Implicit Association Test for Aggressiveness: Further evidence of validity and resistance to desirable responding

Personality and Individual Differences 129 (2018) 95–103

Contents lists available at ScienceDirect

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

Implicit Association Test for Aggressiveness: Further evidence of validity and resistance to desirable responding

T

⁎

Maja Parmač Kovačić , Zvonimir Galić, Mitja Ružojčić Department of Psychology, Faculty of Humanities and Social Sciences, University of Zagreb, Croatia

A R T I C LE I N FO

A B S T R A C T

Keywords: Implicit personality Explicit personality Implicit Association Test Aggressiveness Socially desirable responding

This paper reports the results of three interrelated studies investigating the validity and resistance to desirable responding of the Implicit Association Test for Aggressiveness (IAT-A). In Studies 1 and 2, we tested its validity by correlating it with an established explicit measure of aggressiveness, the conceptually closest measure of socially desirable responding (SDR), and various aggression-related criteria in two large samples of participants. The results supported the validity of IAT-A. It had satisfactory reliability, it was non-significantly or weakly related to an explicit measure of aggressiveness and unrelated to the SDR measure, and it explained different aggression-related behaviors over and above the explicit aggressiveness measure. In Study 3, we examined the IAT-A's susceptibility to deliberate response distortion by comparing the IAT-A and self-reported aggressiveness between situations of honest responding and simulated personnel selection. The results revealed that the IAT-A is less susceptible to deliberate response distortion than the self-report measure of explicit aggressiveness. The mean result on the IAT-A was almost identical between the two response situations, whereas for the self-report measure of aggressiveness, participants scored significantly lower in the simulated selection situation. Altogether, the results suggest that IAT-A is a valid and potentially useful implicit aggressiveness measure.

1. Introduction Although aggressive behavior may be triggered by numerous factors, a primary cause of such behavior is the personality trait of aggressiveness (Bergman, McIntyre, & James, 2004). Aggressive individuals tend to perceive ambiguous behavior of others as malicious and hostile (Dodge, 1980). Their attitudes, values and norms favor aggressive behavior (Guerra, Huesmann, & Hanish, 1995) and, instead of alternative options, they are more prone to activate, select and implement aggressive behavior scripts (Banse, Messer, & Fischer, 2015; Huesmann, 1988). According to the dual-process model of social information processing (e.g., Strack & Deutsch, 2004), personality traits similar to aggressiveness have explicit and implicit components. The explicit component is a part of the personality of which the person is aware, consists primarily of self-ascribed characteristics that are available for introspection and predicts immediate decisions and specific behaviors (McClelland, Koestner, & Weinberger, 1989). Implicit personality refers to the dynamic mental structures and processes that influence individuals behavioral adjustments to their environments that are not accessible through introspection (James & LeBreton, 2012), such as implicit motives and defense mechanisms. They result from repeated

⁎

and/or important experiences and are better at predicting long-term outcomes and spontaneous behavior (McClelland et al., 1989). According to the dual process models of personality (McClelland et al., 1989), explicit and implicit personality are only weakly intercorrelated. Whereas explicit aggressiveness is easily captured with self-report questionnaires (e.g., the Aggression Questionnaire, Buss & Perry, 1992), implicit aggressiveness measurement requires more sophisticated instruments. Over the past decades, researchers have developed different indirect measurement tools to assess implicit personality. One of the best known and most promising is the Implicit Association Test (IAT) for self-concept measurement (Schnabel, Asendorpf, & Greenwald, 2008). The fundamental idea behind the IAT procedure for personality assessment is that implicit self-concept consists of clusters of associations between the concept of the self and various psychological attributes. Individuals have formed those associations based on their everyday experiences and the strength of these associations can be measured with a double-discrimination response latency task. In a typical self-concept IAT, such as the one for measuring aggressiveness, participants need to sort stimuli from two contrasted target categories (e.g., self vs. others) and two contrasted attribute categories (e.g., aggressive vs. peaceful), using two response keys. The key assumption underlying the IAT is that if the target and the attribute concepts are

Corresponding author at: Department of Psychology, Faculty of Humanities and Social Sciences, University of Zagreb, I. Lučića, 3, 10 000 Zagreb, Croatia. E-mail address: mparmac@ffzg.hr (M. Parmač Kovačić).

https://doi.org/10.1016/j.paid.2018.03.002 Received 20 August 2017; Received in revised form 27 February 2018; Accepted 2 March 2018 0191-8869/ © 2018 Elsevier Ltd. All rights reserved.

Personality and Individual Differences 129 (2018) 95–103

M. Parmač Kovačić et al.

Additionally, in our study, we wanted to test the interaction between the IAT-A scores and self-reported aggressiveness in predicting aggressive behavior. Recent research that used the conditional reasoning paradigm (James & LeBreton, 2012) for implicit aggressiveness measurement showed that implicit and explicit aggressiveness interact in explaining aggressive behavior, with participants who are high on both measures showing the highest levels of aggressive behavior (Bing et al., 2007). We expected to replicate the findings using the IAT paradigm for implicit aggressiveness measurement and demonstrate that the interaction between the IAT-A scores and self-reported aggressiveness will provide incremental validity in predicting aggressive behavior (H3). Finally, in our research, we wanted to test the relationship between the IAT-A scores and desirable responding. An inconclusive relationship between IAT-A scores and SDR scores from the study by Banse et al. (2015) could be attributed to the use of a set of different SDR measures that were treated as measures of the same construct. Recent developments in SDR conceptualization and measurement (Paulhus, 2002) reveal that the construct has a complex structure whose components can be classified according to the content of presentation into egoistic and moralistic bias. Since moralistic bias refers to denying socially deviant impulses, such as a tendency to behave aggressively, only the moralistic component of socially desirable responding (M-SDR) should theoretically be related to aggressiveness. Therefore, in our study, we explored the relationship between IAT-A scores and a measure of M-SDR. Additionally, in order to test the relationship between the IAT-A and desirable responding, we tested its susceptibility to deliberate SDR (i.e., faking). So far, several studies have dealt with the problem of resistance to faking on self-concept IAT (e.g., Egloff & Schmukle, 2002; Stieger, Göritz, Hergovich, & Voracek, 2011). The results showed that the IAT is much less fakable than self-report measures; it is only slightly fakable under explicit self-presentation instructions (e.g., “try not to appear anxious”, Stieger et al., 2011). However, to the best of our knowledge, none of the studies tested susceptibility of IAT-A to faking. In accordance with the described studies, we expected that the IAT-A scores should neither be related to SDR scores (H4a) nor susceptible to deliberate response distortion when participants are instructed to fake their responses in order to make a good impression (H4b). To test our hypotheses, we conducted three interrelated studies. In Studies 1 and 2, we tested the relationships among IAT-A scores and self-reported aggressiveness, aggressive behavior and SDR scores using large student samples. In Study 1, a sample of psychology students completed the IAT-A, self-reported aggressiveness and an SDR scale, provided peer ratings of their aggressive behavior, and participated in a game that measured their antisocial behavior. In Study 2, we tried to replicate Study 1 findings by correlating IAT-A scores with self-reported aggressiveness, aggression-related behavior in sports and SDR scores, using a sample of kinesiology students. Finally, in Study 3, we examined the IAT-A's susceptibility to deliberate response distortion by comparing the IAT-A and self-reported aggressiveness between situations of honest responding and a simulated personnel selection.

