Abstract
Individuals differ in how they weigh their own utility versus others’. This tendency codefines the dark factor of personality (D), which is conceptualized as the underlying disposition from which all socially and ethically aversive (dark) traits arise as specific, flavored manifestations. We scrutinize this unique theoretical notion by testing, for a broad set of 58 different traits and related constructs, whether any predict how individuals weigh their own versus others’ utility in proactive allocation decisions (i.e., social value orientations) beyond D. These traits and constructs range from broad dimensions (e.g., agreeableness), to aversive traits (e.g., sadism) and beliefs (e.g., normlessness), to prosocial tendencies (e.g., compassion). In a large-scale longitudinal study involving the assessment of consequential choices (median N = 2,270; a heterogeneous adult community sample from Germany), results from several hundred latent model comparisons revealed that no meaningful incremental variance was explained beyond D. Thus, D alone is sufficient to represent the social preferences inherent in socially and ethically aversive personality traits.
Keywords
Socially aversive and unethical behaviors—from everyday individual antisocial or illegal activities to large-scale cases of exploitation, fraud, or worse—pose severe threats on various levels. In an effort to explain corresponding behaviors of individuals, personality psychology has long focused on aversive (sometimes called “dark”) traits (Furnham et al., 2013; Marcus & Zeigler-Hill, 2015; Paulhus & Williams, 2002)—that is, relatively enduring tendencies of affect, thought, and behavior, such as greed (Krekels & Pandelaere, 2015), Machiavellianism (Christie & Geis, 1970), or sadism (O’Meara et al., 2011).
Many aversive traits show considerable conceptual and empirical overlap, and there is strong evidence for a common denominator among them (Bertl et al., 2017; McLarnon & Tarraf, 2017; Muris et al., 2017; O’Boyle et al., 2015; Schermer & Jones, 2020; Schreiber & Marcus, 2020; Vize et al., 2018; Watts et al., 2017). Indeed, this is unsurprising given that all these traits are intended to account for some forms of aversive behavior. Correspondingly, their common core, the so-called dark factor of personality (D), has been defined as “the tendency to maximize one’s individual utility—disregarding, accepting, or malevolently provoking disutility for others—accompanied by beliefs that serve as justifications” (Moshagen et al., 2018, p. 657). In line with this definition, D predicts diverse behaviors that involve utility maximization at others’ expense (Bader et al., in press; Moshagen et al., 2018; Moshagen, Zettler, Horsten, & Hilbig, 2020) and beliefs that can serve to justify such behaviors (Hilbig et al., in press; Horsten et al., 2021; Moshagen, Zettler, & Hilbig, 2020).
Importantly, D is theoretically conceptualized as the (fluid) underlying disposition from which all aversive traits arise as specific, flavored manifestations (Moshagen et al., 2018; Moshagen, Zettler, & Hilbig, 2020; Zettler et al., 2021). Thus, much as the g factor of intelligence captures the general-mental-ability component of any ability-related task, D is assumed to represent what makes any aversive trait aversive. In turn, whereas aversive traits by their very nature entail features of the general disposition of D (i.e., subjectively justified utility maximization at others’ expense), the traits may differ in (a) how prominently a particular defining feature of D is represented in a trait and (b) which nonaversive features that lie outside the scope of D a trait may additionally entail (if any). For example, the defining feature of D that individuals malevolently provoke disutility on others is particularly prominent in sadism, whereas the defining feature of D that individuals hold justifying beliefs is particularly prominent in narcissism. Moreover, narcissism may involve the unique and per se nonaversive aspect of vulnerability (Crowe et al., 2019; Starlinger et al., 2022), which lies outside the scope of D.
Consequently, the crucial theoretical notion prescribes that individuals’ general tendency to engage in utility maximization at others’ expense is essentially determined by D and not by whatever aspects a trait may encompass beyond D. Stated simply, to the extent that any aversive trait accounts for how much individuals weigh their own utility over others’, it does so because of D. Arguably, this is a bold theoretical stance and thereby a clear advantage, as it sets a high empirical hurdle for the theory to overcome (Meehl, 1967; Platt, 1964). Moreover, this claim is unique to the theory of D in that it is not shared by any other constructs previously suggested as representations of the commonalties of aversive traits—for example, low levels of Big Five Agreeableness (e.g., Vize et al., 2021) or HEXACO Honesty-Humility (Hodson et al., 2018).
