Sage Journals: Discover world-class research

Abstract

Sadism represents a predisposition towards enjoying the suffering that an individual causes to others. This conceptualization of Sadism has clear overlap with schadenfreude—finding pleasure in others’ suffering. However, the relationship between trait Sadism and trait schadenfreude has gone understudied. Using exploratory and confirmatory latent construct modeling with a cross-sectional and diverse sample of 314 undergraduate participants, we found that the bi-factor model of Sadism and schadenfreude that best fit the data articulated Sadism as a sub-facet of schadenfreude. However, an EFA did not find the same relationship, and instead suggested a four-factor solution that split harm-enjoying traits into higher order Sadism and schadenfreude domains that each had lower-order facets—including a malicious component of schadenfreude that most closely abutted Sadism. Future research should seek to identify the mechanisms that translate a passive, schadenfreudic disposition into actual acts of Sadistic aggression.

Keywords

antisocial behavior personality scales factor analysis sadism schadenfreude

“I have never killed anyone, but I have read some obituary notices with great satisfaction”- Clarence Darrow

Sadism reflects the tendency for some people to enjoy the harm that they inflict upon others (Foulkes, 2019). Research into this topic has resulted in significant progress, but also some inconsistency in the conceptualization of which traits are and are not central to Sadism. The present study combined a novel factor analytic approach with bi-factor models to bring clarity to this literature by empirically examining (I) whether dominance-seeking and callous-unempathic responding belong in the core Sadism construct, (II) whether vicarious Sadism is redundant with schadenfreude, and (III) whether Sadism is a lower-order facet of schadenfreude. In addition, the present study used a hierarchical factor analysis in order to examine the relationships between trait Sadism and schadenfreude.

Sadistic personality

Sadism, first conceptualized as a form of psychopathology and criminal behavior outlined in the Diagnostic and Statistical Manual of Mental Disorders revised third edition (American Psychiatric Association, 1987), was once considered a purely sexual phenomena (i.e., sexual Sadism), and indeed its very name comes from Marquis de Sade, who wrote sexually explicit fiction depicting extreme violence and suffering. Modern conceptualizations of Sadism, however, have broadened the definition to “the deliberate infliction of pain for the sake of enjoyment” (Nell, 2006, p. 227). Sadism was previously considered a psychiatric disorder, and Sadistic personality disorder was included in the revised third edition of the diagnostic and statistical manual of mental disorders (American Psychiatric Association, 1987). However, more recently, researchers have conceptualized Sadism as a broader personality trait, and have demonstrated the existence of sub-clinical, so-called “everyday” Sadism that exists within the broader population (e.g., Buckels et al., 2013; Paulhus & Dutton, 2016).

Sadism is of key interest to researchers, clinicians, policy-makers, and the American legal system, as it shows some of the strongest relationships among personality traits with acts of aggression (Chester et al., 2019). The relationship between Sadism and aggression depends upon the perceived suffering of the target—greater perceived harm leads to greater pleasure derived from the act (Chester et al., 2019).

At its core, Sadism represents a predisposition towards enjoying the infliction of harm onto others. However, other conceptual approaches to Sadism expand this construct to include other psychological traits and tendencies. Specifically, the Assessment of Sadistic Personality (ASP; Plouffe et al., 2017) includes subjugation (i.e., a disposition towards seeking power over others) and callousness (called “unempathicness” in the scale; i.e., an unemotional disposition towards the feelings of others) facets of Sadism within its model of Sadism. Sadistic individuals may indeed also be predisposed towards dominating others and having a callous response to others’ suffering. However, it remains unseen whether these constructs are inherent aspects of the Sadism construct or if they are correlated, but ultimately distinct traits.

Other conceptualizations of Sadism (e.g., those manifested in the Comprehensive Assessment of Sadistic Tendencies; CAST; Buckels & Paulhus, 2013) include separate constructs describing the direct inclination to cause harm for enjoyment (i.e., direct sadism) and the passive enjoyment of the suffering of others while not inflicting the harm directly (i.e., vicarious sadism). In summary, we conceptualize Sadism as a core tendency to enjoy the harm that people inflict on others with several peripheral elements that are under debate. This conceptual core is reflected in the pleasure-seeking and direct Sadism components of the ASP and CAST models, respectively. Peripheral aspects that are under debate are the subjugation and unempathicness aspects of the ASP model and the vicarious Sadism factor of the CAST model. We deem these latter factors to be peripheral because they are specific to their own conceptual models and do not re-appear across different measures and models. They are also considered peripheral because they are core aspects of other constructs (e.g., unempathicness is at the conceptual core of psychopathy). It is critical to empirically determine whether these peripheral elements deserve to be retained in the Sadism construct. Additionally, we must also disambiguate Sadism (its core and peripheral elements) from another construct that it may also be closely related to—if not directly redundant with—the construct of schadenfreude.

Sadism and schadenfreude

Schadenfreude reflects the pleasure people experience from the suffering of others (van Dijk et al., 2011) and has been compared to Sadism in the past. Trait Sadism has been previously linked to greater momentary experiences of schadenfreude (Lee, 2019; Schumpe & Lafrenière, 2016). However, the relationship between schadenfreude and Sadism as personality traits remains understudied. At a prima facie level, there is clear overlap between Sadism and schadenfreude—both involving joy derived from others’ suffering. Indeed, there may be grounds to claim that these may have a nested relationship rather than a mere correlational one. Sadism, as a trait, may be nested under a broader disposition towards schadenfreude.

Schadenfreude has also been conceptualized as a context-specific response to a situation. Individuals are likely to experience schadenfreude when they have some instrumental or emotional gain from the victim’s suffering (e.g., positive self-esteem from downward comparison), when the victim is an outgroup member, when the victim deserves their harm, or when the individual envies the victim (e.g., Smith et al., 1996; Smith et al., 2009; Wang et al., 2019). These conditions are not necessarily unique to schadenfreude, as individuals high in Sadism are still likely to receive pleasure from these same scenarios. However, most definitions of Sadism emphasize that the pleasure that a Sadistic individual gains from harm is an end unto itself, and does not rely on situational context (e.g., Chester et al., 2019). This non-contextual “harm for the sake of harm” may differentiate schadenfreude from Sadistic pleasure and propel Sadistic individuals to directly cause the harm that they enjoy. Furthermore, the influence of these situational factors does not preclude stable dispositions towards experiencing schadenfreude. Schadenfreude, like all affective states, occurs both momentarily and dispositionally. Indeed, there is clear evidence for trait schadenfreude. Crysel and Webster (2018) developed and validated a measure of stable dispositions towards experiencing schadenfreude. This scale showed adequate test-retest reliability and stability over time, clearly indicating that individuals have a stable disposition towards or away from experiencing schadenfreude. This measure of trait schadenfreude was positively correlated with Machiavellianism, psychopathy, Narcissism, physical aggression, anger, and hostility. However, the relationship between trait schadenfreude and trait Sadism remains understudied and unclear.

The key distinction between schadenfreude and Sadism may lie in the tendency to passively experience the suffering of another versus the tendency to be actively involved in causing that suffering. Sadism inherently entails the active perpetration of harm (Chester et al., 2019), whereas schadenfreude does not. Whereas individuals high in schadenfreude may enjoy passively watching someone slip on a banana peel, Sadistic individuals may only enjoy this person’s suffering if they were the one who actively placed the banana peel in the path of their hapless victim. But this distinction does not entail that schadenfreude and Sadism are inherently alienated from one another. Instead, they may still share an intimate conceptual connection. It may be that Sadism is best defined as an active form of schadenfreude.