highly associated, the classification task will be easier when the associated concepts share the same response key than when they require different response keys (Schnabel et al., 2008). This assumption means that an aggressive individual will have faster reactions and make fewer errors when sorting the stimuli referring to the self/aggressive with one response key and others/peaceful with the other response key than when sorting stimuli referring to self/peaceful with one response key and others/aggressive with the other response key. The situation will be reversed for a non-aggressive individual. 1.1. Psychometric properties of IAT for Aggressiveness (IAT-A) Studies that tested the IAT-A resulted in promising but not conclusive findings. The IAT-A yields satisfactory internal consistencies ranging from 0.66 to 0.91 (Banse et al., 2015; Gollwitzer, Banse, Eisenbach, & Naumann, 2007; Grumm, Hein, & Fingerle, 2011). Testretest reliabilities were somewhat lower but still substantial. For example, Banse et al. (2015) reported a test-retest correlation over a week of 0.56. However, findings regarding the relationship of the IAT-A with explicit aggressiveness and its potential in predicting aggressive behavior remain somewhat inconsistent. For example, though most of the studies found that the IAT-A is unrelated to explicit measures of aggressiveness (Richetin, Richardson, & Mason, 2010; Banse et al., 2015 Studies 2, 3 and 4), some studies found surprisingly high correlations between the two measures (e.g., 0.48 for a subsample of volleyball players in Study 1, Banse et al., 2015). Similarly, equivocal findings were observed in the relationship between the IAT-A and aggressive behavior. On the one hand, some studies showed that the IAT-A predicted different types of aggressive behavior at least at the level of selfreport measures of aggressiveness and reported incremental validity over explicit measures (Banse et al., 2015 Studies 1, 2 and 3; Grumm et al., 2011). On the other hand, Banse et al. (2015) found no significant correlation between the IAT-A and observable aggressive behavior, captured with coach aggressiveness ratings, in a subsample of volleyball players (Study 1). Additionally, Richetin et al. (2010) showed that IATA scores predicted aggressive behavior only when the participants were provoked. In addition to the mentioned psychometric properties, an important remaining issue is the relationship between IAT-A scores and socially desirable responding (SDR). Considering that the IAT-A is supposed to be an implicit aggressiveness measure, it should mostly reflect automatic processes and not be susceptible to SDR. Until now, the relationship between the IAT-A and SDR was examined only by testing the relationship between IAT-A scores and SDR scale scores. Again, the findings did not reveal a clear pattern of relations between implicit aggressiveness as measured with the IAT-A and desirable responding as captured with SDR scales. In the study by Banse et al. (2015), IAT-A scores were correlated significantly with SDR scores in Study 1 (−0.36; subsample of volleyball players) and Study 2 (−0.38; both p < 0.01) but were below the p < 0.05 threshold in Study 1 (subsample of icehockey players) and Study 3. In all, current findings with the IAT-A are inconsistent and indicate that more research on the IAT-A's psychometric properties is needed before we can draw reliable conclusions about its usefulness.

2. Study 1 The aim of Study 1 was to test the validity of IAT-A by correlating it with an established explicit measure of aggressiveness, a measure of MSDR and two aggression-related criteria - ratings of aggressive behavior and antisocial behavior in a laboratory situation. We used other-ratings of behavioral aggressiveness as an indicator of aggressive behavior. Although they are not actual behavioral observations, they can be used as a proxy of aggressive behavior in the natural social context because they reflect participants' observable aggressive behavior across multiple situations (Banse et al., 2015). Finally, as an indicator of antisocial behavior we used allocated resources in the dictator game (DG), which has already been used in previous research as an objective measure of antisocial behavior (e.g., Millet & Dewitte, 2009).

1.2. Our study In our study, we aimed to test the relationship of the IAT-A with self-reported aggressiveness and aggressive behavior using several aggression-related criteria with samples that were large enough to obtain stable effects (i.e., N over 100). Based on earlier findings and the theory behind the IAT, we expected that the IAT-A will have little to no association with self-report measures of aggressiveness (H1) and that the IAT-A scores will provide incremental validity in predicting aggression-related behaviors over and above self-report measures of aggressiveness (H2). 96

Personality and Individual Differences 129 (2018) 95–103

M. Parmač Kovačić et al.

(hostility, 8 items, “I am sometimes eaten up with jealousy.”). The 15 items were rated on a seven-point scale ranging from 1 (extremely uncharacteristic of me) to 7 (extremely characteristic of me). The Cronbach's alpha in this study was 0.87.

2.1. Materials and methods 2.1.1. Participants A sample of psychology students from a large Croatian university participated in the study (N = 171). All participants were in their 1st and 3rd year of studying and their mean age was 20.47 years (SD = 1.96). The majority of them were female (77.8%). They were all White and Croatian citizens.

2.1.2.3. M-SDR. As a measure of M-SDR, we used a scale by Parmač Kovačić, Galić, and Jerneić (2014) that consists of 10 items describing socially undesirable behaviors that people who want to present themselves in a positive light deny (e.g., “I have some pretty awful habits.”). The participants' task was to rate their agreement with each item on a five-point scale ranging from 1 (completely false) to 5 (completely true). The Cronbach's alpha of the M-SDR scale in this study was 0.77.