Here, our overall goal is to test the deductively derived prediction that aversive traits do not account meaningfully for how individuals weigh their own versus others’ utility above and beyond D. To this end, one must rely on a to-be-predicted criterion that reflects, as directly as possible, this relative weighting of own versus others’ utility. Fortunately, this very relation is directly reflected in the assessment of social value orientations, or SVOs, representing the proactive aspects of social preferences (Fehr & Klaus, 1999; Messick & McClintock, 1968; Murphy & Ackermann, 2014; Van Lange, 1999; Van Lange et al., 1997). Specifically, to measure these preferences in a directly observable and consequential manner (so as to represent actual behavior; Baumeister et al., 2007; Murnighan & Wang, 2016; Thielmann et al., 2021), individuals are repeatedly asked to unilaterally distribute a valued resource (typically money) between themselves and another anonymous individual. The decisions are structurally designed to produce a single continuous measure representing the relative weights assigned to own versus others’ utilities (Balliet et al., 2009; Fiedler et al., 2013; Murphy et al., 2011). This relative weight will thus serve as the to-be-predicted criterion for the present study, operationalized by the well-established Social Value Orientation (SVO) Slider Measure (Murphy et al., 2011) that has previously been linked to D (Horsten et al., 2021; Moshagen, Zettler, & Hilbig, 2020).
Statement of Relevance
People differ in their tendency to engage in socially and ethically aversive behaviors such as aggression, cheating, crime, manipulation, violence, and many more. In part, these are thought to be driven by various aversive (“dark”) traits such as psychopathy, narcissism, greed, or spitefulness. However, despite some nuanced differences, all these traits can be understood as manifestations of a single, underlying disposition: the dark factor of personality (D). The present work tests this claim by showing that none of the 58 different traits and dispositional beliefs we considered meaningfully add to the explanation of the extent to which individuals place their own gains above those of other people. In conclusion, D represents the shared core across a plethora of aversive personality traits.
Importantly, the broader or more comprehensive the set of possible aversive traits considered, the higher the probability of falsifying the to-be-tested prediction that aversive traits account for own versus others’ utility entirely due to D. Thus, a strict test ought to encompass any potentially aversive trait, that is, any stable tendency that conceptually relates to how individuals weigh their own versus others’ utility. Therefore, positively connoted, prosocial traits—such as Honesty-Humility (Ashton et al., 2014) or compassion (Hwang et al., 2008)—are also relevant, given that their low pole may also imply utility maximization at others’ expense. Moreover, one may consider stable attitudes and beliefs, both prosocial and aversive in naming—for example, moral idealism (Forsyth, 1980) or social dominance orientation (SDO; Sidanius & Pratto, 1999). In fact, because any stable tendency conceptually overlapping with how individuals weigh their own versus others’ utility could have an aversive component, the above prediction is tested most strictly if practically all such tendencies are put to the test.
As detailed below, we considered 58 candidate traits (for simplicity were herein use the term for all contructs considered, including stable beliefs). Although perfect comprehensiveness necessarily remains an elusive goal, this is—to best of our knowledge—the most comprehensive set of potentially aversive traits ever studied in combination to scrutinize the prediction that aversive traits do not meaningfully predict how individuals weigh their own versus others’ utility beyond D.
Method
Data were collected in the context of the Prosocial Personality Project (PPP), a large-scale longitudinal project assessing various traits (and similar constructs; see below) conceptually related to prosocial versus aversive behavior. Traits were selected on the basis of (a) a recent meta-analysis on the link between 51 personality traits and prosocial behavior (Thielmann et al., 2020) and (b) a thorough literature search to identify further potentially relevant traits and constructs linked to prosociality, very broadly defined. This resulted in the inclusion of a total of 73 traits in the PPP, which were assessed at different measurement occasions and which provided the basis for the current investigation. Moreover, the PPP included the SVO Slider measure (described below) and thus a consequential measure of the relative weight individuals assign to own versus others’ utility. A detailed documentation of the PPP—including information on all constructs assessed per measurement occasion, documentation of publications based on the PPP, and information on sample compositions, data quality checks, and a priori defined exclusion criteria—is available on the Open Science Framework (OSF; https://osf.io/m2abp/). The data used here stem from seven different measurement occasions of the PPP (see Table 1). All measurement occasions involved voluntary participation based on fully informed consent. Studies followed protocols and procedures preapproved by the local institutional review board, and there was no deception of participants (Hilbig et al., 2022).