Present study

The first aim of the present study was to empirically examine the conceptual core of Sadism.¹ We examined the most commonly-included subfacets of Sadism that are present in the literature: pleasure-seeking, subjugation, unempathicness, and physical Sadism, verbal Sadism, and vicarious Sadism. It is important to investigate these different conceptualizations of Sadism and their proposed lower-order facets in order to more broadly understand contributors to antisocial behavior. We aimed to investigate whether the specific facets of subjugation, unempathicness, and vicarious Sadism should remain under the broader umbrella of Sadism, or are better conceptualized as related, but separate constructs. More specifically, we used a novel latent construct modeling technique, which we call “Dance Partner Modeling” to examine Sadism’s relation to its more peripheral facets: dominance, callousness, and schadenfreude. Dance Partner Modeling uses confirmatory factor analysis (CFA) by comparing the model fit between two different structural permutations of the same items. This approach seeks to identify whether a given construct is better paired with one conceptual ‘dance partner’ or another. To do so, the construct takes turns paired with each ‘dance partner’ and then we compare their performance to determine who the construct best belongs with. In CFA terms, Dance Partner Modeling compares the fit of a pair of models that each specify the same two latent and correlated factors but differ in the observed indicators that define those two factors. More specifically, in one model the target construct is an observed indicator that loads onto the first latent factor and in the other model the target construct is an observed indicator that loads onto the second latent factor. For example, Dance Partner Modeling would first pair the subjugation subscale of the ASP with a latent ‘Dominance’ factor indicated by other measures of dominance-seeking. Then, it would pair this subscale with a latent ‘Sadism’ factor indicated by other Sadism subscales and compare the fits of the two models. If we observed better fit when the subjugation subscale was paired with ‘Dominance’ than ‘Sadism’, then we could argue that it is a better facet of the former construct and should be excluded from the latter construct. We expected that the models that would best fit the data would be those that modeled the peripheral elements of Sadism (i.e., subjugation, unempathicness, and vicarious Sadism) as indicators of non-Sadism constructs (i.e., dominance, callousness, and schadenfreude) compared to those that modeled these peripheral elements as indicators of Sadism. We made no predictions about which of these three different sets of models would fit the best, our focus was on the relative fit of the two models within each instance of Dance Partner Modeling, not between them.

Our preregistered hypotheses were as follows:

H1: Including Sadistic subjugation with dominance will show better model fit compared to including Sadistic subjugation items with Sadism.

H2: Including Sadistic callousness with callousness will show better model fit compared to including Sadistic callousness with Sadism.

H3: Including vicarious Sadism items with schadenfreude will show better model fit compared to including vicarious Sadism items with Sadism items.

The second aim of the present study was to clarify the relationship between Sadism and schadenfreude, specifically investigating if these constructs were empirically redundant, or if they might be nested within one another. We predicted that:

H4: The bi-factor model including Sadism and schadenfreude as a general factor and Sadism as a lower-order facet factor should show the best fit, suggesting that trait Sadism can be considered a lower-order facet of trait schadenfreude.

Finally, we aimed to clarify the nomological networks of Sadism and schadenfreude, we also conducted and compared their bivariate correlations with several theoretically-relevant personality constructs, including the Big 5 personality dimensions, aggression, behavioral approach and inhibition, and a tendency towards holding contradictory responses with others’ emotions (i.e., affective dissonance). We did not have specific predictions concerning these correlations. All methods used in this study, including scales, consent, and debriefing materials were reviewed and accepted by Virginia Commonwealth University’s institutional review board.

Method

Statistical power statement

Sample size was derived via a power analysis using the semPower package in R version 1.2.0 (Moshagen & Erdfelder, 2016) for a S-1 bi-factor model (i.e., model 6) with an expected RMSEA of .07, α = .05, and 80% power. Due to the large number of items in each scale of the intended model, we preregistered using the 16 component subscales as indicators instead of raw items. This power analysis suggested a minimum of 219 participants. In order to account for a possibly inflated effect size in our power analysis and to ensure sufficient power, we over-recruited to 322 participants. This decision was made prior to accessing any data. Eight participants did not answer a single question after consenting to the study, and were thus removed from the dataset, leaving a final sample of 314.

Participants

Participants consisted of 314 undergraduates from a large southeastern university in the United States. Demographic information for the sample can be found in Table 1.

Table 1.

Sample demographic information.

Demographic information	M	SD	Range
Age	19.80	2.86	18–42
Race	Percentage
White	38.89%
Asian	21.91%
Black/African American	18.32%
Non-white Hispanic/Latinx	6.17%
Middle-Eastern/Arabic	4.94%
Native American	0.93%
Pacific Islander/native Hawaiian	0.62%
“Other racial identity”	5.56%
Did not disclose	2.66%
Gender identity	Percentage
Cisgender women	65.74%
Cisgender men	22.53%
Non-binary/other gender identity	3.09%
Prefer to self-describe	2.16%
Transgender women	0.62%
Transgender men	0.31%
Did not disclose	5.55%

Materials

Sadism scales

Assessment of sadistic personality (ASP; Plouffe et al., 2017)

The ASP is a nine-item scale assessing trait Sadism consisting of three subscales: subjugation (three items; e.g., “I never get tired of pushing people around”), pleasure-seeking (four items; e.g., “When I mock someone, it is funny to see them get upset), and unempathic (two items; e.g., “I think about hurting people who irritate me”). Participants were asked to rate the degree to which they agree or disagree with each statement on a scale from 1 (Strongly disagree) to 5 (Strongly agree).

Short sadistic impulse scale (SSIS; O’Meara et al., 2011)

The SSIS is a 10-item scale of dispositional Sadism (e.g., “I would enjoy hurting someone physically, sexually, or emotionally”). Participants were asked to rate the degree to which they agree or disagree with each statement on a scale from 1 (Strongly disagree) to 5 (Strongly agree).

Comprehensive assessment of sadistic tendencies (CAST; Buckles & Paulhus, 2014)

The CAST is a measure of trait Sadism, and comprises three subscales, two measuring direct Sadism: verbal (six items; e.g., “I have purposely tricked someone and laughed when they looked foolish”) and physical (five items; e.g., “I enjoy physically hurting people”) and one subscale measuring vicarious Sadism (seven items; e.g., “In video games, I like the realistic blood spurts”). Participants were asked to rate the degree to which they agree with each statement on a scale from 1 (Strongly disagree) to 7 (Strongly agree).

Secondary construct scales

Computerized adaptive test of personality disorder (CAT-PD; Simms et al., 2014)

The CAT-PD is a measure of personality disorders based on the Diagnostic and Statistical Manual of Mental Disorders fifth edition (DSM-V). It contains 33 subscales, but only the domineering subscale (six items; e.g., “I like having authority over others.”) and callousness (seven items; e.g., “I do not care how my actions affect others.”) were used in the present study. Participants were asked to respond to each item on a scale from 1 (Very untrue of me) to 5 (Very true of me).

Trait schadenfreude scale (Crysel & Webster, 2018)

The Crysel and Webster (2018) Schadenfreude Scale measures trait dispositions towards experiencing pleasure from viewing others’ pain. This scale contains two theoretically-derived subscales of benign schadenfreude, in which the harm done to the target was minimal (six items; e.g., “I have laughed at someone who has fallen before helping them up”), and malicious schadenfreude, in which serious harm to the target occurred (six items; e.g., “I like to see someone successful get fired”). Participants were asked to rate the degree to which they agree or disagree with each statement on a scale from 1 (Strongly disagree) to 9 (Strongly agree).

Discriminant validity scales

Affective and cognitive measure of empathy (ACME; Vachon & Lynam, 2016)

The ACME is a 36-item self-report measure of empathy that includes 3 subscales, however, only the 12-item affective dissonance (e.g., “I get a kick out of making other people feel stupid”) was used in the present study. Participants were asked to rate the degree to which they agree or disagree with each statement on a scale from 1 (Disagree strongly) to 5 (Agree strongly).

Behavioral inhibition system/behavioral activation system scale (BIS/BAS; Carver & White, 1994)

The BIS/BAS scale was designed to measure individual sensitivities to two affective/behavioral motivation systems, the behavioral inhibition system (BIS; i.e., regulation of aversive motivations) and the behavioral activation systems (BAS; i.e., regulation of appetitive motivations). The BIS/BAS consists of four subscales: BAS drive (four items; e.g., “If I see a chance to get something I want I move on it right away”), BAS fun-seeking (four items; e.g., “I will often do things for no other reason than that they might be fun”), BAS reward responsiveness (five items; e.g., “When I’m doing well at something I love to keep at it”), and BIS (seven items; e.g., “I feel worried when I think I have done poorly at something important”). Participants were asked to rate the degree to which they agree or disagree with each statement on a scale from 1 (Very true of me) to 4 (Very false of me).

Buss-Perry aggression questionnaire (BPAQ; Buss & Perry, 1992)

The BPAQ is one of the most widely used trait aggression questionnaires among non-clinical populations. It comprises four subscales: physical aggression (nine items; e.g., “once in a while I can’t control the urge to strike another person”), verbal aggression (five items; e.g., “I can’t help getting into arguments when people disagree with me”), anger (seven items; e.g., “I have trouble controlling my temper”), and hostility (eight items; “I know that ‘friends’ talk about me behind my back”). Participants were asked to respond to each item from 1 (Extremely uncharacteristic of me) to 5 (Extremely characteristic of me).