2.1.2. Instruments 2.1.2.1. Implicit aggressiveness. The Croatian version of the Implicit Association Test for Aggressiveness based on the IAT paradigm (Greenwald, McGhee, & Schwartz, 1998) was developed for this study. Since the IAT-A is intended to measure the strength of associations between implicit self-concepts and aggressiveness, two contrasted target categories (“Self” vs. “Others”) and two contrasted aggressiveness-related attribute categories (“Aggressiveness” vs. “Peacefulness”) were used. Stimuli in the IAT-A related to target categories were words that we took from previous research (e.g., Banse et al., 2015; Uhlmann & Swanson, 2004) and adapted to Croatian (10 per category). Words describing attribute categories were selected with the help of a sample of psychology students (N = 73). They were instructed to write down all words (nouns, verbs, or adjectives) that first came to their mind when they thought about aggressiveness/peacefulness. For each attribute category, we selected the 10 most frequently mentioned words for the stimuli. The English translation of the attribute stimuli used in the study is reported in the Appendix. The IAT-A was created using Inquisit 4 Lab (2014). It consisted of seven standard blocks (Schnabel et al., 2008) that we list in Table 1. Blocks 1, 2 and 5 were used as practice blocks, whereas the other blocks (3 + 4 and 6 + 7) were considered test blocks. The order of the test blocks was counterbalanced between the participants. Within every block, word stimuli were presented to participants in the middle of the computer screen and their task was to sort those stimuli into categories written at the left or the right top of the screen. Participants sorted stimuli using keys “E” and “I” on the keyboard. Correct sorting of stimuli was followed with the presentation of new stimuli, whereas incorrect sorting was followed with a big red “x” indicating an error that has to be corrected by pressing the opposite key (the so called built-in error penalty). To calculate the score on the IAT-A, we used the improved D-scoring algorithm with a built-in error penalty and no treatment of response latencies below 400 ms (labeled as D1 in Greenwald, Nosek, & Banaji, 2003). A positive D-score indicated stronger association between the self-concept and aggressiveness, whereas a negative D-score indicated stronger association between the self and peacefulness. The Cronbach's alpha of this newly constructed IAT-A was satisfactory (0.78).

2.1.2.4. Aggression-related behavior. Ratings from other people on the behavioral component of The Aggression Questionnaire (Buss & Perry, 1992) were used as the indicator of aggressive behavior. All items were written in the third person and rated on a seven-point scale ranging from 1 (extremely uncharacteristic of him/her) to 7 (extremely characteristic of him/her). According to Buss and Perry (1992), the behavioral component of aggressiveness involves hurting or harming others physically (9 items, e.g., “If somebody hits him/her, he/she hits back.”) and verbally (5 items, e.g., “He/She can't help getting into arguments when people disagree with him/her.”). We calculated the estimation of aggressive behavior for each participant by averaging the estimates from two raters (correlation between the raters was 0.43; p < 0.01). In this study, the Cronbach's alpha for the averaged ratings from two raters was 0.87. As a measure of antisocial behavior, we used a hypothetical dictator game (DG) in which the participants' task was to assume the role of a “dictator” and to imagine a situation in which they were assigned 100 Kunas (approximately 13.5 EUR) to distribute between themselves and an arbitrary other. The participants had to answer how much they would give to the arbitrary other from 0 to 100 Kunas. As a measure of antisocial behavior, we used the amount of money participants decided to keep for themselves, which we calculated by simply deducting the amount of money allocated to the other from 100. Since it has been shown that real DG, with actual money and recipients, does not differ from hypothetical DG in terms of amount of money allocated to the other (Ben-Ner, Kramer, & Levy, 2008), we believe that the procedure we used in this study adequately captured the construct of antisocial behavior. 2.1.3. Procedure The study took part at a university laboratory where the participants completed the IAT-A and computerized versions of the self-report instruments. The testing was conducted individually, with a trained test administrator who read the instruction before each instrument. All participants completed the IAT-A, the explicit measure of aggressiveness and the M-SDR scale. Before completing implicit and explicit measures of aggressiveness, a portion of the sample (n = 69) participated in the dictator game. Other's ratings of aggressive behavior were collected through an online questionnaire. As their raters, the participants needed to select two people with whom they were in a close contact for more than six

2.1.2.2. Explicit aggressiveness. Self-ratings on the cognitive-emotional component of The Aggression Questionnaire (Buss & Perry, 1992) were used as an explicit measure of aggressiveness. According to Buss and Perry (1992), this component of aggressiveness involves physiological arousal and preparation for aggression (anger, 7 items, sample item, “I have trouble controlling my temper.”) and feelings of ill will and injustice Table 1 Block sequence in the IAT-A. Block

Task

No. of trials

Left key “E”

Right key “I”

1 2 3+4 5 6+7

Sorting of target stimuli I Sorting of attribute stimuli Simultaneous sorting of target and attribute stimuli (compatible tasks I + II) Sorting of target stimuli II Simultaneous sorting of target and attribute stimuli (incompatible tasks I + II)

20 20 20 + 40 20 20 + 40

Self Aggressiveness Self + Aggressiveness Others Others + Aggressiveness

Others Peacefulness Others + Peacefulness Self Self + Peacefulness

97

Personality and Individual Differences 129 (2018) 95–103

M. Parmač Kovačić et al.

(2011), this correlation was higher than the one obtained between an explicit measure of aggressiveness and other's ratings of aggressive behavior. Contrary to our expectations, the IAT-A was not correlated with antisocial behavior expressed in the DG, whereas the correlation between self-reported aggressiveness and behavior in the DG game was marginally significant (p < 0.10). To test whether the IAT-A has incremental validity in predicting aggression-related behaviors above the explicit measure of aggressiveness and to test the interaction between the IAT-A and explicit aggressiveness in predicting aggression-related behaviors, we conducted two hierarchical regression analyses. In the analyses, other reports of aggressive behavior and DG scores served as the criteria; explicit aggressiveness was entered in the first block, IAT-A scores in the second block, and the interaction between these two variables in the third block. The results of conducted regression analyses are reported in Table 3.1 In line with our expectations, the IAT-A was a significant predictor of other's reports of aggressive behavior over and above the explicit measure of aggressiveness, explaining an additional 5% of the variance. At the same time, aggressiveness self-reports were non-significant predictors of other-reports of aggressive behavior. This finding is consistent with previous findings in which incremental validities ranged from 5 to 15% of variance in aggressive behavior (Banse et al., 2015; Grumm et al., 2011; Teubel, Asendorpf, Banse, & Schnabel, 2011). The interaction of the IAT-A and explicit aggressiveness did not significantly predict other's reports of aggressive behavior. When the DG score was used as the criterion, neither explicit aggressiveness nor implicit aggressiveness was significant predictors of aggression-related behavior. However, their interaction significantly predicted the antisocial behavior in the DG, explaining 9% of the variance. The participants who were both implicitly and explicitly aggressive expressed higher levels of antisocial behavior by keeping more money to themselves (see Fig. 1). Obviously, only one type of aggressiveness was insufficient for displaying antisocial behavior in the lab situation. Overall, the results suggest that the IAT-A could be a valid measure of implicit aggressiveness. It had satisfactory reliability, was only weakly related to an explicit measure of aggressiveness and was unrelated to the conceptually closest measure of SDR. Moreover, IAT-A scores explained the other's reports of aggressive behavior over and above the explicit measure, and their interaction with the explicit measure predicted antisocial behavior in a lab situation. However, having in mind the relatively small effect sizes, before making any firm conclusions, we sought to replicate the findings using different criteria for aggressive behavior with a different sample of participants.