Candidate Traits Along With Their Definitions, Operationalizations, and Measurement Occasion, and Whether They Were Also Assigned to the Strict Set of Traits Involving Unequivocal Conceptual Overlap With the Criterion
Note: The definitions provided in the second column are verbatim citations from the respective references. The average temporal distance between measurement of the traits and assessment of social value orientation was 110 days for T1, 69 days for T2, 49 days for T3, 26 days for T4, 0 days for T5 (the SVO Slider was part of this measurement occasion), 61 days for follow-up 2020-05a, and 58 days for follow-up 2020-05b. ω = latent factor reliability; SDO = Social dominance orientation; SD3 = Short Dark Triad; NARQ = Narcissistic Admiration and Rivalry Questionnaire; SRP-III = Self-Report Psychopathy Scale–III.
The two items from the short scale were collected at T3, and the remaining items were collected at T4.
Data for the PPP were collected online via a professional panel provider in Germany, and this allowed us to recruit a large and heterogenous sample in terms of age, gender, profession, income, and educational background. The first measurement occasion (T1) was completed by 4,585 participants (2,356 female, 2223 male, 6 diverse) covering a broad age range from 18 to 78 years (M = 40.2 years, SD = 13.0). T1 participants were then reinvited to take part in the subsequent measurement occasions.
The SVO Slider
The SVO Slider (Murphy et al., 2011) is a well-established paradigm to assess social value orientations, providing a consequential measure of how individuals weigh their own relative to others’ utility. It was included at T5 of the PPP and completed by 2,707 participants (1,284 female; aged 18–96, M = 43.5 years, SD = 12.5). 1 In this paradigm, participants make six independent decisions, allocating outcomes (i.e., points that are worth money) to themselves and an unknown, randomly determined other individual. The six items (displayed in Fig. 1) are designed to allow for the exact identification of any combination of decision weights assigned to own versus others’ outcomes and, thus, utility. In particular, individuals consistently maximizing joint outcomes (prosocial type) must have assigned positive weight (1) to their own and others’ outcomes; individuals consistently maximizing own outcomes while disregarding others’ (individualistic type) must have assigned positive weight (1) to their own but zero weight (0) to others’ outcomes; and an individual consistently maximizing the difference between own and others’ outcomes (competitive type) must have assigned positive weight (1) to their own but negative weight (−1) to others’ outcomes (Murphy & Ackermann, 2014). All other combinations of weights are theoretically also possible—for example, zero weight (0) to own outcomes and positive weight to others’ outcomes (1), that is, pure altruism—but these make up fewer than 3% of all cases empirically (Murphy et al., 2011).
Items of the SVO Slider measure. Crossmarks indicate the choices of prototypical competitors—that is, assigning positive weight (1) to their own outcomes, but negative weight (−1) to others’ outcomes. SVO = social value orientation.
We explained to participants that the task would incur a bonus payment (€0.20 per 10 points ultimately earned). Participants were informed that half of them would be randomly assigned to each possible role (sender or receiver) and that exactly one of the six decisions was going to be drawn at random to determine their payment. Participants earned a bonus payment between €0.30 and €2.00 (M = €1.55, SD = €0.30) paid out separately and privately by the panel provider, thus preserving full anonymity among participants and between participants and researchers. Items were presented to participants one at a time in random order.
In line with common procedures and as recommended (Murphy et al., 2011), participants’ decisions were combined into a single continuous index of the relation between own versus others’ outcomes, the so-called SVO angle, by
where
D and Candidate Traits
At T1 of the PPP (and thus with the largest temporal distance from the SVO Slider, namely 110 days on average), D was measured via the German version (Bader et al., 2022) of the item set identified by Moshagen, Zettler, & Hilbig (2020), involving 70 items in total (35 reverse keyed). Example items included “I’ll say anything to get what I want” or “I cannot imagine how being mean to others could ever be exciting” (reverse keyed).
