 .08 DIS-I Considerateness .13 .48 .38 .15 .11 NEO PI-R (N) Impulsiveness −.34 .47 −.17 .42 .09 DIS-I Risk Taking −.34 .47* −.33 −.27 .12 NEO PI-R (A) Compliance −.04 .00 .77 −.12 −.09 NEO PI-R (A) Straightforwardness .04 .11 .76 .01 −.09 DIS-I Manipulativeness −.10 .00 −.73* .06 −.13 NEO PI-R (A) Modesty −.14 −.04 .65 .15 .04 NEO PI-R (N) Angry Hostility .11 −.13 −.64 .56 .11 GTS Antisocial Behavior −.25 .14 −.56 −.08 .01 NEO PI-R (A) Trust −.05 .45 .45 −.37 −.11 NEO PI-R (A) Tender Mindedness −.02 .31 .45 .23 .20 NEO PI-R (E) Assertiveness .31 .37 −.41 −.28 .12 NEO PI-R (N) Anxiety .09 .15 .07 .79 −.04 GTS Negative Temperament .15 −.05 −.25 .79 −.03 NEO PI-R (N) Depression −.12 −.19 .04 .73 .18 NEO PI-R (N) Self-Consciousness .04 −.15 .15 .73 .09 NEO PI-R (N) Vulnerability −.21 −.13 .07 .68 −.13 NEO PI-R (O) Ideas .10 −.17 −.10 −.11 .80 NEO PI-R (O) Aesthetics .12 −.06 .02 .13 .73 NEO PI-R (O) Fantasy −.22 .20 −.04 .13 .58 NEO PI-R (O) Actions −.26 −.13 .03 −.42 .49 NEO PI-R (O) Values −.19 .26 .19 .03 .36 Note: N = 565. Correlations ≥ .35 are in bold. *Indicates DIS-I scales that loaded on a single factor. Loadings of DIS-I scales that split across factors are italicized. DIS-I = Disinhibition Inventory; GTS = General Temperament Survey; NEO PI-R = Revised NEO Personality Inventory; C = Conscientiousness; E = Extraversion; A = Agreeableness; N = Neuroticism. O = Openness; with primary traits Manipulativeness, Aggression, respectively), particularly its Antisocial Behavior and Mistrust. Notably, these three scales often split subscale (rs = .60 and .35, respectively) and between the NT and Disinhibition domains (Clark Impulsivity (rs = .35 and .50, respectively). DIS-I et al., in press). Finally, DIS-I total was unrelated to Distractibility, Prosociality, and Goal Orientation SNAP PT (r = –.21) and its associated primary traits. were moderately related to SNAP Disinhibition Among the primary DIS-I scales, Manipulativeness (r range = |.32| to |.39|) and its Carefree Orientation and Risk Taking exhibited the strongest associations subscale (r range = |.34| to |.41|), as well as modestly with SNAP Disinhibition (rs = .51 and .44, related to SNAP Impulsivity (r range = .30 to .33). Dindo et al. / Development of a Measure of Disinhibition 13

Table 7 Correlations of the Disinhibition Inventory With the Schedule for Nonadaptive and Adaptive Personality-2

Pros Mean

Total Pros Con GO Man Distr Risk Taking Ordrly Pros Subscales

NT .17 −.06 −.03 −.06 .14 .34a −.13 .08 .15 .13 Mistrust .35a −.13 −.08 −.13 .30 .33 .05 −.10 .18 .17 Man .59 −.29 −.30 −.20 .64ab .26 .26 −.20 .34 .32 Aggression .49 −.27 −.37b −.09 .51a .23 .23 −.14 .28 .27 Self-harm .28a −.23 −.14 −.25 .17 .28a −.01 −.03 .15 .15 EccPerc .27a −.12 −.11 −.09 .27 .17 .13 −.07 .15 .14 Dependency .12 −.18 −.07 −.23a .04 .22 −.22 −.01 .13 .13 PT −.21 .29 .15 .34a −.07 −.22 .24 .16 .20 .20 Exhibitionism .07 .10 .02 .14 .16 −.03 .32a .07 .14 .13 Entitlement .05 .07 −.04 .15 .27a −.13 .13 .04 .13 .13 Detachment .12 −.16 −.16a −.11 .15 .07 −.16a −.05 .12 .12 Disinhibition .67ab −.33 −.23 −.32 .51 .39 .44 −.28 .39b .37b CO .57a −.34b −.16 −.41 .31 .41b .29 −.31b .33 .32 AB .56 −.20 −.22 −.11 .61a .25 .35 −.18 .33 .30 Impulsivity .58a −.33 −.25 −.30 .35 .31 .50b −.28 .35 .33 Propriety −.28a .22 .13 .25 −.11 −.11 −.20 .23 .18 .17 Workaholism −.23 .30 .07 .43ab .03 −.19 .08 .28 .18 .18 Note: N = 309. Correlations ≥ .35 are in bold. Pros = Prosociality total score; Con = Considerateness subscale of Prosociality; GO = Goal Orientation subscale of Prosociality; Man = Manipulativeness; Distr = Distractibility; Ordrly = Orderly; NT = Negative Temperament; EccPerc = Eccentric Perception; PT = Positive Temperament; CO = Carefree Orientation; and AB = Antisocial Behavior. a. Highest correlation for each SNAP Scale. b. Highest correlation for each Disinhibition Inventory Scale.