Table 2 Descriptive statistics and intercorrelations of all measures used in Study 1. M (SD) 1. IAT-A 2. 3. 4. 5.

Aggressiveness (self-reports) M-SDR Aggression (other-ratings)a Antisocial behavior in dictator gameb

−0.48 (0.34) 3.14 (0.94) 3.18 (0.64) 2.63 (0.69) 56.94 (14.99)

1

2

3

4

0.20⁎ −0.10 0.25⁎⁎ 0.12

−0.46⁎⁎ 0.14† 0.21†

−0.04 −0.15

−0.03

Notes: a Correlations were based on participants with two ratings (n = 153). b Correlations were based on the n = 69 respondents who participated in the dictator game. † p < 0.10. ⁎ p < 0.05. ⁎⁎ p < 0.01.

months. The participants forwarded the link of the on-line questionnaire to the raters. We collected both other person reports for 152 participants (89% of the sample). Participation in the study was voluntary and anonymous. All participants completed the instruments with a password. The raters completed the rating forms under the same password that was provided to them by the participants. In this way, we secured participants' anonymity and were able to connect the data from the same participant with the different instruments/sources. 2.2. Results and discussion Descriptive statistics for the measures in this study and their intercorrelations are reported in Table 2. The mean IAT-A score obtained in this study was −0.48 (SD = 0.34). This result, which is negative and moderate in size, indicates low levels of implicit aggressiveness in our sample. On average, our participants have stronger associations between implicit self-concept and peacefulness than between implicit self-concept and aggressiveness. This finding is in line with findings from previous studies conducted in high school and college students (Grumm et al., 2011; Uhlmann & Swanson, 2004). To test the validity of the IAT-A, we examined the correlation of the IAT-A scores with the explicit measure of aggressiveness. As expected, we obtained a significant but small correlation between these two measures. This result is in accordance with previous findings on the IAT-A (e.g., Banse et al., 2015; Richetin et al., 2010), which suggested that it captures a part of aggressiveness that is related but distinct from explicit aggressiveness. Additionally, this result is in line with the metaanalysis by Hofmann, Gawronski, Gschwendner, Le, and Schmitt (2005) that showed that, on average, IAT scores have low to moderate correlations with explicit measures of the same construct (average r = 0.24). Our expectations regarding the relation of the M-SDR scale with the two aggressiveness measures were also met. While the correlation between the explicit measure of aggressiveness and the M-SDR was moderately high, the relation between the IAT-A and M-SDR scales was non-significant. These results indicate that, while scores on explicit aggressiveness significantly overlap with moralistic SDR, the results of the implicit aggressiveness measure (i.e., IAT-A) are completely unrelated to SDR in that domain. Finally, we correlated the IAT-A scores with two indicators of aggression-related behavior (other's ratings of behavioral aggressiveness and antisocial behavior in a laboratory situation). The results showed that the IAT-A was significantly correlated with other's ratings of aggressive behavior. Individuals with higher scores on the IAT-A were described as more aggressive by their raters. Though somewhat lower than we expected from the work of Banse et al. (2015) and Grumm et al.

3. Study 2 The aim of the second study was to replicate Study 1 findings with another sample of participants. For this study, we decided to use a large sample of kinesiology students that are active sport players for two important reasons. First, implicit aggressiveness measured with the IAT was shown to be positively related to sport performance (Teubel et al., 2011). Therefore, we expected that implicit aggressiveness might be more developed within a sample of active sports players than among psychology students. Relatedly, kinesiology students' cohorts are dominated by males to about the same extent as psychology students 1 We have also conducted regression analyses where age and gender were included in the first step of the analyses as controls. Though some of the effects did somewhat decrease in size, they did not put into question our general conclusion about the validity of the IAT-A since all effects remained significant or marginally significant. Considering that inclusion of control variables without clear theoretical rationale in analyses is inconsistent with recent recommendations about the use of control variables in correlational research (see Becker et al., 2016; Spector & Brannick, 2011), we decided to report the results of the analyses without them.

98

Personality and Individual Differences 129 (2018) 95–103

M. Parmač Kovačić et al.

Table 3 Results of regression analyses for predicting aggression-related behaviors in Study 1. Aggression (other-ratings)a

Criterion

Predictor Aggressiveness (self-reports) IAT-A Aggressiveness × IAT-A R2 Adjusted R2 ΔR2

Antisocial behavior in dictator gameb

1st block

2nd block

3rd block

1st block

2nd block

3rd block

β

β

β

β

β

β

0.21

0.19 0.07

0.04† 0.03

0.05 0.02 0.00

0.13 0.10 0.30⁎ 0.14⁎ 0.10 0.09⁎

†

0.14

0.11 0.23⁎⁎

0.02† 0.01

0.07⁎⁎ 0.06 0.05⁎⁎

0.11 0.23⁎⁎ 0.04 0.07⁎ 0.05 0.00

†

Notes: a Analysis was based on participants with two ratings (n = 147). b Analysis was based on n = 69 respondents who participated in dictator game. † p < 0.10. ⁎ p < 0.05. ⁎⁎ p < 0.01.

Dictator game score

80,00 70,00

(Parmač Kovačić et al., 2014) as in Study 1. The Cronbach's alpha of the M-SDR scale in this study was 0.76.