To arrive at a comprehensive and suitable set of candidate traits for the present investigation (i.e., to test for their ability to predict the relative weights individuals assign to their own versus others’ utility above and beyond D), it was necessary to identify and select relevant candidate traits from the PPP. To this end, all traits measured in the PPP by the end of 2020 were scrutinized in terms of their conceptual overlap with how individuals weigh their own versus others’ utility (as operationalized in the SVO Slider), with a view to both their definition and their item content. To avoid biases, we sought corresponding ratings from (a) two of the present authors who neither had access to the data from the PPP nor had been involved in the selection of traits for the PPP and (b) two entirely independent experts studying social value orientations. All raters independently judged each trait for its conceptual overlap with the criterion (i.e., the SVO angle), assigning it either (a) clear overlap, (b) potential overlap, or (c) clearly no overlap. On the basis of these ratings, we excluded all traits that were judged by at least three of the four raters to involve no overlap. This led to exclusion of 15 traits 2 and thus 58 were retained. Table 1 provides an overview of all 58 traits, along with definitions, questionnaires and example items, reliabilities (as estimated in the present data; see the statistical analysis section below), and measurement occasion (i.e., when each included in the PPP). However, because some of these traits received mixed ratings and because the inclusion of only marginally relevant traits might bias the overall picture (reducing the mean proportion of explained incremental variance of the traits beyond D), we additionally checked the results for a strict subset of traits rated as having unequivocal conceptual overlap with the criterion. For this strict set, we retained only those 37 traits that were judged to involve clear overlap by at least three raters (see the final column of Table 1).
Statistical Analysis
All analyses reported herein were based on structural equation modeling of raw scores using lavaan (Rosseel, 2012). First, for each of the 58 candidate traits, we estimated a baseline model, specifying D as a single latent factor, the candidate trait as a single latent factor (each indicated by their respective items), and the SVO angle as an observed variable, always using pairwise complete data. This model served to gauge the latent zero-order associations of D, the candidate trait, and the SVO angle as well as the reliability (ω) of D and the candidate trait. Model fits were acceptable with a median root-mean-square error of approximation of .06 and a median standardized root-mean-square residual of .05. For detailed results, see the OSF supplement (https://osf.io/jnmda).
To model D, we used only the short 16-item set (Moshagen, Zettler, & Hilbig, 2020) because this was comparable with the candidate traits in terms of number of items and reliability (ω = .86 for D and median ω = .84 for candidate traits; for reliability of each candidate trait, see Table 1). Thus, given that we accounted for measurement error through latent modeling and computed the incremental variance explained by each candidate trait above a measure of D that was similar in terms of reliability, we can rule out the possibility that any such differences biased the comparisons. Nonetheless, we also reran all analyses and double-checked results using the full 70-item set for D, 3 which led to results confirming all conclusions (despite generally smaller increments of variance explained by candidate traits, as must be expected).
To test our main hypothesis, three extensions of the baseline model involving latent multiple regression were estimated and compared: one predicting the SVO angle by D alone, a second predicting the SVO angle by the candidate trait alone, and a third predicting the SVO angle by D and the candidate trait in combination. From these models, the increment in the proportion of variance explained by adding the candidate trait (ΔR²) was computed and served as the key criterion of interest. Given that our median sample size (N = 2,270) already yielded very high statistical power (> .95) for even a miniscule and negligible effect (ΔR² = .005; Jobst et al., in press), we did not consider statistical significance. Instead, we set what J. Cohen (1988) defined as a small effect (f ² = .02; i.e., ΔR² = .018), “just barely escaping triviality” (p. 413), as the lower bound for a relevant effect of a candidate trait beyond D. The complete code for all analyses and supplementary results, including detailed scale descriptives for and observed correlations between all variables, are available on the OSF (https://osf.io/jnmda). The study was not preregistered.
Results
Participants’ decisions on the SVO Slider were largely typical (Murphy et al., 2011), resulting in a median SVO angle of 35° (M = 32°, SD = 11°) and a range from −16.26° (perfectly consistent competitiveness; own weight 1, others’ weight −1) to 61.39° (perfectly consistent prosociality; own weight 1, others’ weight 1). Neither age nor gender had any influence on the SVO angle (r = −.03 and Cohen’s d = 0.05, respectively). The median (latent) zero-order correlations were |r| = .51 (range: .06 ≤ |r| ≤ .83) between D and candidate traits, r ~ −.27 (median R² = .08) between D and the SVO angle, and |r| = .17 (range: .02 ≤ |r| ≤ .33) between candidate traits and the SVO angle. The median total multiple R² (i.e., total variance explained in the SVO angle by D and a candidate trait) was .08 (range: .07 ≤ R² ≤ .11).
Exact results for every single candidate trait and each model (comparison) are available in the OSF supplement (https://osf.io/jnmda). In summary, and as displayed in Figure 2, the incremental variance explained by candidate traits in the SVO angle above D was virtually zero (mean ΔR² = .003, median ΔR² = .001) and well below what was considered even a minimally relevant effect, that is, a small effect according to J. Cohen (1988). Essentially the same result was also found in the strict set of 37 traits involving unequivocal conceptual overlap with the criterion (mean ΔR² = .005, median ΔR² = .002), so we ruled out the possibility that consideration of too many irrelevant traits biased the results in favor of the hypothesis. Results were also independent of the reliability of candidate traits (r = −.02, p = .91).
