Development and validation of a short form ...

Personality and Individual Differences 87 (2015) 130–135

Contents lists available at ScienceDirect

Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid

Development and validation of a short form aggressive beliefs and attitudes scale Jesse S. Michel ⁎, Paige Hartman, Seth Gitter Department of Psychology, Auburn University, United States

a r t i c l e

i n f o

Article history: Received 31 May 2015 Received in revised form 28 July 2015 Accepted 30 July 2015 Available online xxxx Keywords: Aggression Deviance Measurement

a b s t r a c t This research outlines six studies (total N = 3867) that develop and validate an Aggressive Beliefs and Attitudes — Short Form scale for use within the normal nonclinical adult population (e.g., organizational psychology). In the first two samples, exploratory factor analysis reduced the original 30 item scale to a shorter, more parsimonious, eight item measure. In the third sample, confirmatory factor analysis found that the proposed model and items fit the data extremely well. Examination of the nomological network underlying the new measure in Samples 3–4 displayed relationships with positive affect, negative affect, agreeableness, conscientiousness, and neuroticism ranging from r = .23 to −.23, while relationships with anger, hostility, physical aggression, and verbal aggression ranged from r = .49 to .30. Finally, in two criterion-related validity studies the relationships between the aggressive beliefs and attitudes scales, both original and short forms, and workplace deviance were examined with independent samples of employed adults. The short form scale was significantly related to workplace deviance in both samples (r = .37 and .43). Furthermore, these relationships were of identical magnitude to the full 30 item measure, suggesting that the short form scale comparably captures aggression-related behaviors with a greatly reduced number of items. © 2015 Elsevier Ltd. All rights reserved.

Social-cognition is a dominant theoretical perspective for understanding how personality and individual differences result in coherent patterns of human behavior (Greenwald & Banaji, 1995; Mischel & Shoda, 1995). Social-cognition emphasizes how individuals relate with and interpret their social world. Recent research has shown that implicit and explicit social cognitions are important components in explaining how personality and attitudes affect behavior. Implicit and explicit social cognitions are separate constructs at the theoretical and empirical levels, and are operationally distinct components of basic personality structure. Indeed, recent research has shown that implicit and explicit social cognitions each have unique and potentially interactive explanatory prediction in various criteria (e.g., dishonesty, organizational deviance, traffic violations) in basic and applied research (Bing, LeBreton, Davison, Migetz, & James, 2007; Bing, Stewart, et al., 2007; Frost, Ko, & James, 2007; Gawronski & Bodenhausen, 2006). Implicit social cognitions are generally classified as effortless, automatic, and unconscious reasoning leading to a person's beliefs, attitudes, and subsequent behavioral tendencies (Greenwald & Banaji, 1995). Considering that implicit social cognitions operate unconsciously, researchers largely endorse the use of indirect personality assessment (Karpinski & Hilton, 2001; Nosek, Greenwald, & Banaji, 2005; Olson &

⁎ Corresponding author at: Department of Psychology, Auburn University, 226 Thach Hall, Auburn, AL 36849-5214, United States. E-mail address: [email protected] (J.S. Michel).

http://dx.doi.org/10.1016/j.paid.2015.07.041 0191-8869/© 2015 Elsevier Ltd. All rights reserved.

Fazio, 2003), such as the Thematic Apperception Test (Lilienfeld, Wood, & Garb, 2000) or Implicit Association Test (Greenwald, McGhee, & Schwartz, 1998). Within the industrial and organizational psychology literature, James and colleagues (e.g., James, 1998; James & LeBreton, 2012; James & McIntyre, 2000; James et al., 2005) advocate the use of conditional reasoning methods (e.g., Conditional Reasoning Test of Aggression or CRT-A) to assess underlying cognitive biases (i.e., motive-based biases in reasoning and inference). For example, the fundamental idea underlying the CRT-A is that aggressive people, versus non-aggressive or prosocial people, believe that their aggressive actions or reactions are reasonable and appropriate (Baron & Richardson, 1994; Baumeister, Smart, & Boden, 1996). Accordingly, people high in dispositional aggression rely on implicit cognitive biases to rationalize their behavior, thus reflecting their personality and underlying implicit social cognitions. Explicit social cognitions refer to corresponding (i.e., in relation to implicit social cognitions) introspective, organized, and conscious reasoning leading to a person's beliefs, attitudes, and subsequent behavioral tendencies (Greenwald & Banaji, 1995). Considering that explicit social cognitions take place within conscious awareness, direct methods such as self-reported questionnaires are typically used in assessment (McClelland, Koestner, & Weinberger, 1989). Though a number of general aggression measures currently exist with items that capture aggressive emotions and behavioral tendencies, such as “I have trouble controlling my temper” or “I have become so mad that I have broken things” (e.g., Angry Hostility Scale from the NEO personality inventory