Low explicit aggressiveness

High explicit aggressiveness

3.1.2.4. Aggression-related behavior. For measuring aggression-related behaviors, we used two different scales. First, we used Maxwell's (2004) Aggression Scale (AS), which consisted of four items assessing the frequency of the expression of aggressive acts that might appear in virtually any sport. This included both aggressive acts toward the opponent (e.g., “I intentionally anger my opponents”) and towards the officials (“I argue with officials”). The second scale we used was the Antisocial Behavior in Sport Scale (ABSS; Kavussanu & Boardley, 2009), which consisted of eight items describing poor sport behaviors directed toward opponents (e.g., “Deliberately fouled an opponent.”). For both scales, the participants' task was to rate how often they engage in aggression-related behaviors on a scale anchored by 1 (never) and 5 (very often). The reliability (Cronbach's alpha) of the AS and ABSS were 0.75 and 0.87, respectively, in this study.

60,00 50,00 40,00

Low IAT-A score

High IAT-A score

Fig. 1. Interaction between the IAT-A and explicit aggressiveness in predicting the amount of money participants kept for themselves in a dictator game in Study 1.

are dominated by females. We believed that a replication of Study 1 findings in such a sample, using measures of aggressive behavior in sports, would be an important step in testing the validity of the IAT-A.

3.1.3. Procedure The study was conducted in the classroom during the participants' class session. All participants completed the IAT-A on a computer and the set of paper-and-pencil instruments consisting of an explicit measure of aggressiveness, the M-SDR scale and the two measures of aggression-related behaviors. The testing was conducted in small groups (approximately 20 participants per group) with a trained test administrator who read the instructions before completing each of the instruments. Participation in the study was voluntary and anonymous. To secure participants' anonymity and connect the data on the same participant from different instruments, all participants completed the instruments with a password.

3.1. Materials and methods 3.1.1. Participants A sample of 105 kinesiology students from a large Croatian university participated in this study. Only students who participated in at least one sports competition during the last six months were included in the sample. They were all in their 2nd year of studying and their mean age was 20.63 (SD = 1.08). The sample mainly consisted of male students (77.1%). They were all White and Croatian citizens. 3.1.2. Instruments 3.1.2.1. Implicit aggressiveness. The IAT-A described in the Study 1 was used for measuring implicit aggressiveness. The Cronbach's alpha of the IAT-A in this study was also satisfactory (0.81).

Table 4 Descriptive statistics and intercorrelations of all measures used in Study 2.

3.1.2.2. Explicit aggressiveness. Self-ratings of the cognitive-emotional component of the short-form of The Aggression Questionnaire (Webster et al., 2014) were used as an explicit measure of aggressiveness. In this brief measure of aggressiveness, the cognitive-emotional component is assessed with a total of six items that best represent anger and hostility (three items each). As in the original version, all items were rated on a seven-point scale ranging from 1 (extremely uncharacteristic of me) to 7 (extremely characteristic of me). The Cronbach's alpha in this study was 0.72.

1. IAT-A 2. Aggressiveness (selfreports) 3. M-SDR 4. Aggression in sports 5. Antisocial behavior in sports Note: ⁎⁎ p < 0.01.

3.1.2.3. M-SDR. We measured M-SDR scores using the same scale 99

M (SD)

1

−0.49 (0.39) 3.41 (1.07)

0.00

3.04 (0.64) 2.38 (0.88) 2.59 (0.87)

−0.13 0.30⁎⁎ 0.27⁎⁎

2

3

4

−0.59⁎⁎ 0.25⁎⁎ 0.32⁎⁎

−0.42⁎⁎ −0.43⁎⁎

0.75⁎⁎

Personality and Individual Differences 129 (2018) 95–103

M. Parmač Kovačić et al.

Table 5 Results of regression analyses for predicting aggression-related behaviors in Study 2. Criterion

Predictor Aggressiveness self-reports IAT-A Aggressiveness × IAT-A R2 Adjusted R2 ΔR2

Aggression in sports

Antisocial behavior in sports

1st block

2nd block

3rd block

1st block

2nd block

3rd block

β

β

β

β

β

β

⁎⁎

⁎⁎

0.26

0.26 0.30⁎⁎

0.07⁎⁎ 0.06

0.16⁎⁎ 0.14 0.09⁎⁎

⁎

0.24 0.30⁎⁎ 0.12 0.17⁎⁎ 0.15 0.01

0.31

⁎⁎

0.10⁎⁎ 0.09

⁎⁎

0.31 0.26⁎⁎ 0.17⁎⁎ 0.15 0.07⁎⁎

0.30⁎⁎ 0.26⁎⁎ 0.07 0.17⁎⁎ 0.15⁎⁎ 0.01

Notes: ⁎ p < 0.05. ⁎⁎ p < 0.01.

the variance in the indicators. Aggressiveness self-reports remained significant predictors of both indicators of aggression-related behavior once the IAT-A scores were entered in the equation. The interaction of implicit and explicit aggressiveness predicted neither of the criteria. As in Study 1, Study 2 results suggested that the IAT-A is a valid measure of implicit aggressiveness. It had satisfactory reliability, it was not related to M-SDR scale scores or an explicit measure of aggressiveness, but explained two different aggression-related behaviors over and above the explicit aggressiveness measure. However, before declaring the IAT-A a valid measure of implicit aggressiveness, we wanted to address one more important characteristic with which an implicit measure of personality should comply – resistance to faking.

3.2. Results and discussion Descriptive statistics and the correlations of the IAT-A scores with the explicit measure of aggressiveness, the M-SDR scale, and the two indicators of aggression-related behavior in sports are presented in Table 4. Contrary to our expectations, the mean D value in the sample of kinesiology students was almost identical to the one obtained in psychology students (Mkinesiology = −0.49; Mpsychology = −0.48; t (273) = 0.22; p > 0.05). On average, they also had stronger associations between implicit self-concept and peacefulness then between implicit self-concept and aggressiveness. On the explicit measure of aggressiveness, although kinesiology students had significantly higher scores than psychology students (Mkinesiology = 3.41 vs. Mpsychology = 3.14; t (266) = 2.17; p < 0.05; d = 0.27), their mean score was still below the mean scale value. With the aim of testing the validity of the IAT-A, we first explored the correlations among the IAT-A scores with an explicit measure of aggressiveness. In line with our expectations, the IAT-A was not significantly correlated with the self-ratings of aggressiveness, again suggesting that the IAT-A captured the part of aggressiveness that is distinct from explicit aggressiveness. The pattern of correlations between the M-SDR scale and the two aggressiveness measures was also similar to Study 1. In terms of Cohen's (1988) effect sizes, there was a large overlap between self-reported aggressiveness and M-SDR, while the correlation between IAT-A and M-SDR was non-significant. The correlations between the IAT-A and aggression-related behavior in sports are also in line with our expectations. The IAT-A scores showed significant correlations with both measures. This correlation means that implicitly aggressive individuals, as measured by the IAT-A, tend to express more aggressive and more antisocial behaviors in sport. It is also important to note that, in accordance with the literature (Schnabel et al., 2008), these correlations are at the level of those obtained between the explicit measure of aggressiveness and the criteria. To test whether IAT-A has incremental validity in predicting aggression-related behavior above the explicit measure of aggressiveness and examine importance of the interaction between implicit and explicit aggressiveness in prediction of this behavior, as in Study 1, we conducted two regression analyses with AS and ABSS scores as dependent variables. In both analyses, explicit aggressiveness was entered in the first block, the IAT-A in the second block and the interaction term in the third block. The results of performed regression analyses are shown in Table 5.2 In line with our expectations, IAT-A scores were a significant predictor of both indicators of aggression-related behavior over and above explicit measure of aggressiveness, explaining an additional 7–9% of