International personality item pool NEO-60 (IPIP-NEO-60; Goldberg, 1999; Goldberg et al., 2006; Maples-Keller et al., 2019)

The IPIP NEO-60 is a 60-item scale measuring the Big Five model of personality traits: agreeableness (e.g., “I sympathize with the homeless”), conscientiousness (e.g., “I like order”), extraversion (e.g., “I love large parties”), neuroticism (e.g., “I get irritated easily.”), and openness to experience (e.g., “I prefer variety to routine”). Participants were asked to rate the degree to which they agree or disagree with each statement on a scale from 1 (Strongly disagree) to 5 (Strongly agree).

Procedure

Participants were recruited through a university participant pool system. After signing up for the study through the system, they were directed to an online Qualtrics survey. After consenting, participants were asked to complete the study questionnaires in a random order, randomized by Qualtrics, in order to reduce possible order effects. They then read a full debriefing form through the online survey describing the purpose of the study, the initial withholding of hypotheses, and were referred to the first and second authors if they had any questions or concerns. Participants were compensated with course credit.

Data analytic strategy

Data preparation

In accordance with our preregistration, missing data were replaced with multiple imputation predictive mean metric using the mice package version 45i03 (Van Buuren & Groothuis-Oudshoorn, 2011) in R version 4.2.2 (R Core Team, 2023). We then created indices for each applicable Sadism and schadenfreude subscale (e.g., benign schadenfreude, physical Sadism) for use as indicators within our latent models by converting all individual item scores into Z-scores and then averaging appropriate items together. All subsequent analyses were performed with R version 4.2.2 (R Core Team, 2023), including the lavaan package version 0.6–11 (Rosseel, 2012).

Dance partner modeling

Several facets of Sadism, which conceptually extend past the core feature of Sadistic pleasure, have been put forth (i.e., subjugation or dominance, unempathicness or callousness, and vicarious Sadism). However, these traits may not necessarily be intrinsic parts of Sadism. We applied a novel factor analytic approach developed for this study to empirically estimate whether these peripheral traits truly belong within the Sadism construct or not—a technique we term Dance Partner Modeling. Dance Partner Modeling is a form of latent variable analysis that relies upon the same model fit comparison techniques that most CFAs rely upon. By comparing model fit between two competing models, we can examine if (for example) subjugation “dances better” (i.e., has better model fit) when it is partnered with Sadism or when it is partnered with items from a similar construct (i.e., dominance). If model fit is better when, for example, Sadistic subjugation items are loaded onto a dominance factor rather than a Sadism factor, it implies that subjugation as measured by the ASP is not an inherent part of Sadism, and should thus be excluded from the Sadism construct and measures thereof. Its novelty is not derived from the underlying statistics, but from the way we systematically compare how a given construct fits as an indicator of two different latent factors. In sum, the novelty is broadly conceptual and configural and not statistical.

We conducted three sets of latent variable analyses (please see Figure 1 for a conceptual example diagram; Models 1–3). Within each of the three sets were two models that modeled the same two latent factors. The first latent factor was labeled Sadism and was indicated by all available Sadism subscales. The second latent factor changed for each set of models, reflected the given construct of interest (e.g., dominance), and was indicated by a measure of that construct of interest (e.g., the Dominance subscale of the CAT-PD). The first model within each set modeled the construct of interest’s (e.g., dominance) corresponding Sadism subscale (e.g., ASP subjugation) as an indicator for the Sadism factor. The second model within each set modeled the construct of interest’s (e.g., dominance) corresponding Sadism subscale (e.g., ASP subjugation) as an indicator for the factor that reflected the construct-of-interest (e.g., dominance). These models used maximum likelihood estimation and set the loading of the first indicator for each factor to one, in order to set the scale of each latent factor. Error terms were left uncorrelated and latent factors were allowed to correlate. In accordance with our preregistration plan, we inferred that a model had superior fit to another model if the χ² difference test was statistically significant and the comparative fit index (CFI) increased by at least .02.

Figure 1.

Example conceptual diagram of dance partner modeling technique. Note. CAST: comprehensive assessment of sadistic tendencies; SSIS: short sadistic impulse scale; ASP: assessment of sadistic personalities; CAT-PD: computerized adaptive test of personality disorder; Subj: subjugation; Pleasure: pleasure-seeking; Phys: physical; Verb: verbal; Vica: vicarious.

For the Dance Partner Models, we hypothesized that model fit for the CFA in which ASP Subjugation is included with Dominance would be better than when it is included with Sadism. Similarly, we hypothesized that model fit for the CFA in which ASP Unempathicness is included with Callousness would be better than when it is included with Sadism, and when Vicarious Sadism is included with schadenfreude than when Vicarious Sadism is included with Sadism.

Bi-factor models

Bi-factor models are useful for testing hypotheses that one construct is subsumed under another. By separating the variance into a general effect (G) and a number of lower-order specific effects, we can understand nested relationships of constructs. However, commonly used symmetrical bi-factor models may be biased towards the bi-factor model, favoring it over competing models (Bonifay et al., 2017). Other studies have shown that symmetrical bi-factor models often produce impossible or unusual results such as small or negative factor variances (Heinrich et al., 2020). Instead, S-1 bi-factor models, which sets one of the specific factors as a reference factor compared to the other specific factors, have been shown to reduce this bias and produce more easily interpretable results (e.g., Heinrich et al., 2020). We used S-1 bi-factor modeling to examine if Sadism is a facet of schadenfreude. In order to do this, we constructed four different models (Models 4–7; Figure 2). Models 4 and 5 were set as baseline models, showing Sadism and schadenfreude as orthogonal and correlated factors, respectively. We then test two empirical S-1 bi-factor models. Model 6 tested our hypothesis that Sadism can be understood as a facet of schadenfreude, including all facets of schadenfreude and Sadism in the overall G factor, and Sadism alone in the specific factor. Finally, Model 7 tested an alternative hypothesis that schadenfreude is a facet of Sadism, including all facets of schadenfreude and Sadism in the overall G factor and schadenfreude alone in the specific factor. In accordance with our preregistration plan, we inferred that a model had superior fit to another model if the χ² difference test was statistically significant and the comparative fit index (CFI) was larger by at least .02

Figure 2.

Conceptual diagrams of bi-factor models used for models 4, 5, 6, and 7. Note. CAST: comprehensive assessment of sadistic tendencies; SSIS: short sadistic impulse scale; ASP: assessment of sadistic personalities; CAT-PD: computerized adaptive test of personality disorder; Subj: subjugation; Pleasure: pleasure-seeking; Phys: physical; Verb: verbal; Vica: vicarious.

Discriminant validity

Finally, we conducted bivariate correlations between Sadism and schadenfreude and several antagonistic traits. In order to compare Sadism and schadenfreude, we created a composite Sadism index by standardizing scores from the CAST, SSIS, and ASP into z-scores, then averaging those scores. We further used the overall Crysel and Webster (2018) Schadenfreude scale, rather than the benign and malicious subscales separately. We used Fisher’s r-to-Z transformation to then test the difference in strength between these correlation coefficients in order to get a better sense of discriminant validity.

Results

Descriptive statistics and reliability for model indicators can be found in Table 2 and model fit parameters can be found in Table 3. Although our preregistration specified that we would use the component subscales of each measure as our model parameters, we were unable to establish adequate fit for any model using this subscale-level approach (RMSEAs > .10; CFI <.07). Thus, we deviated from our preregistered plan and used individual items instead of subscales as indicators for each latent factor. Full factor loadings and residual variances for each item from each model can be found in Supplemental Materials (S1).

Table 2.

Descriptive statistics for model parameter scales.

Variable	Min	Max	Mean	Std. dev	Skew	Kurtosis	α
CAST physical	1.00	5.40	1.62	0.87	1.77	2.83	0.74
CAST verbal	1.00	6.17	2.43	1.20	0.63	−0.47	0.76
CAST vicarious	1.00	6.71	2.50	0.98	1.05	1.04	0.73
SSIS	1.00	3.67	1.40	1.55	1.80	4.09	0.83
ASP pleasure-seeking	1.00	4.25	1.59	0.71	1.26	0.90	0.75
ASP unempathic	1.00	5.00	2.18	1.07	0.50	−0.67	0.36
ASP subjugation	2.67	5.67	3.02	0.61	2.18	4.55	0.73
Benign schadenfreude	1.00	9.00	3.76	1.87	0.44	−0.39	0.81
Malicious schadenfreude	1.00	6.50	1.72	1.01	1.79	3.13	0.81
CAT-PD dominance	1.00	4.67	1.83	0.76	0.84	0.21	0.85
CAT-PD callousness	1.00	3.86	1.95	0.49	1.10	0.96	0.66

Note. CAST: comprehensive assessment of sadistic tendencies; SSIS: short sadistic impulse scale; ASP: assessment of sadistic personalities; CAT-PD: computerized adaptive test of personality disorder.