J.S. Michel et al. / Personality and Individual Differences 87 (2015) 130–135

[Costa & McCrae, 1992]; Anger, Hostility, Physical Aggression, and Verbal Aggression from the Aggression Questionnaire [Buss & Perry, 1992]), as well as several instruments that measure aggressive cognitions within the clinical literature (e.g., Grisso, Davis, Vesselinov, Appelbaum, & Monahan, 2000; Nagtegaal, 2008), only one measure comprehensively assesses the six explicit social cognitions associated with aggressive biases outlined by James and colleagues within the organizational psychology literature (James, 1998; James & Mazerolle, 2002; James et al., 2005). Specifically, Michel, Pace, Edun, Sawhney, and Thomas (2014) developed a 30 item measure that taps into the aggressive biases of hostile attribution (tendency to see harmful intent in the actions of others), potency (tendency to frame and reason using the contrast of strength versus weakness), retribution (tendency to confer logical priority to retaliation over reconciliation), victimization by powerful others (tendency to frame oneself as a victim and as being exploited by the powerful), derogation of target (attempt to make the target more deserving of aggression), and social discounting (tendency to call on socially unorthodox and antisocial beliefs to interpret and analyze social events and relationships). In doing so, this measure assesses the “explicit aggressive beliefs and attitudes that influence individual patterns of appraisals, attributions, and behavior across situations” (Michel et al., p. 328; cf. Gawronski & Bodenhausen, 2006; Greenwald & Banaji, 1995; Karpinski & Hilton, 2001). The objective of the current research is to develop and validate a short form aggressive beliefs and attitudes scale for use within the normal nonclinical adult population for basic and applied research (e.g., organizational psychology). This is an important contribution to the literature as, in many scenarios, the use of the full 30 item measure may not be possible (e.g., survey length requirements). Additionally, much like the CRT-A, researchers and practitioners are generally interested in an overall assessment of aggression as opposed to facet level examination (e.g., retribution, derogation of target, social discounting). Accordingly, the goals of this research are as follows. First, the proposed series of studies will develop a short form measure based on item commonalities across multiple samples that best represents an overall construct of explicit aggressive beliefs and attitudes. Second, strong psychometric properties will be shown in multiple samples based on confirmatory factor analysis (CFA), coefficient alpha estimates, and mean inter-item correlations. Third, the nomological network of the short form items will be examined with trait affect, the Five Factor Model (FFM) of personality, and multiple forms of aggression (anger, hostility, physical aggression, and verbal aggression). This is an important contribution as the original validation work displayed strong convergent and discriminant validity evidence for all 30 items with other measures of implicit and explicit aggression and the FFM; however, examination of relationships with positive and negative affect remains unexplored. Additionally, we will reexamine patterns of covariance with the most highly related FFM traits in the original validation work (i.e., agreeableness, conscientiousness, and neuroticism), as well as other forms of aggression (i.e., anger, hostility, physical aggression, and verbal aggression), to further support the distinctiveness of the short form items. Fourth, criterion-related validity evidence will be shown for the Aggressive Beliefs and Attitudes — Short Form scale with real world aggressive criteria in multiple working adult samples. 1. Method 1.1. Participants and procedure 1.1.1. Samples 1–2 Two independent samples were recruited through Amazon's Mechanical Turk, which is a large crowd sourcing internet marketplace (database currently consists of over 500,000 individuals from 190 countries) shown to produce demographically diverse samples and reliable data (see Buhrmester, Kwang, & Gosling, 2011; Mason & Suri, 2012; Paolacci, Chandler, & Ipeirotis, 2010). We recruited participants living