2

4. Study 3 The aim of this study was to test the susceptibility of the IAT-A to deliberate response distortion (i.e., faking), and, thus, contribute to the validity of the IAT-A as an implicit aggressiveness measure. The first two studies showed that, unlike self-report measures of aggressiveness, the IAT-A does not overlap with a measure of SDR. However, a stronger test of the IAT-A's resistance to response distortion should come from a study where participants are instructed to fake their scores on that test. In Study 3, in a subsample of participants from Study 1, we also collected responses on the IAT-A and an explicit measure of aggressiveness in a situation in which they were instructed to “fake” the tests. In comparison to some other research that used generic instructions such as “try to make a very good impression” (e.g., Egloff & Schmukle, 2002), we instructed participants to present themselves as ideal candidates in simulated personnel selection for a position where aggressiveness is a highly undesirable trait – a nurse position. Taking together the assumption that the IAT-A should assess automatic evaluation and the faking literature that clearly shows that participants can fake personality questionnaires when instructed to do so (e.g., Viswesvaran & Ones, 1999), we expected larger differences between honest responding and simulated selection on the aggressiveness questionnaire than on the IAT-A. 4.1. Materials and methods 4.1.1. Participants This study was conducted with a subsample of psychology students that participated in Study 1 (n = 70). All participants were in their 1st year of studying and their mean age was 19.34 years (SD = 1.93). The majority of them were female (75.7%). They were all White and Croatian citizens. 4.1.2. Instruments 4.1.2.1. Implicit personality. The IAT-A, constructed and described in the Study 1, was used to measure implicit aggressiveness. In the

See Footnote 1.

100

Personality and Individual Differences 129 (2018) 95–103

M. Parmač Kovačić et al.

the three relationships and replicated the findings using different samples and different research paradigms. Second, the relationship between the IAT-A and SDR, important for a measure that should dominantly assess automatic aspects of aggressiveness, has until now been tested only by examining the relationship of IAT-A scores with the scores on SDR scales. In our study, in addition to testing the relationship using an SDR scale that should theoretically be the closest to the IAT-A, we employed the faking paradigm in which we compared the IAT-A and self-reported aggressiveness scores between situations where participants responded honestly or were instructed to fake their scores. Finally, most of the studies on the IAT for self-concept measurement in general, and the IAT-A in particular, have been conducted using English and German language stimuli. Though there is no reason to expect much different findings cross-culturally, a test in another language (i.e., Croatian) certainly contributes to the universal validity of the research paradigm. Our three interrelated studies showed that the IAT paradigm might indeed be useful for implicit aggressiveness measurement. In addition to displaying decent basic psychometric characteristics such as internal consistency, most of our hypotheses that followed from the theoretical work on the IAT were supported. We showed that the IAT-A developed for this study (1) was non-significantly or weakly related to self-reported aggressiveness (H1), (2) predicted aggression-related behavior over and above self-reported aggressiveness (H2), (3) was unrelated to SDR, as measured with the SDR scale, and (4) was unsusceptible to deliberate response distortion (H4a & H4b). The only hypothesis that was only partially supported was about the interaction between the IAT-A and self-reported aggressiveness in predicting aggression-related behavior, which was confirmed only for antisocial behavior in the Dictator Game. Additionally, we believe that some aspects of our findings contribute to the general validity of IAT research paradigm. For example, obtained mean values on our IAT-A were completely consistent with the studies that already used the IAT paradigm for aggressiveness measurement (e.g., Grumm et al., 2011; Uhlmann & Swanson, 2004). Moreover, the correlations of the IAT-A with explicit measures of aggressiveness in our research were in line with the metaanalytical findings about the relationship between IATs and self-reported measures of the same construct (Hofmann et al., 2005). These findings suggest that our findings are probably not limited by the fact that the used stimuli belonged to a specific language. The results have a clear, practical implication for personality assessment. Although psychologists dominantly measure aggressiveness using self-report measures, our results suggest that measurement of implicit aggressiveness with the IAT-A might also be valuable. Though in most of the cases the two types of measures should be used together, our findings suggest that in the case of some criteria and certain situations, the IAT-A might even be preferable to self-report measures of aggressiveness. The latter especially refers to situations where respondents are motivated to present themselves in a desirable light. There are several main limitations and related suggestions for future research. First, in our study, we used self-ratings and other's ratings as indicators of aggression-related behaviors. Since self-ratings and other's ratings are subject to different biases, such as self-enhancement in the case of self-reports or the halo effect in the case of other's reports, objective indicators of aggressive behavior in real or laboratory situations should be used in future studies in order to strengthen the conclusion about the IAT-A's validity. Second, susceptibility to faking was determined by comparing participants' results on the IAT-A between honest responding and a simulated selection situation. In comparison to self-report scales with which participants are well acquainted and where it is evident to them how to distort the answers in order to present themselves in a positive light, the IAT is a relatively new assessment method with which participants have no experience and do not know how it functions. Maybe, if we gave them specific instructions on how to fake IAT (as, for example, Stieger et al., 2011 for anxiety), they would be more successful. However, we still believe that, even

Table 6 Comparison of IAT-A scores and aggressiveness self-report results between honest responding and simulated selection situations in Study 3.