Table 3.

Summary of model fit parameters for dance partner models and bi-factor models.

Models	χ² (df)	p	RMSEA (90%CI)	CFI	Latent inter-factor correlations
Dance partner models
Model 1
Model 1a	2471.63 (818)	<.001	.08 (.08, .08)	.72	.59
Model 1b	2696.17 (818)	<.001	.09 (.08, .09)	.68	.75
Model 2
Model 2a	2137.66 (739)	<.001	.08 (.07, .08)	.72	−.54
Model 2b	2230.10 (739)	<.001	.08 (.08, .08)	.70	−.60
Model 3
Model 3a	2847.13 (944)	<.001	.08 (.08, .08)	.69	.78
Model 3b	2857.42 (944)	<.001	.08 (.08, .08)	.69	.81
Bi-factor models
Model 4	3059.01 (945)	<.001	.08 (.08, .09)	.65	.00
Model 5	2847.13 (944)	<.001	.08 (.08, .08)	.69	.78
Model 6	2702.96 (912)	<.001	.08 (.08, .08)	.71	.00
Model 7	2768.25 (933)	<.001	.08 (.08, .08)	.70	.00

Dance partner modeling

Model 1: Dominance and sadistic subjugation

Our first model examined whether model fit was better when ASP subjugation was included with Sadism (Model 1a) measures or with another dominance measure (Model 1b). Both Model 1a and Model 1b showed adequate fit by RMSEA but not by CFI. Due to the same degrees of freedom, we were unable to interpret the χ² difference test, however, the difference in CFI exceeded our established threshold, with greater CFI in Model 1a than Model 1b. Thus, model fit was better when Sadistic subjugation was included with Sadism than when paired with dominance.

Model 2: Callousness and sadistic unempathicness

Our second model compared model fit when ASP unempathicness was included with other Sadism items (Model 2a) and when it is included with other callousness items (Model 2b). Model 2a and Model 2b had adequate fit by RMSEA but not by CFI. Due to the same degrees of freedom in each model, we were unable to interpret the χ² difference test. However, the difference in CFI exceeded our established threshold, with greater CFI in Model 2a than Model 2b. This suggests that unempathicness is better when paired with Sadism than when paired with callousness.

Model 3: Schadenfreude and vicarious sadism

Our third model compared model fit when CAST vicarious Sadism was included with other Sadism items (Model 3a) and when it is included with trait schadenfreude (Model 3b). Model 3a and Model 3b both had adequate fit by RMSEA but not by CFI. Due to the degrees of freedom being equal in each model, we were unable to calculate a chi-square difference test. In addition, the difference in CFI between models did not exceed our threshold, suggesting that vicarious Sadism is just as good when paired with Sadism as it is when paired with schadenfreude.

Bi-factor models

The relationship between trait Sadism and schadenfreude has gone understudied, and may be deeper than previously suggested. It may be that Sadism and schadenfreude are not merely correlated, but represent a nested hierarchical structure. That is to say, trait Sadism is a facet under a broader umbrella of schadenfreude that captures not only a tendency towards enjoying the pain of others, but an inclination to directly cause it. Our bi-factor models investigated this.

Our first baseline model (Model 4)—setting Sadism and schadenfreude as orthogonal—had adequate fit by RMSEA but not by CFI. Our second baseline model (Model 5)—setting Sadism and schadenfreude as correlated but separate constructs—showed adequate fit by RMSEA but not by CFI. Model 5 showed improved fit over Model 4, χ² difference (1) = 211.89, p < .001. This was expected, as other research has shown that individuals high in trait Sadism are more likely to experience schadenfreude (e.g., Schumpe & Lafrenière, 2016). Model 6, setting Sadism as a facet of schadenfreude, showed adequate fit by RMSEA but not by CFI, as did Model 7, setting schadenfreude as a facet of Sadism. Most important for our hypothesis, Model 6 had significantly better fit compared to Model 4, χ² difference (33) = 356.06, p < .001, Model 5, χ² difference (32) = 144.17, p < .001, and Model 7, χ² difference (21) = 65.29, p < .001. Model 7 showed significantly better fit compared to Model 4, χ² difference (12) = 290.76, p < .001, and Model 5, χ² difference (11) = 78.88, p < .001. These results suggest that Sadism may be a facet of trait schadenfreude.

Discriminant validity

Reliability and descriptive statistics for discriminant validity measures can be found in Table 4 and bivariate correlations between schadenfreude, Sadism, and our discriminant validity measures can be found in Table 5.

Table 4.

Descriptive statistics for discriminant validity scales.

Scale	Min	Max	Mean	Std. dev	Skew	Kurtosis	α
Affective dissonance	1.00	3.92	1.53	0.65	1.52	1.69	0.90
BAS drive	1.00	4.00	2.26	0.57	0.03	−0.18	0.73
BAS fun-seeking	1.00	3.75	1.96	0.55	0.16	−0.41	0.64
BAS reward responsiveness	1.00	3.40	1.54	0.47	0.97	1.04	0.76
BIS	1.00	3.29	1.83	0.47	0.32	−0.36	0.71
BPAQ anger	1.14	7.57	3.33	1.20	0.80	−0.02	0.57
BPAQ hostility	1.00	7.00	3.60	1.27	0.06	−0.47	0.81
BPAQ physical aggression	1.00	7.00	2.73	1.16	0.65	−0.02	0.82
BPAQ verbal aggression	1.00	7.00	3.37	1.26	0.30	−0.25	0.74
Agreeableness	2.42	4.92	3.84	0.54	−0.32	−0.55	0.72
Conscientiousness	1.75	5.00	3.71	0.61	−0.38	−0.19	0.81
Extraversion	1.50	5.00	3.47	0.64	−0.32	0.25	0.80
Neuroticism	1.50	4.67	3.11	0.64	−0.14	−0.42	0.77
Openness	2.08	4.42	3.34	0.39	−0.16	0.09	0.68
Sadism aggregate	−0.59	2.00	0.00	0.59	1.42	1.47	0.93

BAS: behavioral activation system; BIS: behavioral inhibition system; BPAQ: Buss-Perry aggression questionnaire.

Table 5.

Bivariate correlations between sadism, schadenfreude, and discriminant validity scales.

Scale	Sadism aggregate	Schadenfreude scale total	Z-score comparison
Sadism aggregate	1.00	.65^***
Affective dissonance	.75^***	.75^***	0.00
BAS drive	−.04	.03	−1.48
BAS fun-seeking	−.07	−.08	0.21
BAS reward responsiveness	.15^***	.02	2.76^**
BIS	.22^***	.05	3.63^***
BPAQ anger	.43^***	.25^***	4.10^***
BPAQ hostility	.41^***	.40^***	0.24
BPAQ physical aggression	.57^***	.33^***	5.87^***
BPAQ verbal aggression	.42^***	.35^***	1.63
Agreeableness	−.58^***	−.36^***	5.45^***
Conscientiousness	−.26^***	−.19^***	1.52
Extraversion	.01	−.02	0.63
Neuroticism	.10	.16^**	−1.28
Openness	.04	.06	−0.42

Note. **p < .01, ***p < .001.

BAS: behavioral activation system; BIS: behavioral inhibition system; BPAQ: Buss-Perry aggression questionnaire.

Both Sadism and schadenfreude fit within a common antagonistic nomological network: both were positively correlated with affective dissonance—a predisposition towards experiencing the opposite emotional and psychological states as others (e.g., pleasure from another’s pain and pain from another’s pleasure)—and all facets of trait aggression, and negatively correlated with agreeableness and conscientiousness. However, schadenfreude and Sadism diverged on several traits. Sadism was more strongly correlated than schadenfreude with physical aggression, trait anger, agreeableness, BAS reward responsivity, and BIS. In addition, schadenfreude was significantly correlated with neuroticism, while Sadism was not, although this difference was not significant. This pattern of correlations suggests that Sadism can be differentiated from its broader construct of schadenfreude by inclinations towards direct aggression, greater anger, lower agreeableness, responsiveness towards rewards, and an ability to inhibit behavior.