131

in the U.S. and 19 years of age or older. To encourage participation, individuals received a small monetary incentive ($0.25). For Sample 1, we received completed cleaned data from 945 respondents.1 The average participant was 33.3 years of age (SD = 11.4) and male (53.5%). Ethnic/racial breakdown of the sample was: 76.7% Caucasian or White (non-Hispanic), 5.7% African American or Black, 6.2% Hispanic, 9.6% Asian American or Pacific Islander, .3% Native American, and 1.4% other. The sample was well educated with the majority of participants having an advanced degree (10.1%), a Bachelor's degree (38.2%), or some college education (35.9%). For Sample 2, we received completed cleaned data from 712 independent respondents. The average participant was 31.3 years of age (SD = 10.0) and male (53.2%). Ethnic/racial breakdown of the sample was: 76.5% Caucasian or White (non-Hispanic), 6.7% African American or Black, 7.0% Hispanic, 7.7% Asian American or Pacific Islander, .8% Native American, and 1.1% other. The majority of participants had an advanced degree (8.4%), a Bachelor's degree (36.9%), or some college education (39.2%). 1.1.2. Samples 3–4 Two independent samples of undergraduate students were recruited from a large university in the southeastern U.S. participated in the study for extra credit. Sample 3 participants were solicited through the university SONA system and completed the survey online. We received completed clean data from 670 participants. Demographic information indicated that the sample was diverse (12.7% Caucasian or White [non-Hispanic], 6.3% African American or Black, 65.4% Hispanic, 3.9% Asian American or Pacific Islander, and 11.7% other), predominately female (71.0%), and ranged in age from 18 to 56 (M = 21.1, SD = 4.2). Sample 4 participants were recruited via classroom lectures and completed the survey in-person (i.e., paper-and-pencil). We received completed clean data from 341 participants. The average participant was 20.9 years of age (SD = 4.6) and female (67.6%). Ethnic/racial breakdown of the sample was: 15.5% Caucasian or White (nonHispanic), 8.4% African American or Black, 68.1% Hispanic, 3.4% Asian American or Pacific Islander, and 4.6% other. 1.1.3. Samples 5–6 Two independent samples were recruited using a peer-nomination web-based sampling methodology similar to approaches used by Matthews, Kath, and Barnes-Farrell (2010) and Martins, Eddleston, and Veiga (2002). Information about the study was presented to students at a large southeastern U.S. university enrolled in advanced undergraduate courses in organizational psychology. Individuals were instructed to forward the study information to others who might qualify (an email invitation was provided). In order to be eligible, participants were required to be 18 years of age or older, work at least 20 h per week, and not identify themselves as a college student. Participants meeting these requirements followed a link to an online survey that verified their eligibility, collected contact information (to verify the accuracy of their data), and obtained consent to participate. Students received nominal course credit while participants received no compensation. Sample 5 consisted of 339 participants. The average participant was 29 years of age (SD = 10.5), worked 36 h per week (SD = 9.48), had a job tenure of 3.38 years (SD = 4.73), was not married or living as 1 We followed recommendations in the survey methods literature and included quality control items in each survey across the six samples (Huang, Liu, & Bowling, 2015; Meade & Craig, 2012). Participants engaging in careless or insufficient effort responding were identified and removed prior to analyses. Specifically, when examining the factor structure of the revised scale (Samples 1–3) we used an inclusion rule of zero missed items (i.e., if a participant missed one or more quality control items we excluded their data due to careless or insufficient effort responding). For nomological network and criterion-related validity evidence (Samples 4–6) we used a slightly relaxed inclusion rule of up to one missed item (i.e., if a participant missed two or more quality control items we excluded their data due to careless or insufficient effort responding).

132

J.S. Michel et al. / Personality and Individual Differences 87 (2015) 130–135

married (68.1%), and female (62.2%). Additionally, many participants had an advanced degree (9.1%), a Bachelor's degree (24.8%), or some college education (42.7%). Ethnic/racial breakdown of the sample was: 13.9% Caucasian or White (non-Hispanic), 11.5% African American or Black, 69.0% Hispanic, 1.2% Asian American or Pacific Islander, .6% Native American, and 3.8% other. Sample 6 consisted of an independent sample of 860 participants with data collection occurring approximately six months after Sample 5. The average participant was 29 years of age (SD = 11.2) and worked 36 h per week (SD = 10.75). Participants had an average job tenure of 3.42 years (SD = 5.07), were not married or living as married (68.1%), and were predominately female (59.2%). Many participants had an advanced degree (9.3%), a Bachelor's degree (25.3%), or some college education (46.3%). Ethnic/racial breakdown of the sample was: 15.9% Caucasian or White (non-Hispanic), 8.3% African American or Black, 68.1% Hispanic, 2.0% Asian American or Pacific Islander, .1% Native American, and 5.6% other.

characteristic of me). Coefficient alphas for Sample 4 were .79 (anger), .77 (hostility), .83 (physical aggression), and .57 (verbal aggression).