DIS-I Orderliness and Considerateness exhibited composed of five scales: Prosociality, with correlated generally low associations with SNAP Disinhibition subscales Goal Orientation and Considerateness, (rs = –.28 and –.23, respectively). Manipulativeness, Risk Taking, Distractibility, and Consistent with Study 1 results, only DIS-I Goal Orderliness. The DIS-I total score is strongly associated Orientation related moderately to the domain of PT (r = with each of these scales. As predicted based on previ- .34); with few exceptions, the DIS-I scales otherwise ous research, disinhibition is a somewhat heteroge- were unrelated to lower order traits associated with PT. neous construct and, accordingly, the facets of this trait As before, only DIS-I Distractibility correlated moder- vary in their degree of intercorrelations. ately with SNAP NT (r = .34). However, DIS-I Initial examination of the DIS-I in relation to Big- Manipulativeness exhibited strong associations with pri- Three and Big-Five personality measures demonstrated mary SNAP scales Manipulativeness and Aggression (rs that the DIS-I total score correlated most highly with = .64 and .51, respectively), and DIS-I Considerateness conceptually related constructs (GTS Disinhibition and exhibited moderate associations with these same scales. NEO PI-R A and C). In sharp contrast, correlations with theoretically unrelated traits (N/NT and E/PT) were generally weak. This pattern of correlations Discussion supports the construct validity of the overall DIS-I score and demonstrates that the DIS-I is broadly rel- Summary of Findings evant to the disinhibition domain. The DIS-I scales To explicate the nature of disinhibition and provide a also were largely unrelated to conceptually distinct more thorough means of its assessment, we developed constructs: Only DIS-I Distractibility correlated a multifaceted measure of the construct. The DIS-I is (moderately) with N/NT, whereas the “positive” 14 Assessment

DIS-I scales (Prosociality and its subscales) plus Risk Considerateness, and Orderliness—and high end— Taking correlated moderately with E/PT. Manipulativeness, Distractibility, and Risk Taking— Factor analyses of the DIS-I, GTS, and NEO PI-R of this higher order trait. Moreover, they suggest that resulted in a five-factor solution, in which each DIS- Risk Taking exhibits strong associations to E. As we I scale loaded on one of three factors, representing discuss subsequently, this result is consistent with (a) discipline, order, and dutifulness versus a carefree recent work into the multifaceted nature of impulsiv- approach; (b) interpersonal warmth and thrill seek- ity (e.g., Smith, Fischer, Cyders, Annus, & Spillane, ing; and (c) agreeableness, including compliance and 2007; Whiteside & Lynam, 2001, 2003). straightforwardness versus manipulativeness and A third, related, impetus for this study was the nature hostile behavior. The final two factors, reflecting of the trait itself: Disinhibition poses a particular mea- N/NT scales and Openness, did not include any dis- surement challenge given the complex relations of its inhibition markers. Thus, the DIS-I scales show a facets to various domains of personality. For example, clear convergent/discriminant with other measures of whereas Tellegen (in press) places aggression within the disinhibition and (un)related constructs. domain of negative emotionality, Zuckerman (1979) places it within a broad factor he labeled Impulsive- Significance of Disinhibition and the Unsocialized-Sensation-Seeking, and, in the SNAP, the Contributions of This Study construct splits across these two factors (Clark et al., in press). More generally, as we have seen, although the The impetus for this examination of disinhibition domain as a whole is broadly related to externalizing was threefold. First, there is a recent emergence of psychopathology, its components, as represented in the interest in the construct of disinhibition, in large part DIS-I scales, relate significantly to three of the Big-Five because of its significant relations with externalizing factors (see Table 5). This diversity of content can lead psychopathology. As noted previously, researchers to inconsistent results in studies aimed at clarifying the have found strong associations between trait disinhi- causes, correlates, and consequences of disinhibition bition, drug and alcohol abuse, and antisocial