Hierarchical factor structure of sadism and schadenfreude

Our bi-factor models give some preliminary evidence that Sadism may be a subfacet of broader schadenfreude, although this evidence is mixed and somewhat weak. Thus, in addition to a confirmatory approach, we also attempted an exploratory approach examining the hierarchical relationships between schadenfreude and Sadism using a “bass-ackwards” factor analysis (Goldberg, 2006). A bass-ackwards factor analysis is a factor analytic technique in which a set of items is subjected to a series of iterative hierarchical factor analyses with each iteration extracting an additional factor. This unique approach allows researchers to examine a hierarchical structure of latent constructs and see how higher-order constructs branch off into sub-facets. Bass-ackwards factor analyses have been previously used to sudy malevolent personality traits such as antagonism (Sleep et al., 2021), Narcissism (Crowe et al., 2019).

We expected to find,² within our bass-ackwards analysis, (H5) a two-factor solution capturing schadenfreude and Sadism, (H6) that the initial factor will have schadenfreude items loading above Sadism items, and (H7) that the initial factor should have stronger correlations with the second-level schadenfreude than second-level sadism. Supporting these hypotheses would give stronger evidence of a possible hierarchical relationship between Sadism and schadenfreude.

Data analytic strategy

The participants and materials were the same as in the previous analyses; specifically, in this analysis the CAST, ASP, SSIS, and Schadenfreude scale were used.

All items were standardized into Z-scores before being entered into the factor analysis. We next correlated all items with each other in order to identify any redundant or excessively overlapping items (i.e., r > .70) that could bias the factor analysis by creating artificial factors that arose merely because there were a large number of conceptually-redundant items. We then conducted Velicer’s minimum average partial (MAP) test using the EFA.dimensions package version 0.1.7.6 (O’Connor, 2023) for R statistical software (version 4.2.2, R Core Team, 2023) in order to establish an optimal factor solution.

Next, we conducted a bass-ackwards factor analysis, extracting factors in a stepwise fashion until we reached the number of factors suggested by the MAP analysis. Then, we extracted and examined factor loadings and item content at each level. If there was a factor onto which no items significantly loaded (i.e., <|.35|), or there was a factor determined by the research team to be either so conceptually unclear that it was not conceptually meaningful or interpretable, we reverted back to the previous factor solution (i.e., stage of the factor analysis with one fewer factor) and re-examined the loadings and content.

Once the final level was determined, we conducted bivariate correlations between each factor at each level and our discriminant validity measures in order to better understand each factor. Finally, we used Fisher’s r-to-Z transformation to compare the relative strength of each correlation.

Data preparation

Only one pair of items had a correlation greater than our preregistered threshold (i.e., r>|.70|) for redundant correlations, so we elected to retain both items. As recommended by previous bass-ackwards analyses, we removed all items that had weak loadings (<|.35|) onto a single factor in a single-factor EFA. This removed a total of four items, one from the CAST Vicarious scale (“There’s way too much violence in sports”), one from the CAST Verbal scale (“I would never purposely humiliate someone”), one from the SSIS (“I wouldn’t intentionally hurt anyone”), and one from the ASP Unempathic scale (“I would not purposely hurt anybody, even if I didn’t like them”). These items all represented the reverse-scored items from each of the scales. This left a total of 45 items for the bass-ackwards analysis. The results of our MAP analysis suggested a total of five factors.

Bass-ackwards analysis

We then conducted a bass-ackwards factor analysis with a maximum of five factors. However, after examining factor loadings and item content, the researcher team decided the fifth factor in the final model was too conceptually incoherent to be of use. Thus, we decided to retain the fourth-level model. This four-factor solution showed adequate fit via RMSEA (RMSEA = .06, 90% CI: .05, .06), but not by Tucker-Lewis Index (TLI = .81), and explained a total of 42% of the variance. Full factor loadings can be found in Supplemental Materials (S2).

First-level

At the first level of our analysis, factor one (1.1, see Figure 3) explained 31% of the variance, and consisted of 45 items. The item content included both Sadism and schadenfreude content, capturing a general enjoyment of the suffering of others. We named this broad, top-level factor Harm Enjoyment. At this level, the four highest loading items were items capturing the active enjoyment and perpetration of harm (i.e., Sadism; e.g., “Being mean to others can be exciting”) although the fifth and six highest-loading items captured the passive enjoyment of harm (i.e., schadenfreude; e.g., “Watching people get into fights excites me”). This gives mixed evidence for H5, but is not consistent with our broader expectations that schadenfreude items would load highest onto a single factor.

Figure 3.

Final model bass-ackwards factor analysis structure. Note. Only correlations going from one level of the hierarchy to the next are depicted. Correlations less than .65 were excluded from the figure. Full within- and between-level correlations can be found in Supplemental Materials S3.

Second-level

At the second level (i.e., two factors) of our analysis, the first-level, higher-order Harm Enjoyment factor split into two factors that explained a total of 36% of the variance, supporting H4. Factor 2.1 (see Figure 3) explained 22% of the variance, and captured the tendency to inflict harm on others and enjoy causing others to suffer (e.g., “I enjoy physically hurting people”). These items came from the ASP Pleasure-seeking (three items), ASP Subjugation (three items), ASP Unempathic (one item), CAST Physical (five items), CAST Verbal (three items), and SSIS (nine items). This fits under most conceptualizations of Sadism, thus we called this factor Sadism. Factor 2.2 explained 14% of the variance, and captured the tendency to enjoy passively viewing the suffering of others (e.g., “I take pleasure in another’s failure.”). These items came from the ASP Pleasure-seeking (one item) CAST Verbal (two items), CAST Vicarious (one item), and schadenfreude scale (12 items). We called this factor Schadenfreude. Also at this level, four items did not load onto either factor; these items all came from the CAST Vicarious scale. Disconfirming H6, our first-level, higher-order Harm Enjoyment factor was more strongly correlated with second-level Sadism, r = .94, than second-level Schadenfreude, r = .84.

Examining the component scales for each factor, the second-level Sadism factor was positively correlated with all Sadism and schadenfreude measures (Supplemental Materials S4), and most strongly correlated with the SSIS. 2.2 Second-level Schadenfreude was moderately positively correlated with all Sadism and schadenfreude measures, and most strongly with the trait schadenfreude scale and the CAST vicarious Sadism scale (Supplemental Materials S4).

Third-level

At the third-level, the second-level Sadism factor (i.e., 2.1) split into two lower-order factors, while the second-level Schadenfreude factor (i.e., 2.2) remained mostly intact. Factor 3.1 included Sadistic items capturing the positive affect and hedonistic experience of causing harm (e.g., “Hurting people would be exciting”). This factor contained fifteen items, and explained 14% of the variance. These items came from the ASP Pleasure-seeking (one item), ASP Subjugation (one item), ASP Unempathic (one item), CAST Physical (three items), CAST Verbal (three items), CAST Vicarious (three items), and SSIS (seven items). We thus named this factor Sadistic Hedonism. Factor 3.2 also emerged from the second level Sadism factor, and consisted of ten items explaining an additional 14% of the variance. This factor included Sadism items capturing a desire for dominating and bullying others (e.g., “I have made fun of people so that they know I am in control”). These items came from the ASP Pleasure-seeking (three items), ASP Subjugation (two items), CAST Verbal (four items), and SSIS (one item). We named this factor Sadistic Dominance. Factor 3.3 was composed of nine items that explained 11% of the variance, and which captured the passive enjoyment of harm. These items came from the schadenfreude scale. Thus, we called this factor Schadenfreude (e.g., “I enjoy watching segments of videos where people fall”), similar to Factor 2.2. At this level, there were four items with problematic cross-loadings and seven that did not significantly load onto any factor.