1.2. Measures

To reduce the original 30 item scale and develop a shorter more parsimonious measure, exploratory factor analysis (EFA) was conducted on Samples 1 and 2. EFAs were performed through principle axis factoring with oblique rotation and Kaiser normalization. Eigenvalues from both samples suggest the presence of a six factor structure within the full 30 item scale (Sample 1 had an eigenvalue of 1.47 for 6 factor solution versus .83 for 7 factor solution; Sample 2 had an eigenvalue of 1.39 for 6 factor solution versus .82 for 7 factor solution). The six factor solution also explained the majority of the total item variance in both samples (55.06% in Sample 1; 53.90% in Sample 2). Finally, with the exception of one derogation of target bias item (“In general, people are either good or evil.”), which was the poorest performing derogation item in the original validation work (see Michel et al., 2014), each item exhibited strong factor loadings (above .45) and small cross-loadings (below the .32 factor loading cutoff criteria recommendation of Tabachnick & Fidell, 2001) on the original factors across both samples. The original scale development and validation work of Michel et al. (2014) proposed and found support for a superordinate aggressive beliefs and attitudes construct. To identify the best overall items that represent this higher-order superordinate construct, we examined extraction communalities of our EFA results. Communalities are the squared multiple correlation of the item as predicted from the set of factors, and these values reflect the proportion of item variance accounted for by the factors in the solution (Tabachnick & Fidell, 2001). Researchers have suggested that large samples with communalities less than .40 are not highly correlated with one or more of the factors in the solution (Worthington & Whittaker, 2006) and in the social sciences communality magnitudes generally range from .40 to .70 (Costello & Osborne, 2005). In our EFA we found that seven items fell below .43 in Samples 1 and 2. Examining the communality magnitudes in Sample 1 there was a noticeable drop in communality magnitudes after the 11th highest communality value (.63). In Sample 2, these 11 items retained the highest communality magnitudes, though we found another noticeable decrease after the 8th strongest item (.67). Considering the high communality magnitudes across the two samples, as well as the replication of the items across the two samples, we retained eight items for the final Aggressive Beliefs and Attitudes — Short Form scale. These items and communalities are presented in Table 1. The final short form scale with participant and administrator instructions is provided in Appendix A.

1.2.1. Aggressive beliefs and attitudes Explicit aggressive beliefs and attitudes were assessed in Samples 1–6 with the original 30 item scale developed and validated by Michel et al. (2014). Sample items include: “People gain others' trust to betray them” (hostile attribution bias); “Life presents challenges that separate the weak from the strong” (potency bias); “If I am betrayed then I have the right to retaliate” (retribution bias); “The rich get richer by taking advantage of the poor” (victimization by powerful others bias); “Some people are just bad people” (derogation of target bias); and “Laws are meant to be broken” (social discounting bias). Items were rated on a seven-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). Coefficient alphas for Samples 1–6 were .91, .92, .91, .89, .95, and .95. 1.2.2. Positive and negative affect Positive and negative affect were assessed in Sample 3 with the 20 item Positive and Negative Affect Schedule (PANAS) from Watson, Clark, and Tellegen (1988). Sample positive affect adjective descriptors include: “active,” “alert,” “determined,” “excited,” and “strong,” while sample negative affect adjective descriptors include: “irritable,” “hostile,” “guilty,” “upset,” and “distressed.” Items were rated on a five-point frequency scale ranging from 1 (very slightly or not at all) to 5 (extremely). Coefficient alpha for Sample 3 was .92 of positive affect and .86 for negative affect. 1.2.3. Agreeableness, conscientiousness, and neuroticism The constructs of agreeableness, conscientiousness, and neuroticism were assessed in Sample 3 with the public domain International Personality Item Pool developed by Goldberg and colleagues (Goldberg, 1999; Goldberg et al., 2006). Each personality construct consisted of 10 items, with sample items including “I insult people” (agreeableness — reverse coded), “I make a mess of things” (conscientiousness — reverse coded), and “I get upset easy” (neuroticism). Items were rated on a 5-point scale ranging from 1 (very inaccurate) to 5 (very accurate). Coefficient alpha reliabilities for Sample 3 were .78 (agreeableness), .83 (conscientiousness), and .88 (neuroticism). 1.2.4. Anger, hostility, physical aggression, and verbal aggression The constructs of anger, hostility, physical aggression, and verbal aggression were assessed in Sample 4 with Buss and Perry's (1992) 29 item Aggression Questionnaire. Sample items include “Some of my friends think I am a hothead” (anger), “I sometimes feel that people are laughing at me behind my back” (hostility), “Once in a while, I can't control the urge to strike another person” (physical aggression), and “I can't help getting into arguments when people disagree with me” (verbal aggression). Each item was rated on a five-point scale ranging from 1 (extremely uncharacteristic of me) to 5 (extremely

1.2.5. Workplace deviance Workplace deviance was assessed in Samples 5–6 with the 19 item measure developed by Bennett and Robinson (2000). This measure assesses interpersonal and organizational deviance within the workplace. Sample items include “Said something hurtful to someone at work,” “Publicly embarrassed someone at work,” “Taken property from work without permission,” and “Dragged out work in order to get overtime.” Each item was rated on a seven-point frequency scale from 1 (never) to 7 (daily). Coefficient alpha for the workplace deviance scale was .95 in Sample 5 and .90 in Sample 6. 2. Results 2.1. Exploratory factor analysis (Samples 1–2)

2.2. Confirmatory factor analysis and nomological network (Samples 3–4) CFA was performed in Sample 3 to examine the fit of the Aggressive Beliefs and Attitudes — Short Form scale. Following the proposed model of Michel et al. (2014), we examined a hypothesized hierarchical model in which the superordinate aggressive beliefs and attitudes construct affects the factor-level constructs of victimization by powerful others

J.S. Michel et al. / Personality and Individual Differences 87 (2015) 130–135 Table 1 Short form items and communalities from Sample 1 and Sample 2.

