Abstract
Sex differences in reactions to partner infidelity have often been studied by comparing emotional reactions to scenarios of sexual versus emotional infidelity. Men, relative to women, tend to react with more distress to partner sexual infidelity than to emotional infidelity. Evolutionary theorists interpret this difference as evidence of sexually dimorphic selection pressures. In contrast, focusing only on the simple effects within each sex, social-cognitive theorists suggest that men and women do not differ in their reactions to partner infidelity. As evidenced by recent rival meta-analytic reports, these diverging perspectives remain largely unresolved and contentious. The present study was designed to take a new approach by measuring attitudes toward partner infidelity. Results were consistent with the evolutionary perspective: Men, to a significantly larger degree than women, evaluated partner sexual infidelity more negatively than emotional infidelity.
Introduction
In one of the most cited evolutionary psychology studies, Buss, Larson, Westen, and Semmelroth (1992) asked college students which would distress or upset them more: Imagining their romantic partner forming a deep emotional attachment to another person, or enjoying passionate sexual intercourse with that person. Evolutionary theory posits that due to sexually dimorphic selection pressures, men and women have evolved different sensitivities to emotional versus sexual infidelity. Indeed, 60% of men but only 17% of women selected sexual infidelity as most distressing. Although the specific percentages have varied somewhat across subsequent studies, this pattern of results has been replicated many times (see meta-analyses by Carpenter, 2012; Harris, 2003).
Despite these consistent results, the evolutionary account of sex differences in jealousy has been questioned. For example, Desteno and Salovey (1996) offered the “double shot hypothesis” as an alternative explanation. When forced to choose only one option that is most distressing, they argue that people reasonably select the type of infidelity that they believe most likely implies the co-occurrence of both types of infidelity. If men tend to believe that women will only engage in sex after falling in love (Harris and Christenfield, 1996), men will perceive sexual infidelity as most distressing. Thus, the reasoning goes, if a woman has formed a sexual relationship with another man she likely also has strong emotional feelings toward the man. On the other hand, if women believe that men eagerly have sex prior to, and often without ever, falling in love, women will likely select emotional infidelity as worse. Stated differently, when a man is emotionally attached to a woman it implies a much more serious relationship – one that has moved beyond sexual activity. However, the “double shot” hypothesis has not been supported when carefully pitted against evolutionary predictions (Buss et al., 1999).
Following the lack of support for the double-shot hypothesis, DeSteno and colleagues (DeSteno, Bartlett, Braverman, and Salovey, 2002; DeSteno, Bartlett, and Salovey, 2006; DeSteno, 2010) more recently argued that sex differences on forced-choice measures are the result of a format-induced measurement artifact. Specifically, DeSteno and colleagues posit that the forced-choice scenario leads participants to engage in an effortful and gender socialized decision strategy concerning the likely outcomes and tradeoffs of each form of infidelity. Unfortunately, the specific considerations and trade-offs are not specifically detailed. In fact, and despite the results of Buss et al. (1999), the only explicit reference of a possible gender socialized decision strategy remains the aforementioned “double shot hypothesis” (DeSteno et al., 2002).
To support the measurement artifact hypothesis, DeSteno and colleagues argue that sex differences disappear under cognitions of cognitive load and when participants separately rate their levels of distress to sexual and emotional infidelity on continuous scales. In response, evolutionary theorists have demonstrated sex differences under both deliberative and automatic (cognitive load) conditions (Barrett, Frederick, Haselton, and Kurzban, 2006; Schutzwohl, 2007, 2008), and argued that smaller sex differences on continuous scales can be accounted for because of ceiling (Buss et al., 1999; Edlund, Heider, Scherer, Farc, and Sagarin, 2006) and scale labeling effects (Edlund and Sagarin, 2009).
Meta-analytic results
Attempting to resolve this continuing debate (DeSteno, 2010; Edlund, 2011), two recently published meta-analyses were independently conducted (Carpenter, 2012; Sagarin et al., 2012). Importantly, both focused on the results of studies employing continuous scales. However, because of disagreements regarding the critical statistical analyses needed to test evolutionary predictions, these meta-analyses reached opposite conclusions.
In line with social-cognitive theorists (e.g., DeSteno and Salovey, 1996; Harris, 2005, Harris and Christenfeld, 1996), Carpenter (2012) performed his meta-analysis with the view that the evolutionary perspective implies two critical, and separate, predictions: 1) Men will report significantly more distress to sexual infidelity than they do to emotional infidelity, and 2) Women will report significantly more distress to emotional infidelity than they do to sexual infidelity. With this focus on simple effects within participant sex, the Carpenter meta-analysis found that both men and women tend to rate sexual infidelity as more distressing than emotional infidelity, leading Carpenter to conclude that his results were consistent with the social-cognitive hypothesis of no sex differences in infidelity distress. Not emphasized by Carpenter is the fact that the sexual minus emotional difference was larger for men than it was for women. As described below, this pattern of results is actually consistent with evolutionary predictions.