behav- (e.g., associations with certain disorders, relations to ior that can be explained by a shared latent sexual behavior, and links to academic performance) vulnerability reflecting a strong genetic influence unless the nature of the particular measure(s) used is (e.g., Krueger et al., 2002). Numerous other studies well understood. Thus, given disinhibition’s complex- also have found a common genetic diathesis to under- ity, research explicating its essential nature and bound- lie associations between externalizing disorders and aries, as well as its relations to other traits, is essential. personality traits related to disinhibition, including For example, recent developments in the psychopa- novelty seeking (Young, Stallings, Corley, Krauter, & thy literature indicate that the affective-interpersonal Hewitt, 2000), behavioral undercontrol (Slutske dimension of is specifically and nega- et al., 2002), and social deviance and excitement tively associated with the harm-avoidance dimension of seeking (Mustanski, Viken, Kaprio, & Rose, 2003). disinhibition (i.e., engaging in dangerous behaviors— Furthermore, prospective studies suggest that trait measured by Risk Taking in the DIS-I), whereas the disinhibition, in fact, predates and predicts the emer- impulsive-antisocial dimension of psychopathy is gence of externalizing psychopathology (e.g., Caspi associated with low control (i.e., nonplanfulness; DIS- et al., 1996; Masse & Tremblay, 1997). In summary, I’s [low] Goal Orientation) and traditionalism disinhibition is clearly important in the study of psy- (Benning, Patrick, Hicks, Blonigen, & Krueger, 2003). chopathology, particularly in the development of Such findings have contributed to a greater under- externalizing disorders. standing of the risk factors associated with subtypes of Second, research was needed to clarify the structure psychopathy, and this kind of research now needs to be of disinhibition, as there is little consensus regarding extended to other externalizing disorders. the basic constituent elements of this broad higher order domain (e.g., Tellegen, in press; Zuckerman Unique Contributions of DIS-I et al., 1988). Thus, we conducted a comprehensive survey of the relevant literature with the goal of The DIS-I also includes scales that tap into disin- improved understanding and measurement of the hibition-related constructs that have been neglected lower order structure of this complex trait. Our results in current adult measures of this domain. First, dis- suggest there are five facets of disinhibition, com- tractibility/poor attentional control is a commonly prising both the low end—Goal Orientation, assessed trait in measures of childhood temperament Dindo et al. / Development of a Measure of Disinhibition 15 that has received less attention in the adult personal- the Disinhibitory factor further breaks down into ity literature. For example, the Children’s Behavior dimensions of Disagreeable Disinhibition and Questionnaire (Rothbart, Ahadi, & Hershey, 1994) Unconscientious Disinhibition, with the former includes an Attentional Focusing scale as a lower reflecting callousness, manipulativeness, and aggres- order facet of Effortful Control which, in turn, has sion and the latter reflecting impulsivity and under- been linked to Disinhibition in adults (Rothbart & controlled behavior. These two factors clearly parallel Ahadi, 1994). Similarly, the Child Behavior the FFM traits of A and C, respectively. Thus, the hier- Checklist (Achenbach, 1991) includes a trait measure archical structure of personality and the superordinate of attention problems that has been shown to have the trait of alpha can help explain some of the complex highest discriminating power for diagnosing attention associations between the DIS-I scales and markers of deficit hyperactivity disorder (Chen, Faraone, Big-Three and Big-Five traits (e.g., the moderate cor- Biederman, & Tsuang, 1994), a disorder that shares a relations with markers of N/NT). Interestingly, genetic vulnerability with other externalizing disor- whereas SNAP Manipulativeness shares significant ders, such as substance