133

Table 3 Descriptive statistics, correlations, and reliabilities for Sample 4.

Item

Sample 1

Sample 2

Variables

M

1. The wealthy capitalize on those who are less fortunate. 2. Some people are just bad people. 3. The rich get richer by taking advantage of the poor. 4. Getting back at others makes me feel better. 5. I believe that large corporations exploit their employees. 6. If I am betrayed then I have the right to retaliate. 7. If someone disrespects me, I feel the need to get even. 8. Some people are simply horrible human beings.

.798 .715 .744 .720 .642 .659 .635 .686

.753 .746 .732 .701 .680 .669 .668 .665

1. Aggressive Beliefs and Attitudes — Short Form 2. Anger 3. Hostility 4. Physical aggression 5. Verbal aggression

4.06 1.03 .77 2.31 2.64 2.31 3.07

SD

1

.83 .79 .86 .89

.39⁎⁎⁎ .40⁎⁎⁎ .49⁎⁎⁎ .30⁎⁎⁎

2

3

4

5

.79 .52⁎⁎⁎ .77 .54⁎⁎⁎ .36⁎⁎⁎ .83 .54⁎⁎⁎ .40⁎⁎⁎ .47⁎⁎⁎ .57

Note. N = 341–333. M = mean; SD = standard deviation. Alpha reliabilities are in italics and appear on the diagonal. ⁎⁎⁎ p b .001.

Note. Sample 1 N = 945. Sample 2 N = 712.

(three items), derogation of target (two items), and retribution (three items). Results indicate that the hypothesized model and items fit the data extremely well: χ2 (17, N = 670) = 17.21, p = .44; RMSEA = .004 (90% CI of RMSEA = .000–.035); SRMR = .017; CFI = 1.00; and TLI = 1.00. To examine the nomological network of the short form scale we examined discriminant and convergent validity with several measures of individual differences and aggression. Specifically, we examined discriminant validity as the extent to which the new short form scale exhibits small effect size magnitudes (see Cohen's, 1988, classification of small [≥.10], medium [≥.30], and large [≥.50]) with measures of dissimilar constructs and convergent validity as the extent to which the new short form scale exhibits medium to large effect size magnitudes with measures of similar constructs (Campbell & Fiske, 1959; Hinkin, 1998). In Sample 3, the Aggressive Beliefs and Attitudes — Short Form scale displayed small but significant patterns of covariance with positive affect, negative affect, agreeableness, conscientiousness, and neuroticism with estimates ranging from r = .23 to r = − .23 (see Table 2). These magnitudes show evidence of discriminant validity as the new measure is related to, but clearly distinct from, theoretically linked constructs such as negative affect (r = .23) and agreeableness (r = −.23); meanwhile, these relationships are much smaller than the relationship found for negative affect and neuroticism (r = .67), which are frequently viewed as overlapping constructs (e.g., Judge, Higgins, Thoresen, & Barrick, 1999). Indeed, it is commonly argued that observed values approaching or exceeding r = .70 indicates a single construct (e.g., Berry, Ones, & Sackett, 2007). In Sample 4, the Aggressive Beliefs and Attitudes — Short Form scale displayed medium effect size magnitudes with anger (r = .39), hostility (r = .40), physical aggression (r = .49), and verbal aggression (r = .30; see Table 3), while the overall relationship between the Aggressive Beliefs and Attitudes — Short Form scale and the Aggression Questionnaire was large (r = .53, p b .001). Accordingly, these magnitudes display evidence of convergent validity with theoretically related forms of aggression, but do not indicate excessive overlap or evidence of a single construct.

in two independent working samples. As outlined in Table 4, the Aggressive Beliefs and Attitudes — Short Form scale was significantly related to workplace deviance in Sample 5 (r = .37, p b .001) and Sample 6 (r = .43, p b .001). This is not surprising considering similar relationships were displayed in the original development and validation work (see Michel et al., 2014). However, it is important to note that these relationships were of identical magnitude to the full 30 item measure, suggesting that the eight items in the revised short form scale adequately capture deviant behavior in the workplace in these samples with a greatly reduced number of items. Additionally, based on Classical Test Theory, our results suggest the eight item measure is more internally consistent based on average inter-item correlations in these two samples (mean inter-item correlation of .415 for the revised eight item short form versus .388 for the original 30 item original form). Finally, the original and short form aggressive beliefs and attitudes measures were related at r = .94 (p b .001) in Sample 5 and r = .93 (p b .001) in Sample 6, clearly indicating a single construct. 3. Discussion The purpose of this research was to develop and validate a short form aggressive beliefs and attitudes scale for basic and applied research settings with normal nonclinical adult samples. In Samples 1–2, we found evidence for a shorter more parsimonious measure based on commonalities from independent EFAs. CFA from Sample 3 found that the resultant Aggressive Beliefs and Attitudes — Short Form scale fits the data extremely well. Additionally, in Samples 3–4 the short form scale displayed discriminant validity evidence with a number of highly researched personality variables (positive affect, negative affect, agreeableness, conscientiousness, and neuroticism), and convergent validity evidence with the Aggression Questionnaire (anger, hostility, physical aggression, and verbal aggression). Finally, in a series of independent working samples (Samples 5–6), we found evidence for strong criterion-related validity. Specifically, the Aggressive Beliefs and Attitudes — Short Form scale was significantly related to workplace deviance in both samples, and these relationships were of similar magnitude across the strongly correlated 30 item and reduced eight item measures, suggesting that the revised short form measure captures workplace deviance and potentially other aggression-related behaviors with a greatly reduced number of items. Additionally, our results