Distribution of incremental proportions of variance explained (ΔR²) by the 58 candidate traits beyond D in the social value orientation (SVO) angle. The dashed line represents the median proportion of incremental variance explained; the dotted line represents a small effect size (ΔR² = .018) according to J. Cohen (1988) and the threshold for an effect to be considered relevant.
As can also be seen in Figure 2, exactly one single candidate trait explained incremental variance larger than what was considered a minimally relevant effect. Specifically, a competitive jungle world view accounted for ΔR² = .028 of incremental variance beyond D (ΔR² = .02 when modeling D on the basis of 70 rather than 16 items), which is approximately one fourth of the total variance explained in combination with D (R²mult = .11). Of note, competitive jungle world view was also particularly strongly associated with D (r = .75). The zero-order association between a competitive jungle world view and the SVO angle was marginally stronger (r = −.33) than the association between D and the SVO angle (r = −.27). In interpreting this difference, it must be stressed that competitive jungle world view was assessed at the same measurement occasion as the SVO slider, whereas D preceded the latter by around 110 days.
Additionally, and beyond our hypothesis, we also computed the extent to which D accounts for incremental variance in the SVO angle beyond each candidate trait. Although such an incremental effect is not strictly implied by the theory of D, it is informative of the extent to which D uniquely accounts for how individuals weigh their own versus others’ utility. Across candidate traits, D added up to ΔR² = .09 of incremental variance (mean and median ΔR² = .04) and above what was considered a minimally relevant effect in 46 out of 58 comparisons with candidate traits (including Big Five Agreeableness, ΔR² = .045). The remaining 12 candidate traits above which D did not account for sufficient incremental variance in the SVO angle (including HEXACO Honesty-Humility, ΔR² < .01) were particularly highly correlated with D (median |r| = .72) and did not correlate with the SVO angle more strongly than D (median |r| = .26 vs. median |r| = .27). Thus, in summary, D outperformed 80% of candidate traits in accounting for the SVO angle and fared on par with the rest.
Discussion
The common core of aversive traits, D, is conceptualized as the (fluid) underlying disposition from which aversive traits arise as specific, flavored manifestations (Moshagen et al., 2018). The crucial implication of this notion is that everything aversive in a trait is due to D and, in turn, that aversive traits do not account meaningfully for how individuals weigh their own versus others’ utility above and beyond D. To critically scrutinize this prediction, we tested whether and to what extent a total of 58 candidate traits and related constructs (such as stable beliefs) meaningfully predict how individuals weigh their own versus others’ utility—as operationalized via a consequential measure of proactive social preferences, the SVO angle (Murphy et al., 2011)—beyond D.
Results revealed that around 10% of variance in the SVO angle can be accounted for by self-reported traits, which is at the upper end of meta-analytical effect-size estimates on the link between personality and prosocial behavior (Thielmann et al., 2020). Crucially, on average, candidate traits predicted practically no incremental variance beyond D, with 57 out of the 58 traits yielding a proportion of incremental explained variance well below a small effect size. Thus, the main prediction held not only for traits that have explicitly been called aversive (or “dark”; e.g., the Dark Tetrad: Machiavellianism, narcissism, psychopathy, and sadism), but also for broad personality dimensions (e.g., Honesty-Humility or Big Five Agreeableness), stable beliefs (e.g., belief in reciprocity, moral idealism vs. relativism), and diverse but more narrow tendencies, both positively (e.g., altruism, empathy, collectivism) or negatively (e.g., exploitativeness, competitiveness, aggressiveness, envy) related to prosociality. Thus, conclusions were consistent independent of construct breadth (broad vs. narrow; e.g., Honesty-Humility vs. humility), specific nature (behavioral tendencies vs. beliefs), or direction (positive vs. negative). Finally, results were no less favorable when D was measured more broadly (70 rather than 16 items) and was practically the same in a strict set of 37 candidate traits involving unequivocal conceptual overlap with the criterion as rated by at least three of four independent experts.