abuse and variance with the N/NT domain, and thus directly rep- (Eaves et al., 1997). The DIS-I Distractibility scale resents alpha of the Big Two (see Digman, 1997; thus represents a distinct contribution to the adult Markon et al., 2005), DIS-I Manipulativeness, in con- personality assessment literature that may help to trast, correlates strongly with SNAP Manipulativeness, connect adult and child models of disinhibition and, but is free of N/NT variance, and thus represents a in turn, adult and childhood forms of externalizing more pure measure of this aspect of disinhibition. psychopathology. To test this hierarchical model of personality, factor Second, DIS-I Manipulativeness is related concep- analyses were conducted using the measures developed tually to various externalizing disorders. For example, in Study 1. PAF analyses with promax rotation were deceitfulness/using others for personal profit or conducted on the NEO PI-R facets, GTS PT and NT pleasure is a diagnostic criteria for both ASPD and scales, GTS Disinhibition subscales (Antisocial conduct disorder, and also is central in descriptions of Behavior and Carefree Orientation), and the six DIS-I psychopathic individuals. Yet this particular aspect of (sub)scales, examining the two-, three-, and four-factor the domain has been relatively neglected in previous solutions. The two-factor solution revealed factors measures of disinhibition (see Clark, 1993; Clark resembling Digman’s (1997) alpha—containing pri- et al., in press, for an exception, discussed further marily facets of N/NT, Disinhibition, C, and A—and below). Although FFM A measures include content beta, a combined E/PT and Openness factor. The three- related to low manipulativeness, low A per se does factor solution contained a similar first factor resem- not capture its essence. Other measures, such as the bling beta, whereas the second factor generally UPPS (Whiteside & Lynam, 2001), are more nar- represented Disinhibition, containing C, A, and rowly focused on “impulsive-like behaviors,” and Disinhibition facets/scales. The third factor contained thus exclude items associated with (low) agreeableness mainly N/NT scales. Thus, the Big-Three model of per- or manipulativeness. sonality generally emerged in the three-factor solution, consistent with the hierarchical model. Finally, in the Disinhibition and Hierarchical four-factor solution, the first factor again resembled beta (DIS-I Considerateness and Risk Taking, NEO PI- Structures of Personality R E, and O facets), the second factor contained mainly Markon et al. (2005) examined the joint factor C and Disinhibition (GTS Carefree Orientation, DIS-I structure of common measures of both normal per- Goal Orientation, Orderliness, and Distractibility) sonality and personality pathology. Consistent with facets, the third factor was composed of N/NT scales previous research, their findings demonstrate that per- and the fourth factor contained mainly A and sonality structure is hierarchical. That is, at the super- Disinhibition (GTS Antisocial Behavior and DIS-I ordinate level, two factors, first identified by Digman Manipulativeness) facets. Thus, at the four-factor level, (1997) and labeled alpha (α) and beta (β), emerge. the broad Disinhibition factor found at the three-factor The former consists of a combination of N/NT and level split into two components. Similar results for each Disinhibition, whereas the latter consists of E/PT and solution were found using varimax rotation as well. Openness. Alpha breaks into its two components at Thus, the two-, three-, and four-factor solutions from the three-factor level and then, at the four-factor level, Study 1 generally resembled previously established 16 Assessment

higher order structures of personality. This hierarchical with Sensation Seeking. Just as Risk Taking was less perspective helps to explain the complex correlations central to Disinhibition in this study, Sensation between facets of disinhibition and other personality Seeking has been shown to be distinct from the other constructs such as N/NT. “Impulsivity” facets in the above-mentioned research.