2.3. Criterion-related validity (Samples 5–6) Relationships between the aggressive beliefs and attitudes scales (both original and short forms) and workplace deviance were examined

Table 2 Descriptive statistics, correlations, and reliabilities for Sample 3. Variables

M

SD

1

2

3

4

5

6

1. Aggressive Beliefs and Attitudes — Short Form 2. Positive affect 3. Negative affect 4. Agreeableness 5. Conscientiousness 6. Neuroticism

4.00 3.82 1.88 4.03 3.58 2.81

1.12 .76 .66 .55 .70 .81

.80 −.21⁎⁎⁎ .23⁎⁎⁎ −.23⁎⁎⁎ −.23⁎⁎⁎ .23⁎⁎⁎

.92 −.29⁎⁎⁎ .28⁎⁎⁎ .35⁎⁎⁎ −.30⁎⁎⁎

.86 −.09⁎ −.26⁎⁎⁎ .67⁎⁎⁎

.78 .27⁎⁎⁎ −.06

.83 −.20⁎⁎⁎

.88

Note. N = 670–668. M = mean; SD = standard deviation. Alpha reliabilities are in italics and appear on the diagonal. ⁎ p b .05. ⁎⁎⁎ p b .001.

134

J.S. Michel et al. / Personality and Individual Differences 87 (2015) 130–135

suggest the eight item measure is more internally consistent based on average inter-item correlations in our criterion-related validity samples.

criterion-related validity for the reduced measure. Second, researchers and practitioners can pair the CRT-A with the Aggressive Beliefs and Attitudes — Short Form to extend the limited research on implicit and explicit aggression and interactions, thus further exploring the unique and interactive processes and mechanisms underlying aggressive behaviors (e.g., mediation, moderation, independent and additive coaction; cf. Bing, LeBreton, et al., 2007; Frost et al., 2007; Gawronski & Bodenhausen, 2006). Third, future research should investigate the Aggressive Beliefs and Attitudes — Short Form scale in relation to an expanded set of contexts and criterion variables within organizational settings and beyond (e.g., conflict resolution, employment decisions, and team composition).

3.1. Limitations

Appendix A. Aggressive beliefs and attitudes — Short form

As with any study, there are several limitations that should be acknowledged in the current work. First, the Aggressive Beliefs and Attitudes — Short Form scale does not capture all six factors outlined in the original measure. Specifically, the new measure includes three victimization by powerful others items, three retribution items, and two derogation of target items. Accordingly, the new measure does not capture as broad of a construct space as the original 30 item measure. This should not be a serious issue, however, as the original development and validation work (see Michel et al., 2014) proposed and found support for a superordinate aggressive beliefs and attitudes construct that affects the six factor-level constructs. Considering our item reduction strategy was based on item communalities across two large independent samples, we were able to retain the most representative items of this superordinate aggressive beliefs and attitudes construct. Additionally, support for the adequacy of the reduced measure was strengthened by strong CFA evidence and criterion-related validity estimates that are equal to the full 30 item scale across multiple independent working samples. Second, this study provided limited convergent and discriminant validity evidence. However, this should not be a major limitation as the original development and validation work (see Michel et al., 2014) displayed extensive convergent and discriminant validity with multiple measures of personality, implicit and explicit aggression, and social desirability (i.e., all of the items in the short form measure included in this original work). Additionally, the current research replicates and extends previous findings through the examination of trait affect (positive affect and negative affect), personality (agreeableness, conscientiousness, and neuroticism), and aggression (anger, hostility, physical aggression, and verbal aggression) in relation to the short form measure.

A.1. Administrator instructions

Table 4 Comparison of original and short forms across Sample 5 and Sample 6. Sample

Original form (30 items) and workplace deviance rxy

Sample 5 (N = 339) Sample 6 (N = 860)

.37⁎⁎⁎ .43⁎⁎⁎

rxx .95 .95

Short form (8 items) and workplace deviance

ryy

rxy

rxx

ryy

.95 .90

.37⁎⁎⁎ .43⁎⁎⁎

.86 .86

.95 .90

⁎⁎⁎ p b .001.