1. IAT-A 2. Aggressiveness (selfreports)

Honest

Simulated selection

M (SD)

M (SD)

−0.44 (0.33) 3.22 (0.99)

−0.42 (0.34) 1.47 (0.58)

t (df)

Effect size (Cohen's d)

−0.44 (69) 15.45 (68)⁎⁎

0.06 −2.16

Note: ⁎⁎ p < 0.01.

subsample used for Study 3, the Cronbach's alpha of the IAT-A in the situation of honest responding was 0.75, and in the simulated selection, it was 0.81. 4.1.2.2. Explicit aggressiveness. As described in Study 1, self-ratings on the cognitive-emotional component of The Aggression Questionnaire (Buss & Perry, 1992) were used as an explicit measure of aggressiveness in both situations. The Cronbach's alpha in this subsample of participants was 0.88 in the situation of honest responding and 0.90 in the simulated selection situation. 4.1.3. Procedure A week after completing the IAT-A and an explicit measure of aggressiveness in the situation of honest responding (described in Study 1), participants completed the same measures in the simulated selection situation for a nurse position. The participants were instructed not to respond honestly, but to present themselves as ideal candidates for a nurse position. This study also took part at the university laboratory, with all aspects of the procedure being the same as in Study 1. 4.2. Results and discussion The comparison of the responses on the IAT-A and the explicit measure of aggressiveness between honest responding and simulated selection is presented in Table 6. As seen in Table 6, the results are in line with our expectations. While in the simulated selection situation the participants scored significantly lower on the explicit measure of aggressiveness than in the situation of honest responding, the mean result on the IAT-A was almost identical between the two response situations. In addition to that, the correlation between two response situations was lower for the explicit measure of aggressiveness (r = 0.38) than for the IAT-A (r = 0.49, both p < 0.01). These results revealed that the IAT-A is less susceptible to deliberate response distortion than the self-report measure of explicit aggressiveness. Given that aggressiveness is an undesirable trait, this finding might have important practical implications in situations where we are motivated to identify individuals with high levels of that trait. However, the IAT-A's stability between honest responding and faking situations was far from perfect. This finding was consistent with Schnabel et al.'s (2008) observation that there are systematic occasionto-occasion variations in IAT scores that stem from changes in either the association strengths being measured or unidentified additional sources in variance in the measurement procedure. 5. General discussion We believe that our study offers several important contributions. First, there are only a few studies that tested the relationship between the IAT-A and self-reported aggressiveness, aggressive behavior and SDR, and their findings were not unambiguous. In our study, we tested 101

Personality and Individual Differences 129 (2018) 95–103

M. Parmač Kovačić et al.

(Perugini, 2005). According to the additive model, implicit and explicit aggressiveness should have unique contributions to the prediction of aggressive behavior (i.e., one of the constructs adds over and above the other), whereas a double dissociation model assumes that implicit and explicit measures predict different types of behavior (e.g., implicit measures predict spontaneous behavior and explicit measures predict controlled behavior; Asendorpf, Banse, & Mücke, 2002). In our research, we found certain support for the additive model in Study 2 and some support for the double dissociation in the case of other's reports of aggressive behavior in Study 1. Future research should consider the three predictive models in the case of the IAT-A and specify the conditions and the criteria where each of these models “work”.

with more specific instruction, faking on the IAT-A would be less pronounced than on an explicit aggressiveness measure. Moreover, our research is based on student samples of participants. Although we tried to ensure that our situations resembled real-life assessment situations, more ecologically valid designs including different samples of participants and real assessment situations would be welcomed in the future studies. Finally, future research should better address the relationship between the IAT-A and self-reported aggressiveness in predicting different criteria. Based on earlier research that tested the interaction between implicit and explicit aggressiveness in predicting aggressive behavior (Bing et al., 2007; Frost, Ko, & James, 2007), we expected to show that individuals high on both types of aggressiveness will be especially prone to aggressive behavior. However, our expectation was met only for one of the criteria, antisocial behavior in the dictator game. For other criteria, a different pattern of relationship between implicit and explicit aggressiveness in predicting aggressive behavior was observed. In addition to the described interaction model, there are two other possible predictive models relating implicit and explicit constructs in predicting different criteria: additive and double dissociation models

Acknowledgments Funding This work was fully supported by Croatian Science Foundation under the project 6719.

Appendix A. Appendix Table A1 The English translation of the attribute stimuli of the IAT-A. Aggressiveness

Peacefulness

Aggression Conflict Violence To fight To attack To argue Anger To hit To threaten To abuse

Peacefulness Gentleness Compromise Agreement To cooperate To calm down To give in To reconcile To please Peacemaker

Appendix B. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.paid.2018.03.002.

Dodge, K. A. (1980). Social cognition and children's aggressive behavior. Child Development, 51, 162–170. http://dx.doi.org/10.2307/1129603. Egloff, B., & Schmukle, S. C. (2002). Predictive validity of an implicit association test for assessing anxiety. Journal of Personality and Social Psychology, 83, 1441–1455. http:// dx.doi.org/10.1037//0022-3514.83.6.1441. Frost, B. C., Ko, C. H. E., & James, L. R. (2007). Implicit and explicit personality: A test of a channeling hypothesis for aggressive behavior. Journal of Applied Psychology, 92, 1299–1319. http://dx.doi.org/10.1037/0021-9010.92.5.1299. Gollwitzer, M., Banse, R., Eisenbach, K., & Naumann, A. (2007). Effectiveness of the Vienna social competence training on explicit and implicit aggression. European Journal of Psychological Assessment, 23, 150–156. http://dx.doi.org/10.1027/10155759.23.3.150. Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. (1998). Measuring individual differences in implicit cognition: The implicit association test. Journal of Personality and Social Psychology, 74, 1464–1480. http://dx.doi.org/10.1037//0022-3514.74.6. 1464. Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding and using the implicit association test: I. An improved scoring algorithm. Journal of Personality and Social Psychology, 85, 197–216. http://dx.doi.org/10.1037/0022-3514.85.2.197. Grumm, M., Hein, S., & Fingerle, M. (2011). Predicting aggressive behavior in children with the help of measures of implicit and explicit aggression. International Journal of Behavioral Development, 35, 352–357. http://dx.doi.org/10.1177/ 0165025411405955. Guerra, N. G., Huesmann, L. R., & Hanish, L. (1995). The role of normative beliefs in children's social behavior. In N. Eisenberg (Vol. Ed.), Review of personality and social psychology. 15. Review of personality and social psychology (pp. 140–158). Newbury Park, CA: Sage Publications.