What About Impulsivity? Further Validation of the DIS-I We initially hypothesized that impulsivity would Research is needed to validate the DIS-I further emerge as a specific facet of the broad disinhibition and to address this investigation’s limitations. One construct. However, in our factor analyses, impulsiv- important direction for future research is validating ity content split across multiple factors rather than the DIS-I’s psychometric properties and structure in forming its own specific dimension. Multiple steps diverse samples, such as clinical populations, given were taken to encourage the emergence of an the clinical implications of this construct. Although Impulsivity factor, including (a) factor analyzing the O’Connor (2002) established that factor structures of impulsivity items in isolation, (b) examining the personality traits and clinical symptoms typically are internal consistency reliability and average interitem highly robust across clinical and nonclinical samples, correlation of a rationally derived Impulsivity scale, an extension to a clinical sample nevertheless would and (c) examining correlations among the impulsivity represent a significant test of the measures’ general- items and other facets of disinhibition. Nevertheless, izability. Given the robust support for the DIS-I’s impulsivity continued to fail to emerge as a specific structure in Study 2, we would expect replication; facet. Thus, our results indicate that impulsivity is not nevertheless, formal tests are in order. In addition, a distinct and coherent construct but comprises con- retest reliability studies of the DIS-I are needed to tent within multiple domains of personality. examine its temporal stability. These results are congruent with an emerging con- Furthermore, future research should identify specific sensus regarding impulsivity. In a study examining behavioral correlates (e.g., in such domains as alcohol the “place of impulsiveness in a dimensional model and drug use, academic and job performance, and sex- of personality,” Eysenck and Eysenck (1977) found ual behavior) of the DIS-I total and scale scores. Future that the four subscales associated with this construct studies should also examine the incremental validity of are differentially associated with the Big-Three the DIS-I in predicting maladaptive and externalizing domains of N, E, and Psychoticism. That is, Risk behaviors beyond other instruments measuring related Taking, Nonplanning, and Liveliness exhibited stronger constructs (e.g., NEO PI-R A and C). Finally, further associations with E, whereas narrow Impulsiveness examinations of the DIS-I’s convergent and discrimi- correlated more strongly with Psychoticism. nant validity across multiple methods (e.g., self-other Whiteside and Lynam (2001) factor analyzed mul- agreement) are needed. For example, Oltmanns and tiple scales related to impulsivity and also concluded Turkheimer (2006) found that self- and peer reports that it was a multifaceted construct. Subsequently, were better at predicting, respectively, internalizing and Whiteside, Lynam, Miller, and Reynolds (2005) repli- externalizing problems. Thus, this last line of research cated their facets in diverse samples, whereas Smith would permit investigation of the convergence between et al. (2007) extended this research using a multitrait raters on the trait of disinhibition, as well as provide multimethod design and demonstrated that distinct interesting data regarding potential sources of disagree- “impulsivity-like traits” were associated with different ment and the utility of peer ratings. aspects of risky behaviors. They stated that “reliance on unspecified or overly broad measures of ‘impulsiv- ity’ invites theoretical and empirical imprecision” Notes (p. 168), and recommended specific assessment of tar- geted constructs. The current study further supports 1. We use these two terms interchangeably. this multidomain conceptualization of impulsivity. 