3.2. Implications and future research There are many potential implications for a self-report explicit aggressive beliefs and attitudes short form scale. First, as previously highlighted, this scale should be a valuable tool for basic and applied research settings with survey space limitations. In many scenarios, the use of the full 30 item Aggressive Beliefs and Attitudes measure may not be possible. In the event that factor level relationships are not hypothesized (e.g., derogation of target, social discounting), and an overall aggressive beliefs and attitudes score is preferred, the new short form measure can be used to reduce survey length. Additionally, the new measure extends other short form aggression scales (e.g., Bryant & Smith, 2001; Webster et al., 2014) as previous short form measures have been based on alternative operationalizations of aggression (e.g., anger, hostility, physical aggression, and verbal aggression from the Aggression Questionnaire) versus explicit aggressive beliefs and attitudes (i.e., hostile attribution, potency, retribution, victimization by powerful others, derogation of target, and social discounting). This new short form measure seems particularly fruitful for future research considering the preliminary evidence of strong psychometric properties, discriminant and convergent validity, and

Items should be presented in random order when possible. The overall Short Form score is an average of all eight items. A.2. Participant instructions Please indicate the extent to which you agree or disagree with each of the following statements. Using the 7 point rating scale shown below, write the number corresponding to your level of agreement or disagreement on the line in front of each statement. Strongly disagree

Disagree Slightly disagree

Neither agree nor disagree

Slightly agree

Agree Strongly agree

1

2

4

5

6

1. 2. 3. 4. 5. 6. 7. 8.

3

7

_____The wealthy capitalize on those who are less fortunate. _____Some people are just bad people. _____The rich get richer by taking advantage of the poor. _____Getting back at others makes me feel better. _____I believe that large corporations exploit their employees. _____If I am betrayed then I have the right to retaliate. _____If someone disrespects me, I feel the need to get even. _____Some people are simply horrible human beings.

References Baron, R., & Richardson, D. (1994). Human aggression. New York: Plenum. Baumeister, R., Smart, L., & Boden, J. (1996). Relation of threatened egotism to violence and aggression: The dark side of high self-esteem. Psychological Review, 103, 5–33. Bennett, R., & Robinson, S. (2000). Development of a measure of workplace deviance. Journal of Applied Psychology, 85, 349–360. Berry, C., Ones, D., & Sackett, P. (2007). Interpersonal deviance, organizational deviance, and their common correlates: A review and meta-analysis. Journal of Applied Psychology, 92, 410–424. Bing, M., LeBreton, J., Davison, H., Migetz, D., & James, L. (2007). Integrating implicit and explicit social cognitions for enhanced personality assessment: A general framework for choosing measurement and statistical methods. Organizational Research Methods, 10, 346–389. Bing, M., Stewart, S., Davison, H., Green, P., McIntyre, M., & James, L. (2007). An integrative typology of personality assessment for aggression: Implications for predicting counterproductive workplace behavior. Journal of Applied Psychology, 92, 722–744. Bryant, F., & Smith, B. (2001). Refining the architecture of aggression: A measurement model for the Buss–Perry Aggression Questionnaire. Journal of Research in Personality, 35, 138–167. Buhrmester, M., Kwang, T., & Gosling, S. (2011). Amazon's Mechanical Turk a new source of inexpensive, yet high-quality, data? Perspectives on Psychological Science, 6, 3–5. Buss, A., & Perry, M. (1992). The Aggression Questionnaire. Journal of Personality and Social Psychology, 63, 452–459. Campbell, D., & Fiske, D. (1959). Convergent and discriminant validation by the multitrait–multimethod matrix. Psychological Bulletin, 56, 81–105. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Mahwah, NJ: Erlbaum. Costa, P., & McCrae, R. (1992). Revised NEO Personality Inventory and NEO Five-Factor Inventory professional manual. Odessa, FL: Psychological Assessment Resources. Costello, A., & Osborne, J. (2005). Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical Assessment Research & Evaluation, 10(7), 1–9.