References Asendorpf, J. B., Banse, R., & Mücke, D. (2002). Double dissociation between implicit and explicit personality self-concept: The case of shy behavior. Journal of Personality and Social Psychology, 83, 380–393 (doi:10.1037%2F0022-3514.83.2.380). Banse, R., Messer, M., & Fischer, I. (2015). Predicting aggressive behavior with the aggressiveness-IAT. Aggressive Behavior, 41, 65–83. http://dx.doi.org/10.1002/ab. 21574. Becker, T. E., Atinc, G., Breaugh, J. A., Carlson, K. D., Edwards, J. R., & Spector, P. E. (2016). Statistical control in correlational studies: 10 essential recommendations for organizational researchers. Journal of Organizational Behavior, 37, 157–167. http:// dx.doi.org/10.1002/job.2053. Ben-Ner, A., Kramer, A., & Levy, O. (2008). Economic and hypothetical dictator game experiments: Incentive effects at the individual level. The Journal of Socio-Economics, 37, 1775–1784. http://dx.doi.org/10.1016/j.socec.2007.11.004. Bergman, S. M., McIntyre, M. D., & James, L. R. (2004). Identifying the aggressive personality. Journal of Emotional Abuse, 4(3–4), 81–93. http://dx.doi.org/10.1300/ j135v04n03_06. Bing, M. N., Stewart, S. M., Davison, H. K., Green, P. D., McIntyre, M. D., & James, L. R. (2007). An integrative typology of personality assessment for aggression: Implications for predicting counterproductive workplace behavior. Journal of Applied Psychology, 92, 722–744. http://dx.doi.org/10.1037/0021-9010.92.3.722. Buss, A. H., & Perry, M. (1992). The aggression questionnaire. Journal of Personality and Social Psychology, 63, 452–459. http://dx.doi.org/10.1037//0022-3514.63.3.452. Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum Associateshttp://dx.doi.org/10.4324/9780203771587.

102

Personality and Individual Differences 129 (2018) 95–103

M. Parmač Kovačić et al.

Perugini, M. (2005). Predictive models of implicit and explicit attitudes. British Journal of Social Psychology, 44, 29–45. http://dx.doi.org/10.1348/014466604x23491. Richetin, J., Richardson, D. S., & Mason, G. D. (2010). Predictive validity of IAT aggressiveness in the context of provocation. Social Psychology, 41, 27–34. http://dx. doi.org/10.1027/1864-9335/a000005. Schnabel, K., Asendorpf, J. B., & Greenwald, A. G. (2008). Using implicit association tests for the assessment of implicit personality self-concept. Personality measurement and testing: . Vol. 2. The SAGE handbook of personality theory and assessment (pp. 508–528). . http://dx.doi.org/10.4135/9781849200479.n24. Spector, P. E., & Brannick, M. T. (2011). Methodological urban legends: The misuse of statistical control variables. Organizational Research Methods, 14, 287–305. http://dx. doi.org/10.1177/1094428110369842. Stieger, S., Göritz, A. S., Hergovich, A., & Voracek, M. (2011). Intentional faking of the single category implicit association test and the implicit association test. Psychological Reports, 109, 219–230. http://dx.doi.org/10.2466/03.09.22.28.pr0.109.4.219-230. Strack, F., & Deutsch, R. (2004). Reflective and impulsive determinants of social behavior. Personality and Social Psychology Review, 8, 220–247. http://dx.doi.org/10.1207/ s15327957pspr0803_1. Teubel, T., Asendorpf, J. B., Banse, R., & Schnabel, K. (2011). Implicit but not explicit aggressiveness predicts performance outcome in basketball players. International Journal of Sport Psychology, 42, 390–400. Uhlmann, E., & Swanson, J. (2004). Exposure to violent video games increases automatic aggressiveness. Journal of Adolescence, 27(1), 41–52. http://dx.doi.org/10.1016/j. adolescence.2003.10.004. Viswesvaran, C., & Ones, D. S. (1999). Meta-analyses of fakability estimates: Implications for personality measurement. Educational and Psychological Measurement, 59, 197–210. http://dx.doi.org/10.1177/00131649921969802. Webster, G. D., DeWall, C. N., Pond, R. S., Deckman, T., Jonason, P. K., Le, B. M., ... Smith, C. V. (2014). The brief aggression questionnaire: Psychometric and behavioral evidence for an efficient measure of trait aggression. Aggressive Behavior, 40, 120–139. http://dx.doi.org/10.1002/ab.21507.

Hofmann, W., Gawronski, B., Gschwendner, T., Le, H., & Schmitt, M. (2005). A metaanalysis on the correlation between the implicit association test and explicit selfreport measures. Personality and Social Psychology Bulletin, 31, 1369–1385. http://dx. doi.org/10.1177/0146167205275613. Huesmann, L. R. (1988). An information-processing model for the development of aggression. Aggressive Behavior, 14, 13–24. http://dx.doi.org/10.1002/10982337(1988)14:1<13::aid-ab2480140104>3.0.co;2-j. Inquisit for Windows 4.0 (2014). Computer software. Seattle, WA: Millisecond Software. Available from http://www.millisecond.com/download/. James, L. R., & LeBreton, J. M. (2012). Assessing the implicit personality through conditional reasoning (1st ed.). Washington: American Psychological Associationhttp://dx.doi. org/10.1037/13095-000. Kavussanu, M., & Boardley, I. D. (2009). The prosocial and antisocial behavior in sport scale. Journal of Sport and Exercise Psychology, 31, 97–117. http://dx.doi.org/10. 1123/jsep.31.1.97. Maxwell, J. P. (2004). Anger rumination: An antecedent of athlete aggression? Psychology of Sport and Exercise, 5, 279–289. http://dx.doi.org/10.1016/s1469-0292(03) 00007-4. McClelland, D. C., Koestner, R., & Weinberger, J. (1989). How do self-attributed and implicit motives differ? Psychological Review, 96, 690–702. http://dx.doi.org/10. 1037//0033-295x.96.4.690. Millet, K., & Dewitte, S. (2009). The presence of aggression cues inverts the relation between digit ratio (2D: 4D) and prosocial behaviour in a dictator game. British Journal of Psychology, 100, 151–162. http://dx.doi.org/10.1348/ 000712608x324359. Parmač Kovačić, M., Galić, Z., & Jerneić, Ž. (2014). Social desirability scales as indicators of self-enhancement and impression management. Journal of Personality Assessment, 96, 532–543. http://dx.doi.org/10.1080/00223891.2014.916714. Paulhus, D. L. (2002). Social desirable responding: The evolution of a construct. In H. I. Brown, D. N. Jackson, & D. E. Wiley (Eds.). The role of constructs in psychological and educational measurement (pp. 49–69). Mahwah NJ: Erlbaum.

103