2. According to Fabrigar, Wegener, MacCallum, and Strahan (1999) and MacCallum, Widaman, Zhang, and Hong (1999), the Specifically, (low) Goal Orientation resembles Lack adequacy of a sample size depends on the number of variables of Premeditation and Persistence, Distractibility is representing each common factor, and the extent to which the related to Urgency, and Risk Taking can be identified individual variables in each domain is related to each other. If Dindo et al. / Development of a Measure of Disinhibition 17 many variables represent the common factors and the variables Clark, L. A., Simms, L. J., Wu, K. D., & Casillas, A. (Ed.). (in are moderately to highly related (i.e., have high communalities), press). Schedule for nonadaptive and adaptive personality: then accurate results may be obtained from samples as small as Manual for administration, scoring, and interpretation. (2nd 100. Similarly, Guadagnoli and Velicer (1988) found that the sta- Edition). Minneapolis, MN: University of Minnesota Press. bility of factor analyses did not vary as a function of the partici- Clark, L. A., & Watson, D. (1990). General temperament survey. pants to variables ratio (i.e., the popular rule of thumb of that Unpublished manuscript. sample size should increase with the number of variables). Clark, L.A. & Watson, D. (1993). Behavioral disinhibition versus Rather, their series of analyses indicate that factor loadings and constraint: A dispositional perspective. In D. Wegner & absolute sample size are the most important determinants of J. Pennebaker (Eds.), Handbook of mental control (pp. 506- factor stability. When Guadagnoli and Velicer examined a similar 527). NJ: Prentice-Hall. number of variables (144), factors (6), and factor loadings (.40) Clark, L. A., & Watson, D. (1995). Constructing validity: Basic as those in our study, excellent comparability was found between issues in objective scale development. Psychological at sample sizes of 150. Assessment, 7, 309-319. 3. Factor analyses also were carried out with principal axis Clark, L. A., & Watson, D. (1999). Temperament: A new paradigm factoring (PAF) and with varimax rotation. Results were consis- for personality. In L. Pervin & O. John (Eds.), Handbook of tent with those reported. personality (2nd ed., pp. 399-423). New York: Guilford. 4. Parallel analysis is a method of selecting the number of fac- Costa, P. T., & McCrae, R. R. (1992). NEO Personality tors to retain based on the rationale that valid factors from real Inventory–Revised. Port Huron, MI: Sigma Assessment Systems. data should have larger eigenvalues than parallel analyses con- Digman, J. M. (1997). Higher-order factors of the Big Five. ducted with random data sets that have the same sample size and Journal of Personality and Social Psychology, 73, 1246-1256. number of variables (Hayton, Allen, & Scarpello, 2004). Eaves, L. J., Silberg, J. L., Meyer, J. M., Maes, H. H., Simonoff, 5. Principal components analyses examining the one-factor E., Pickles, A., et al. (1997). Genetics and developmental psy- solution defined by the DIS-I scales were also carried out with the chopathology: 2. The main effects of genes and environment five scales (i.e., using the overall Prosociality scale rather than the on behavioral problems in the Virginia twin study of adoles- subscales). In Study 1, Manipulativeness had the highest loading cent behavioral development. Journal of Child Psychology (.73) followed by Prosociality (–.70), Distractibility (.56), Orderly and Psychiatry, 38, 965-980. (–.55), and Risk Taking (.41). In Study 2, Manipulativeness also Eysenck, H. J., & Eysenck, S. B. G. (1975). Manual of the Eysenck exhibited the strongest loading (.57) followed by Prosociality Personality Questionnaire. London: Hodder and Stoughton. (–.53), Orderly (–.45), Distractibility (.40), and Risk Taking (.26). Eysenck, H. J., & Eysenck, S. B. G. (1977). The place of impulsive- 6. A similar pattern of factor loadings was obtained when PAF ness in a dimensional