J.S. Michel et al. / Personality and Individual Differences 87 (2015) 130–135 Frost, B., Ko, C., & James, L. (2007). Implicit and explicit personality: A test of a channeling hypothesis for aggressive behavior. Journal of Applied Psychology, 92, 1299–1319. Gawronski, B., & Bodenhausen, G. (2006). Associative and propositional processes in evaluation: An integrative review of implicit and explicit attitude change. Psychological Bulletin, 132, 692–731. Goldberg, L. (1999). A broad-bandwidth, public domain, personality inventory measuring the lower-level facets of several five-factor models. In I. Mervielde, I. Deary, F. De Fruyt, & F. Ostendorf (Eds.), Personality Psychology in Europe, Vol. 7. (pp. 7–28). Tilburg, Netherlands: Tilburg University Press. Goldberg, L., Johnson, J., Eber, H., Hogan, R., Ashton, M., Cloninger, C., et al. (2006). The International Personality Item Pool and the future of public-domain personality measures. Journal of Research in Personality, 40, 84–96. Greenwald, A., & Banaji, M. (1995). Implicit social cognition: Attitudes, self-esteem, and stereotypes. Psychological Review, 102, 4–27. Greenwald, A., McGhee, D., & Schwartz, J. (1998). Measuring individual differences in implicit cognition: The implicit association test. Journal of Personality and Social Psychology, 74, 1464–1480. Grisso, T., Davis, J., Vesselinov, R., Appelbaum, P., & Monahan, J. (2000). Violent thoughts and violent behavior following hospitalization for mental disorder. Journal of Consulting and Clinical Psychology, 68, 388–398. Hinkin, T. (1998). A brief tutorial on the development of measures for use in survey questionnaires. Organizational Research Methods, 1, 104–121. Huang, J., Liu, M., & Bowling, N. (2015). Insufficient effort responding: Examining an insidious confound in survey data. Journal of Applied Psychology, 100, 828–845. James, L. (1998). Measurement of personality via conditional reasoning. Organizational Research Methods, 1, 131–163. James, L., & LeBreton, J. (2012). Assessing the implicit personality through conditional reasoning. Washington, D.C.: American Psychological Association. James, L., & Mazerolle, M. (2002). Personality at work. Thousand Oaks, CA: Sage. James, L., & McIntyre, M. (2000). Conditional reasoning test of aggression: Test manual. Knoxville, TN: Innovative Assessment Technology. James, L., McIntyre, M., Glisson, C., Green, P., Patton, T., LeBreton, J., et al. (2005). A conditional reasoning measure for aggression. Organizational Research Methods, 8, 69–99. Judge, T., Higgins, C., Thoresen, C., & Barrick, M. (1999). The big five personality traits, general mental ability, and career success across the life span. Personnel Psychology, 52, 621–652. Karpinski, A., & Hilton, J. (2001). Attitudes and implicit association test. Journal of Personality and Social Psychology, 81, 774–788. Lilienfeld, S., Wood, J., & Garb, H. (2000). The scientific status of projective techniques. Psychological Science in the Public Interest, 1, 27–66.

135

Martins, L., Eddleston, K., & Veiga, J. (2002). Moderators of the relationship between work–family conflict and career satisfaction. Academy of Management Journal, 45, 399–409. Mason, W., & Suri, S. (2012). Conducting behavioral research on Amazon's Mechanical Turk. Behavior Research Methods, 44, 1–23. Matthews, R., Kath, L., & Barnes-Farrell, J. (2010). Short, valid, predictive measures of work–family interference. Journal of Occupational Health Psychology, 15, 75–90. McClelland, D., Koestner, R., & Weinberger, J. (1989). How do self-attributed and implicit motives differ? Psychological Review, 96, 690–702. Meade, A., & Craig, S. (2012). Identifying careless responses in survey data. Psychological Methods, 17, 437–455. Michel, J., Pace, V., Edun, A., Sawhney, E., & Thomas, J. (2014). Development and validation of an explicit aggressive beliefs and attitudes scale. Journal of Personality Assessment, 96, 327–338. Mischel, W., & Shoda, Y. (1995). A cognitive-affective system theory of personality: Reconceptualizing situations, dispositions, dynamics, and invariance in personality structure. Psychological Review, 102, 246–268. Nagtegaal, M. (2008). Aggression: Its association with dysfunctional thought control processes, cognition, and personality (Unpublished dissertation, Erasmus University Rotterdam). Nosek, B., Greenwald, A., & Banaji, M. (2005). Understanding and using the implicit association test: II. Method and construct validity. Personality and Social Psychology Bulletin, 31, 166–180. Olson, M., & Fazio, R. (2003). Relations between implicit measures of prejudice: What are we measuring? Psychological Science, 14, 636–639. Paolacci, G., Chandler, J., & Ipeirotis, P. (2010). Running experiments on Amazon Mechanical Turk. Judgment and Decision Making, 5, 411–419. Tabachnick, B., & Fidell, L. (2001). Using multivariate statistics. Boston: Allyn and Bacon. Watson, D., Clark, L., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54, 1063–1070. Webster, G., DeWall, C., Pond, R., Deckman, T., Jonason, P., Le, B., et al. (2014). The brief aggression questionnaire: Psychometric and behavioral evidence for an efficient measure of trait aggression. Aggressive Behavior, 40, 120–139. Worthington, R., & Whittaker, T. (2006). Scale development research: A content analysis and recommendations for best practices. The Counseling Psychologist, 34, 806–838.