Fourth-level

At the final, fourth level, Sadistic Dominance remerged from level three as factor 4.1, consisting of 12 items explaining 13% of the variance and capturing a desire to inflict harm in order to gain dominance over or bully others (e.g., “I have made fun of people so that they know I am in control.”). These items came from the ASP Pleasure-seeking (three items), ASP Subjugation (one item), CAST Physical (two items), CAST Verbal (five items), and SSIS (one item). Among component scales, this factor was most strongly correlated with the ASP pleasure-seeking scale and the CAST Verbal Sadism scale (Supplemental Materials S4). In addition, Sadistic Hedonism reemerged as factor 4.2, consisting of seven items explaining 11% of the variance, and capturing the pursuit of positive affect through harm (e.g., “I have hurt people for my own enjoyment”). These items all came from the SSIS. Among component scales, this correlated most strongly with the SSIS. At this final level, the third-level Schadenfreude factor (i.e., 3.3) split into two lower-order factors (Factors 4.3 and 4.4). Factor 4.3 contained seven items explaining 9% of the variance, and captured a passive enjoyment of others’ suffering from relatively minor incidents (e.g., “It’s funny when a person walks into a closed sliding-glass door”). All of these items came from the schadenfreude scale. As such, we called this Benign Schadenfreude. Among component scales, this factor correlated most strongly with the schadenfreude scale. Factor 4.4 contained six items which explained 9% of the variance, and captured a passive enjoyment of others’ suffering from relatively severe harm, both real and fictional (e.g., “I love to watch YouTube clips of people fighting”; “I sometimes replay my favorite scenes from gory slasher films”). These items came from the ASP Pleasure-seeking (one item), CAST Physical (one item), and CAST Vicarious scales (four items). We named this factor Malicious Schadenfreude. Among component scales, this factor correlated most strongly with the CAST Vicarious scale. Also at this level, five items were removed for low loadings (i.e., <|.35|) and five items were removed for problematic cross-loadings (>|.35| difference between factor loadings).

Discriminant validity

Correlations between discriminant validity measures and fourth-level factor scores can be found in Table 6. Discriminant validity correlations for second- and third-level factors can be found in Supplemental Materials (S5). All four factors had expected correlations with antagonistic constructs; however, there were notable differences in the relative strength of correlations.

Table 6.

Bivariate correlations between discriminant validity measures and fourth-level factor scores.

Discriminant validity scale	Sadistic dominance	Sadistic hedonism	Benign schadenfreude	Malicious schadenfreude
Affective dissonance	.68^**	.64^**	.68^**	.57^**
BAS drive	−.06	<.01	.03	.02
BAS fun-seeking	−.07	−.02	−.09	−.03
BAS reward responsivity	.07	.18^**	−.03	.15^**
BIS	.15^**	.14^*	.03	.26^**
BPAQ anger	.41^**	.29^**	.22^**	.25^**
BPAQ hostility	.39^**	.33^**	.37^**	.25^**
BPAQ physical aggression	.42^**	.42^**	.31^**	.52^**
BPAQ verbal aggression	.43^**	.27^**	.33^**	.27^**
Callousness	.41^**	.36^**	.36^**	.30^**
Dominance	.49^**	.43^**	.38^**	.35^**
Agreeableness	−.54^**	−.47^**	−.29^**	−.40^**
Conscientiousness	−.24^**	−.26^**	−.14^*	−.11
Extraversion	.03	−.03	−.01	.02
Neuroticism	.08	.10	.15^**	< −.01
Openness	.05	.05	.08	−.01

Note. *p < .05, **p < .01.

BAS: behavioral activation system; BIS: behavioral inhibition system; BPAQ: Buss-Perry aggression questionnaire.

Comparing factor score correlations, Sadistic Hedonism, Sadistic Dominance, and Benign Schadenfreude were all significantly higher on affective dissonance compared to Malicious Schadenfreude (Z’s = 1.85–2.91, p’s < .050), but did not differ from each other. Sadistic Dominance was most strongly correlated with agreeableness (Z’s = 1.79–5.25, p’s < .050) and anger (Z’s = 2.78–3.76, p’s < .050) compared to the other three factors. Sadistic Hedonism and Malicious Schadenfreude were more strongly correlated with BAS reward responsivity than Sadistic Dominance and Benign Schadenfreude (Z’s = 1.84–3.51, p’s < .050), but did not differ from each other. Malicious Schadenfreude was most strongly correlated with BIS (Z’s = 2.15–4.14, p’s < .050) while Benign Schadenfreude was most weakly correlated with it (Z’s = 1.84–4.14, p’s < .050). Sadistic Dominance was more strongly correlated with callousness compared to Malicious Schadenfreude (Z = 2.28, p = .011), but was just as strongly correlated compared to Sadistic Hedonism and Benign Schadenfreude, which also did not differ from each other (Z’s = < .01–1.24, p’s > .050). Both Sadism factors were more strongly negatively correlated with conscientiousness compared to the two schadenfreude factors (Z’s = 1.89–3.03, p’s < .050), although neither the two Sadism factors (Z = 0.44, p = .329) nor the two schadenfreude factors (Z = 0.53, p = .298) significantly differed from each other. This pattern was similar for dominance, such that Sadistic Dominance was more strongly correlated with dominance compared to both facets of schadenfreude (Z’s = 2.32–3.02, p’s < .050), although not compared to Sadistic Hedonism (Z = 1.48, p = .07). Furthermore, Sadistic Hedonism was more strongly correlated with dominance than Malicious Schadenfreude (Z = 1.48, p = .07). However, the two schadenfreude facets did not significantly differ from each other (Z = 0.58, p = .280). Although all four constructs were positively correlated with trait hostility, Malicious Schadenfreude was most weakly correlated with it (Z’s = 1.67–2.86, p < .050) compared to the other factors. Benign Schadenfreude was the only factor significantly correlated with neuroticism, and more strongly correlated with it than Malicious Schadenfreude (Z = 2.67, p = .004) was, although this correlation was not significantly different from the correlation between Sadistic Dominance and Sadistic Hedonism (Z’s = 0.84–1.30, p’s > .050). Curiously, Malicious Schadenfreude was most strongly correlated with physical aggression compared to all other facets (Z’s = 2.23–4.23, p’s < .05), while Benign Schadenfreude was the weakest correlation (Z’s = 2.01–4.23, p’s < .05), with both Sadism facets not differing from each other (Z <.01, p = .500). This is somewhat unexpected, given Sadism’s strong links to physical aggression, and is most likely due to the unexpected loading of the CAST Physical sadism item “I enjoy physically hurting people” moving from 3.1 to 4.3. Finally, Sadistic Dominance was most strongly correlated with verbal aggression compared to all other facets (Z’s = 2.04–3.71, p’s < .050), which did not differ from each other (Z’s = < .01–1.13, p’s > .050).

Discussion

The primary motivations of this study were two-fold: to clarify whether three peripheral facets of Sadism truly belonged in the conceptual core of this construct, and to more thoroughly map the relationship between Sadism and schadenfreude. Given Sadism’s strong relations to real-world aggression (Chester et al., 2019), it is of critical importance to understand its underlying structure and nomological network. The field currently lacks a consistent definition of Sadism and also lacks an understanding of the aspects of Sadism that are unique or shared with schadenfreude. This lack of conceptual clarity is an obstacle to research on these constructs. Our study sought to provide such clarity to the literature in order to better understand and ameliorate these antisocial tendencies.

The structure of sadism

Our attempts to tease apart the callous, dominant, and vicarious facets of sadism via Dance Partner Modeling were unsuccessful. Against our predictions, it appears that these three ingredients are required in the recipe for the broad Sadism construct. Indeed, our bass-ackwards exploratory factor analysis revealed a distinct facet of Sadism concerned with dominance and bullying. The disposition towards enjoying the suffering of others would also include the enjoyment of dominating others. In order to overcome the natural empathic reaction to another person being harmed, individuals high in Sadism would need to be callous to their victim’s suffering. This supports the inclusion of these constructs in further Sadism studies. Furthermore, it is natural that an individual who enjoys harming others would experience a similar degree of joy from passively seeing others harmed. However, contrary to our predictions, vicarious Sadism was not redundant with schadenfreude. This is somewhat surprising given the results of our bass-ackwards factor analysis, in which vicarious Sadism items from the CAST scale loaded onto lower-order schadenfreude facets. This highlights the importance of using both exploratory and confirmatory techniques.

The relationship between sadism and schadenfreude

To our knowledge this is the first study to compare Sadism to trait schadenfreude, thus putting both constructs on the same phenomenological level. The findings of our bi-factor models imply that Sadism may be a facet of a broader schadenfreude, but these results were somewhat weak and inconclusive. Our bivariate correlations between the two constructs showed that, although both Sadism and schadenfreude were positively correlated with trait aggression, Sadism was more strongly correlated with physical aggression and anger than schadenfreude was, suggesting that it is Sadism that propels direct aggression. This was further supported by Sadism’s correlation with reward responsiveness and schadenfreude’s relationship with neuroticism.