system of personality description. British was used to examine the one-factor solution of the six DIS-I Journal of Social and Clinical Psychology, 16(1), 57-68. scales. In Study 1, five of the scales loaded |.37| to |.71| on the Fabrigar, L. R., Wegener, D. T., MacCallum, R. C., & Strahan, factor, whereas Risk Taking loaded –.14. In Study 2, five of the E. J. (1999). Evaluating the use of exploratory factor analysis scales loaded |.35| to |.59|, whereas Risk Taking loaded –.19. in psychological research. Psychological Methods, 4, 272-299. Frick, P. J., Kuper, K., Silverthorn, P., & Cotter, M. (1995). Antisocial behavior, somatization, and sensation-seeking behavior in References mothers of clinic-referred children. Journal of the American Academy of Child and Adolescent Psychiatry, 34, 805-812. Achenbach, T. M. (1991). Manual for the Child Behavior Gough, H. G. (1987). Manual for the California Psychological Checklist/4-18 and 1991. Burlington: University of Vermont, Inventory. Palo Alto, CA: Consulting Psychologists Press. Department of Psychiatry. Guadagnoli, E., & Velicer, W. F. (1988). Relation of sample size Benning, S. D., Patrick, C. J., Hicks, B. M., Blonigen, D. M., & to the stability of component patterns. Psychological Bulletin, Krueger, R. F. (2003). Factor structure of the psychopathic 103, 265-275. personality inventory: Validity and implications for clinical Hayton, J. C., Allen, D. G., & Scarpello, V. (2004). Factor reten- assessment. Psychological Assessment, 15, 340-350. tion decisions in exploratory factor analysis: A tutorial on par- Briggs, S. R., & Cheek, J. M. (1986). The role of factor analysis allel analysis. Organizational Research Methods, 7, 191-205. in the development and evaluation of personality scales. Kosson, D. S., Steuerwald, B. L., Newman, J. P., & Widom, C. S. Journal of Personality, 54, 106-148. (1994). The relation between socialization and antisocial Buss, A., & Plomin, R. (1984). Temperament: Early personality behavior, substance use, and family conflict in college traits. Hillsdale, NJ: Lawrence Erlbaum. students. Journal of Personality Assessment, 63, 473-488. Caspi, A., Moffitt, T. E., Newman, D. L., & Silva, P. A. (1996). Krueger, R. F., Hicks, B. M., Patrick, C. J., Carlson, S. R., Iacono, Behavioral observations at age 3 years predict adult psychi- W. G., & McGue, M. (2002). Etiologic connections among atric disorders. Longitudinal evidence from a birth cohort. substance dependence, antisocial behavior, and personality: Archives of General Psychiatry, 53, 1033-1039. Modeling the externalizing spectrum. Journal of Abnormal Chen, W. J., Faraone, S. V., Biederman, J., & Tsuang, M. T. Psychology, 111, 411-424. (1994). Diagnostic accuracy of the Child Behavior Checklist Krueger, R. F., Schmutte, P. S., Caspi, A., Moffitt, T. E., scales for attention-deficit hyperactivity disorder: A receiver- Campbell, K., & Silva, P. A. (1994). Personality traits are operating characteristic analysis. Journal of Consulting and linked to crime among men and women: Evidence from a Clinical Psychology, 62, 1017-1025. birth cohort. Journal of Abnormal Psychology, 103, Clark, L. A. (1993). Manual for the Schedule for Nonadaptive 328-338. and Adaptive Personality (SNAP). Minneapolis: University of Loevinger, J. (1957). Objective tests as instruments of Minnesota Press. psychological theory. Psychological Reports, 3, 635-694. 18 Assessment