The only candidate construct predicting a meaningful portion of incremental variance in SVO beyond D was a competitive jungle world view (Sibley & Duckitt, 2009). This particular belief may comprise aversive aspects beyond D, albeit to a limited extent, and therefore not merely represent a manifestation of D. Of note, competitive jungle world view is among the constructs most highly associated with D, and both were similarly related to the SVO angle, suggesting a suppression effect (the difference in their respective zero-order effects implies ΔR² = .004 only if D and competitive jungle world view were independent). Thus, future research will be required to determine whether and to what extent competitive jungle world view comprises unique aversive variance beyond D in SVO and, more importantly, in a broader array of aversive behaviors.
Indeed, particular SVO weights are essentially an abstraction of many, but certainly not all, aversive behaviors; these weights represent the essential ingredients of the proactive side in many situations of interdependence (Thielmann et al., 2021), but they do not immediately translate to situations that do not involve one specific other individual, but rather shared norms of groups or societies (Bazerman & Gino, 2012). Moreover, distributing valued outcomes between oneself and an anonymous other in an entirely calm and unthreatening setting will not fully represent situations involving severe threat, physical violence, or the like. Finally, SVO misses out on reactive aspects of social preferences (i.e., reciprocity); traits such as a negative reciprocity norm endorsement (Eisenberger et al., 2004) or vengefulness (Stuckless & Goranson, 1992) that were excluded here because of their lack of conceptual relevance for SVO may account for aversive behavior beyond D in corresponding criteria.
Of note, the main assumption here is not that SVO covers the plethora of imaginable situations that may produce aversive behavior, let alone that no candidate traits will ever explain incremental variance in any of these. On the contrary, some of the traits we considered involve certain aspects irrelevant for SVO (or social preferences more broadly) and unrelated to D that are nonetheless useful to account for other specific aversive behaviors. For example, although psychopathy adds nothing to the explanation of SVO beyond D, as shown herein, it is known to encompass disinhibition (Patrick et al., 2009). This aspect is largely irrelevant for SVO and conceptually beyond the scope of D, but it will nonetheless be relevant for certain behaviors that are both aversive and impulsive (e.g., reckless driving; Bader et al., 2022). Thus, the generalization that D will make obsolete all specific aversive traits in accounting for any aversive behavior would be premature and a gross oversimplification.
Moreover, some of the candidate traits may interact with features of the situation and they may, under certain conditions, do so beyond or independent of D. For example, in social dilemmas such as the one-shot prisoner’s dilemma (Dawes & Messick, 2000; Rapoport & Chammah, 1965), there is both a possibility for exploitation (temptation) and dependence on others under uncertainty (beliefs in others; i.e., trust vs. fear) involved, both of which determine which traits can be expressed in behavior (Thielmann et al., 2020, 2021). A trait such as HEXACO Honesty-Humility is primarily afforded by a possibility for exploitation and therefore involves an interaction in accounting for social-dilemma behavior, because temptation is necessary for Honesty-Humility to be predictive, whereas fear is neither necessary nor sufficient (Hilbig et al., 2018). Because D, by contrast, also involves justifying beliefs, including those related to distrust and thus fear of exploitation, and indeed substantially more so than Honesty-Humility (Hilbig et al., in press), one would expect no such interaction for D. Thus, the interaction found for Honesty-Humility ought to hold independent of D, which is an important implication given the present finding that Honesty-Humility and D are highly comparable when it comes to explaining SVO. On a broader theoretical level, the argument is that even traits that can be considered (flavored) manifestations of D are neither necessarily fully subsumed by, nor functionally equivalent to, D.
In conclusion, our findings are consistent with the prediction derived from the theory of D that specific traits do not meaningfully subsume “the tendency to maximize one’s individual utility—disregarding, accepting, or malevolently provoking disutility for others” (Moshagen et al., 2018, p. 657) beyond D, at least in a proactive situation as represented by the SVO Slider. In other words, the extent to which any trait accounts for how individuals weigh their own versus others’ utility is due to D, confirming that D captures the proactive social preferences reflected in—and thus much of the socially aversive essence of—such traits. Consequently, D serves to theoretically integrate the vast literature on aversive personality and to unite the diverse and often isolated approaches—from basic traits, personality psychopathology, so-called dark traits, prosocial traits, and diverse beliefs—currently relied on to explain socially and ethically aversive affect, thought, and behavior.
Footnotes
Acknowledgements
We are greatly indebted to Dorothee Mischkowski and Jan Luca Pletzer, who provided us with independent ratings for the selection of traits.
Transparency
Action Editor: Paul Jose
Editor: Patricia J. Bauer
Author Contributions