The results of our bass-ackwards factor analysis showed that Sadism and schadenfreude fit together under a common higher-order construct, but split into the active pursuit of harm (i.e., Sadism) and passive enjoyment of harm (i.e., schadenfreude) at lower levels. Furthermore, these constructs each split into two distinct subfactors in our final level. Sadism seems to contain Dominance and Hedonism facets, while schadenfreude contains Benign and Malicious facets. The specific dominance facet of Sadism further supports its inclusion in conceptual models of sadism (e.g., Plouffe et al., 2017), against our expectations.

These results may help distinguish trait schadenfreude as more passive—experiencing joy from others pain but not necessarily being the cause of it—as opposed to more direct Sadism—enjoying actively harming others, as others have suggested (e.g., Ben-Ze’ev, 2014). Furthermore, the distinction between benign and malicious forms of schadenfreude may help clarify the specific kind of schadenfreude that leads to Sadistic pursuits. Indeed, the Malicious Schadenfreude factor, but not the Benign Schadenfreude factor, was strongly correlated with Sadistic Hedonism, suggesting a type of schadenfreude in response to grievous harm that most closely abuts—and perhaps even overlaps with—Sadism.

However, Sadism does not fit into any of the pre-existing perspectives of schadenfreude. Within the body of scientific literature, there are three broad perspectives of why and when schadenfreude occurs: justice (e.g., Berndsen & Tiggemann, 2020), envy (e.g., van Dijk et al., 2015), and group identity (e.g., Combs et al., 2009). Although some have found that Sadism is linked to envy (e.g., Dinic & Brankovic, 2022) and dominance is a part of Sadism (e.g., Plouffe et al., 2017), the Sadistic enjoyment of harm is not contingent on envying the target, nor is it concerned with in-group benefit—but rather the pure hedonism gained through aggression. Thus, new conceptualizations of schadenfreude that focus on its malicious facet, and can explain its link to Sadism are needed to unify these constructs.

The Malicious Schadenfreude facet may be a clear target for intervention as it likely serves as a bridge between schadenfreude and Sadism. This supposition is based on the fact that it was the only schadenfreude facet to be strongly linked to Sadistic Hedonism. Although we lack the longitudinal data to test whether changes in Malicious Schadenfreude may predict subsequent changes in Sadistic traits, our cross-sectional findings offer preliminary evidence that this may be the case. If interventions are able to reduce this form of schadenfreude, they may then effectively disrupt the development of Sadistic traits. By keeping schadenfreude benign, we may be able to prevent people from seeking to inflict the harm that they tend to enjoy.

Are individuals who enjoy viewing harm destined to commit it themselves? Clearly, the separation of Sadism and schadenfreude as lower-order factors means that although there is a clear overlap between those who enjoy causing harm to others and those who merely enjoy witnessing it, these two dispositions have distinct nomological profiles. How, why, and when do individuals who experience schadenfreude decide to directly pursue Sadistic harm? Future research should leverage longitudinal designs to investigate schadenfreude as a risk factor or predispositional precursor to Sadism. Experimental and computational approaches might further the study of these topics by disentangling the mechanisms that differentially motivate and reinforce Sadism and schadenfreude. Such studies might allow us to prevent a person prone to the passive enjoyment of others’ suffering from seeking to inflict such suffering themselves.

Limitations and future directions

Although our bi-factor models showed some evidence that Sadism is a lower-order facet of schadenfreude, this conceptual structure did not emerge in our bass-ackwards factor analysis. Thus, the exact nature of the overlap or nested hierarchy of Sadism and schadenfreude is still unknown. The present study only investigated these constructs at the trait level, and not at the experiential or behavioral levels, and cannot necessarily comment on the experience of schadenfreude nor the perpetration of Sadistic aggression. Future studies using behavioral paradigms, as well as experiential schadenfreude, may help to elucidate this.

A further limitation is the sample size for our analyses. Although we conducted an a priori power analysis for our Dance Partner Modeling and bifactor modeling based on model degrees of freedom, it is entirely possible that one of the reasons why these models had subpar fit is due to an inadequate sample size.

Another limitation in the current study is its sample characteristics. Although the sample was relatively racially and ethnically diverse, it still consisted of undergraduate students in the United States, and the majority of participants were cisgender women. There is little evidence to suggest that the core construct of Sadism or its relationship with schadenfreude would differ by age, it could very well differ by gender identity and cultural context. In this study, we found that men were higher than women on all facets of Sadism besides ASP Subjugation and were higher than women on schadenfreude. This raises the possible issue of measurement variance between men and women within the scales used in this study. Previous research applying differential item functioning to the ASP found that with the exception of one item from the Pleasure-seeking subscale (“Watching people get into fights excites me”) and one item from the Unempathic subscale (“I would not purposely hurt anybody, even if I didn’t like them”), ASP items functioned equally across men and women (Plouffe et al., 2021). However, this still leaves the gender invariance of the SSIS, CAST, and Schadenfreude scales unexamined. Ultimately, establishing measurement invariance across gender identities is outside of the scope of this study, but future studies should attempt to replicate the models presented in this study within a more gender-balanced sample. In addition, although the experience of schadenfreude exists in other cultures—for instance, Japan has a similar expression capturing the experience of joy at another’s pain: “the misfortunes of others taste like honey” (van Dijk & Ouwerkerk, 2014, p. 4)—individuals in collectivist cultures, or cultures with different social display rules than the United States may experience or report schadenfreude differently. Future research should thus use cross cultural samples.

This study opens the door to a revision of preexisting scales. For instance, several items from Sadism scales more strongly loaded onto the second-level schadenfreude factor than the Sadism factor. These items should be considered for removal from Sadism scales and perhaps added to schadenfreude measures.

Conclusion

Sadism is arguably the most malevolent personality trait studied within psychology, showing the strongest links to aggression (e.g., Chester et al., 2019), and is of critical importance to understand. The present study explored Sadism and schadenfreude in order to examine the latent network of these constructs and clarify the nature of their relationship. By studying how the active perpetration of harm for the sake of pleasure emerges from the passive enjoyment of harm, we hope that our findings help researchers, policy-makers, and practitioners to understand the nuanced ways in which pleasure can lead to the infliction of pain.

Supplemental Material

Supplemental Material - Disentangling sadism from schadenfreude as traits using latent construct modeling

Supplemental Material for Disentangling sadism from schadenfreude as traits using latent construct modeling by Drew M. Parton and David S. Chester in Personality Science

Footnotes

Author Note

Not applicable.

Acknowledgements

Not applicable.

Author contributions

Drew M. Parton: Conceptualization,Data curation,Formal analysis,Investigation,Methodology,Project administration,Visualization,Writing,Writing - review & editing.

David S. Chester: Conceptualization,Methodology,Supervision,Writing.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research,authorship,and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research,authorship,and/or publication of this article: This research was in part supported by the National Institute on Alcohol Abuse and Alcoholism (NIAAA) of the National Institutes of Health under award number K01AA026647 (PI: Chester).

ORCID iD

Drew M. Parton

Data Accessibility Statement

Materials,analysis scripts,and data can be found on the public Open Science Framework page for the project:

Supplemental material

Supplemental material for this article is available online. Depending on the article type,these usually include a Transparency Checklist,a Transparent Peer Review File,and optional materials from the authors.

Notes

References

American Psychiatric Association . (1987). Diagnostic and statistical manual of mental disorders (3rd ed., rev.).

Ben-Ze’ev

(2014). The personal comparative concern in schadenfreude. In van Djik

W. W.

Ouwerkerk

J. W.

(Eds.), Schadenfreude: Understanding pleasure at the misfortune of others (pp. 77–90). Cambridge University Press.

Berndsen

Tiggemann

(2020). Multiple versus single immoral acts: An immoral person evokes more schadenfreude than an immoral action. Motivation and Emotion, 44(5), 738–754. https://doi.org/10.1007/s11031-020-09843-5

Bonifay

Lane

S. P.

Reise

S. P.

(2017). Three concerns with applying a bifactor model as a structure of psychopathology. Clinical Psychological Science, 5(1), 184–186. https://doi.org/10.1177/216770261665706

Buckels

E. E.

Jones

D. N.

Paulhus

D. L.

(2013). Behavioral confirmation of everyday Sadism. Psychological Science, 24(11), 2201–2209. https://doi.org/10.1177/0956797613490749

Buckels

E. E.

Paulhus

D. L.

(2013). Comprehensive assessment of sadistic tendencies (CAST). Unpublished measure. University of British Columbia.

Buss

A. H.

Perry

(1992). The aggression questionnaire. Journal of personality and social psychology, 63(3), 452.