MacCallum, R. C., Widaman, K. F., Zhang, S., & Hong, S. Slutske, W. S., Heath, A. C., Madden, P. A., Bucholz, K. K., (1999). Sample size in factor analysis. Psychological Statham, D. J., & Martin, N. G. (2002). Personality and the Methods, 4, 84-99. genetic risk for alcohol dependence. Journal of Abnormal Markon, K. E., Krueger, R. F., & Watson, D. (2005). Delineating Psychology, 111, 124-133. the structure of normal and abnormal personality: An integra- Smith, G. T., Fischer, S., Cyders, M. A., Annus, A. M., & Spillane, tive hierarchical approach. Journal of Personality and Social N. S. (2007). On the validity and utility of discriminating Psychology, 88, 139-157. among impulsivity-like traits. Assessment, 14, 155-170. Masse, L. C., & Tremblay, R. E. (1997). Behavior of boys in Tellegen, A. (1985). Structures of mood and personality and their kindergarten and the onset of substance use during adoles- relevance to assessing anxiety, with an emphasis on self- cence. Archives of General Psychiatry, 54, 62-68. report. In A. H. Tuma & J. D. Maser (Eds.), Anxiety and the McCormick, R. A. (1993). Disinhibition and negative affectivity anxiety disorders. Hillsdale, NJ: Lawrence Erlbaum. in substance abusers with and without a gambling problem. Tellegen, A. (in press). Multidimensional Personality Addictive Behaviors, 18, 331-336. Questionnaire. Minneapolis: University of Minnesota Press. McGue, M., Slutske, W., Taylor, J., & Iacono, W. G. (1997). Watson, D., & Clark, L. A. (1993). Behavioral disinhibition ver- Personality and substance use disorders: I. Effects of gender sus constraint: A dispositional perspective. In D. M. Wegner & and alcoholism subtype. Alcoholism: Clinical and J. W. Pennebaker (Eds.), Handbook of mental control Experimental Research, 21, 513-520. (pp. 506-527). Upper Saddle River, NJ: Prentice Hall. Miller, J. D., & Lynam, D. R. (2003). Psychopathy and the five- Watson, D., Clark, L. A., & Harkness, A. R. (1994). Structures of factor model of personality: A replication and extension. personality and their relevance to psychopathology. Journal of Journal of Personality Assessment, 81, 168-178. Abnormal Psychology, 103, 18-31. Mustanski, B. S., Viken, R. J., Kaprio, J., & Rose, R. J. (2003). Wensman, L., & Clark, L. A. (2008). Impulsivity: Multiple traits, mul- Genetic influences on the association between personality risk tiple behaviors. Manuscript in preparation, University of Iowa. factors and alcohol use and abuse. Journal of Abnormal Whiteside, S. P., & Lynam, D. R. (2001). The five-factor model and Psychology, 112, 282-289. impulsivity: Using a structural model of personality to understand O’Connor, B. P. (2002). The search for dimensional structure impulsivity. Personality and Individual Differences, 30, 669-689. differences between normality and abnormality: A statisti- Whitesdie, S., & Lynam, D. (2003). Understanding the role of cal review of published data on personality and psy- impulsivity and externalizing psychopathology in alcohol chopathology. Journal of Personality and Social Psychology, abuse: Application of the UPPS Impulsive Behavior Scale. 83, 962-982. Experimental and Clinical Psychopharmacology, 11, 210-217. Oltmanns, T. F., & Turkheimer, E. (2006). Perceptions of self and Whiteside, S. P., Lynam, D. R., Miller, J. D., & Reynolds, S. K. (2005). others regarding pathological personality traits. In R. F. Validation of the UPPS Behaviour Scale: A four-factor model of Krueger & J. L. Tackett (Eds.), Personality and psychopathol- impulsivity. European Journal of Personality, 19, 559-574. ogy (pp. 71-111). New York: Guilford. Young, S. E., Stallings, M. C., Corley, R. P., Krauter, K. S., & Rothbart, M. K., & Ahadi, S. A. (1994). Temperament and the Hewitt, J. K. (2000). Genetic and environmental influences on development of personality. Journal of Abnormal Psychology, behavioral disinhibition. American Journal of Medical 103, 55-66. Genetics, 96, 684-695. Rothbart, M. K., Ahadi, S. A., & Hershey, K. L. (1994). Zuckerman, M. (1979). Sensation seeking: Beyond the optimal Temperament and social behavior in children. Merrill-Palmer level of arousal. Hillsdale, NJ: Lawrence Erlbaum. Quarterly, 40, 21-39. Zuckerman, M., Kuhlman, D. M., & Camac, C. (1988). What lies Sher, K. J., & Trull, T. J. (1994). Personality and disinhibitory beyond E and N? Factor analyses of scales believed to mea- psychopathology: Alcoholism and antisocial personality dis- sure basic dimensions of personality. Journal of Personality order. Journal of Abnormal Psychology, 103, 92-102. and Social Psychology, 54, 96-107.