Carver

C. S.

White

T. L.

(1994). Behavioral inhibition, behavioral activation, and affective responses to impending reward and punishment: the BIS/BAS scales. Journal of personality and social psychology, 67(2), 319. https://doi.org/10.1037/0022-3514.67.2.319

Chester

D. S.

DeWall

C. N.

Enjaian

(2019). Sadism and aggressive behavior: Inflicting pain to feel pleasure. Personality and Social Psychology Bulletin, 45(8), 1252–1268. https://doi.org/10.1177/0146167218816327

10.

Combs

D. J.

Powell

C. A.

Schurtz

D. R.

Smith

R. H.

(2009). Politics, schadenfreude, and ingroup identification: The sometimes happy thing about a poor economy and death. Journal of Experimental Social Psychology, 45(4), 635–646. https://doi.org/10.1016/j.jesp.2009.02.009

11.

Crowe

M. L.

Lynam

D. R.

Campbell

W. K.

Miller

J. D.

(2019). Exploring the structure of narcissism: Toward an integrated solution. Journal of Personality, 87(6), 1151–1169. https://doi.org/10.1111/jopy.12464

12.

Crysel

L. C.

Webster

G. D.

(2018). Schadenfreude and the spread of political misfortune. PLoS One, 13(9), Article e0201754. https://doi.org/10.1371/journal.pone.0201754

13.

Dinić

B. M.

Branković

(2021). Cross-cultural comparison of the benign and malicious envy scale (BeMaS) across Serbian and US samples and further validation. European Journal of Psychological Assessment, 38(1), 49–60. https://doi.org/10.1027/1015-5759/a000643

14.

Foulkes

(2019). Sadism: Review of an elusive construct. Personality and Individual Differences, 151, 109500. https://doi.org/10.1016/j.paid.2019.07.010

15.

Goldberg

L. R.

(1999). A broad-bandwidth, public domain, personality inventory measuring the lower-level facets of several five-factor models. Personality psychology in Europe, 7(1), 7–28.

16.

Goldberg

L. R.

(2006). Doing it all bass-ackwards: The development of hierarchical factor structures from the top down. Journal of Research in Personality, 40(4), 347–358. https://doi.org/10.1016/j.jrp.2006.01.001

17.

Goldberg

L. R.

Johnson

J. A.

Eber

H. W.

Hogan

Ashton

M. C.

Cloninger

C. R.

Gough

H. G.

(2006). The international personality item pool and the future of public-domain personality measures. Journal of Research in personality, 40(1), 84–96. https://doi.org/10.1016/j.jrp.2005.08.007

18.

Heinrich

Zagorscak

Eid

Knaevelsrud

(2020). Giving G a meaning: An application of the bifactor-(S-1) approach to realize a more symptom-oriented modeling of the Beck depression inventory–II. Assessment, 27(7), 1429–1447. https://doi.org/10.1177/1073191118803738.2016.1144192

19.

Lee

S. A.

(2019). The dark Tetrad and callous reactions to mourner grief: Patterns of annoyance, boredom, entitlement, schadenfreude, and humor. Personality and Individual Differences, 137, 97–100. https://doi.org/10.1016/j.paid.2018.08.019

20.

Maples-Keller

J. L.

Williamson

R. L.

Sleep

C. E.

Carter

N. T.

Campbell

W. K.

Miller

J. D.

(2019). Using item response theory to develop a 60-item representation of the NEO PI–R using the international personality item pool: Development of the IPIP–NEO–60. Journal of Personality Assessment, 101(1), 4–15. https://doi.org/10.1080/00223891.2017.1381968

21.

Moshagen

Erdfelder

(2016). A new strategy for testing structural equation models. Structural Equation Modeling, 23, 54–60. https://doi.org/10.1080/10705511.2014.950896

22.

Nell

(2006). Cruelty’s rewards: The gratifications of perpetrators and spectators. Behavioral and Brain Sciences, 29(3), 211–224. https://doi.org/10.1017/S0140525X06009058

23.

O’Connor

B. P.

(2023). EFA.dimensions: Exploratory factor analysis functions for assessing dimensionality. R package version 0.1.7.6. https://CRAN.R-project.org/package=EFA.dimensions

24.

O'Meara

Davies

Hammond

(2011). The psychometric properties and utility of the Short sadistic Impulse scale (SSIS). Psychological Assessment, 23(2), 523–531. https://doi.org/10.1037/a0022400

25.

Paulhus

D. L.

Dutton

D. G.

(2016). Everyday sadism. In Zeigler-Hill

Marcus

D. K.

(Eds.), The dark side of personality: Science and practice in social, personality, and clinical psychology (pp. 109–120). American Psychological Association. https://doi.org/10.1037/14854-006

26.

Plouffe

R. A.

Kowalski

C. M.

Tremblay

P. F.

Saklofske

D. H.

Rogoza

Di Pierro

Chahine

(2021). Gender differences or gender bias? European Journal of Psychological Assessment, 37(6), 440–449. https://doi.org/10.1027/1015-5759/a000634

27.

Plouffe

R. A.

Saklofske

D. H.

Smith

M. M.

(2017). The assessment of sadistic personality: Preliminary psychometric evidence for a new measure. Personality and Individual Differences, 104, 166–171. https://doi.org/10.1016/j.paid.2016.07.043

28.

R Core Team . (2023). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. https://www.R-project.org/.

29.

Rosseel

(2012). lavaan: An R Package for Structural Equation Modelin. Journal of Statistical Software, 48(2), 1–36. https://doi.org/10.18637/jss.v048.i02

30.

Schumpe

B. M.

Lafrenière

M. A. K.

(2016). Malicious joy: Sadism moderates the relationship between schadenfreude and the severity of others’ misfortune. Personality and Individual Differences, 94, 32–37. https://doi.org/10.1016/j.paid.2016.01.005

31.

Simms

L. J.

Goldberg

L. R.

Roberts

J. E.

Watson

Weite

Rotterman

J. H.

(2014). Computerized adaptive assessment of personality disorder: Introducing the CAT-PD project. Personality Assessment in the, DSM-5, 73–82.

32.

Sleep

C. E.

Crowe

M. L.

Carter

N. T.

Lynam

D. R.

Miller

J. D.

(2021). Uncovering the structure of antagonism. Personality Disorders: Theory, Research, and Treatment, 12(4), 300. https://doi.org/10.1037/per0000416

33.

Smith

R. H.

Powell

C. A.

Combs

D. J.

Schurtz

D. R.

(2009). Exploring the when and why of schadenfreude. Social and Personality Psychology Compass, 3(4), 530–546. https://doi.org/10.1111/j.1751-9004.2009.00181.x

34.

Smith

R. H.

Turner

T. J.

Garonzik

Leach

C. W.

Urch-Druskat

Weston

C. M.

(1996). Envy and schadenfreude. Personality and Social Psychology Bulletin, 22(2), 158–168. https://doi.org/10.1177/0146167296222005

35.

Vachon

D. D.

Lynam

D. R.

(2016). Fixing the problem with empathy: Development and validation of the affective and cognitive measure of empathy. Assessment, 23(2), 135–149. https://doi.org/10.1177/1073191114567941

36.

Van Buuren

Groothuis-Oudshoorn

(2011). mice: Multivariate Imputation by Chained Equations in R. Journal of Statistical Software, 45(3), 1–67. https://doi.org/10.18637/jss.v045.i03

37.

van Dijk

W. W.

Ouwerkerk

J. W.

(2014). Introduction to schadenfreude. In van Dijk

W. W.

Ouwerkerk

J. W.

(Eds.), Schadenfreude: Understanding pleasure at the misfortune of others (pp. 1–11). Cambridge University Press.

38.

van Dijk

W. W.

Ouwerkerk

J. W.

Smith

R. H.

Cikara

(2015). The role of self-evaluation and envy in schadenfreude. European Review of Social Psychology, 26(1), 247–282. https://doi.org/10.1080/10463283.2015.1111600

39.

van Dijk

W. W.

van Koningsbruggen

G. M.

Ouwerkerk

J. W.

Wesseling

Y. M.

(2011). Self-esteem, self-affirmation, and schadenfreude. Emotion, 11(6), 1445–1449. https://doi.org/10.1037/a0026331

40.

Wang

Lilienfeld

S. O.

Rochat

(2019). Schadenfreude deconstructed and reconstructed: A tripartite motivational model. New Ideas in Psychology, 52, 1–11. https://doi.org/10.1016/j.newideapsych.2018.09.002

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.68 MB

0.00 MB