Taking a markedly different approach, Sagarin et al. (2012) conducted their meta-analysis to examine the Participant Sex × Infidelity Type interaction. This approach is based on the acknowledgement that many factors other than sexually dimorphic selection pressures influence reactions to partner infidelity. These factors include the tendency for women to report more intense emotions than men, the operationalization of infidelity, and anchoring effects from the wording used on response scales (see Sagarin et al. for a review). Given such factors, in any given study, simple effects may appear to be inconsistent with evolutionary predictions (e.g., women reporting more distress to sexual infidelity than to emotional infidelity). However, the critical test of evolutionary theory according to Sagarin et al.'s perspective is detection of a significant relative sex difference: Men, relative to women, should report a more pronounced negative reaction to sexual infidelity than to emotional infidelity (Buss and Haselton, 2005; Sagarin, 2005). In other words, the sexual infidelity minus emotional infidelity difference should be larger for men than it is for women, resulting in a significant Participant Sex × Infidelity Type interaction. This perspective provides a logical account for the varied patterns of simple effects in previous studies, such as why in some studies both women and men are found to react more negatively in response to partner sexual infidelity and why in some studies women report similar levels of distress to both forms of infidelity. Indeed, with this focus the Sagarin et al. meta-analysis of 45 independent samples found a statistically significant Participant × Infidelity Type interaction in the evolutionary theory supportive direction.
In summary, debates about the existence and meaning of sex differences in response to partner infidelity continue, with little progress to date to resolve the seemingly entrenched evolutionary and social-cognitive positions. Although recent meta-analyses represent important progress, it is likely that they will serve to only bolster their respective polarized positions. The aim of the present study was to take a novel approach to the study of sex differences in reactions to partner infidelity. Specifically, this study was designed to investigate attitudes toward partner infidelity. We describe our rationale below for why we believe the attitude construct can contribute to our understanding of sex differences in response to partner infidelity.
Do men and women differ in their attitudes toward partner infidelity?
Given that much of the research on sex differences in reactions to partner infidelity has been conducted by social psychologists, and the central role of attitudes in social psychological research (Allport, 1935; Albarracín and Vargas, 2010), it is surprising that no previous studies have directly examined attitudes toward partner infidelity. Rather, because evolutionary theory has focused on sex differences in emotional states triggered by relationship threats, researchers have focused almost exclusively on individual affective reactions (e.g., jealousy, distress). Although some theoretical approaches equate attitude with affect, much of modern attitude theory describes attitudes as overall evaluative tendencies that reflect affective, cognitive, and behavioral experiences (Eagly and Chaiken, 1993). This framework is important because while very likely emotionally intense, reactions to partner infidelity are not purely affective. Cognitive responses are also part of the reaction to partner infidelity, such as attributions of blame versus understanding (Bauerle, Amirkhan, and Hupka, 2002; Hall and Fincham, 2006). In a rare evolutionary study examining sex differences in cognitive responses, Schutzwohl and Koch (2004) found that men, relative to women, preferentially process and have better memory for cues to sexual infidelity. Also rarely studied are overt behavioral responses, but they are certainly also an important part of real-world responses to partner infidelity. For example, research suggests that infidelity is the most common presenting problem for dating and married couples entering therapy (Glass and Wright, 1988) and the leading cause of divorce (Amato and Previti, 2003). Thus, examining attitudes toward partner infidelity is arguably a more thorough approach than asking about specific emotional reactions, in terms of understanding overall reactive tendencies. Moreover, it is also a much more efficient approach compared to separately measuring affective, cognitive, and behavioral responses.
Prior studies have assessed attitudes toward personally engaging in infidelity, or attitudes toward infidelity in general, and if such attitudes relate to behavior. Not surprisingly, those who have personally had “extra-relationship involvements” tend to have more positive attitudes and more likely intend to engage in such relationships again in the future (Drake and McCabe, 2000; Thompson, 1984). Similarly, attitudes toward “extradyadic sex” predict personal involvement and willingness to engage in future extradyadic relationships (Buunk and Bakker, 1995). In a study examining gender differences in attitudes toward reasons for extramarital relationships, Glass and Wright (1992) found that men have more positive attitudes toward sexual justifications for their own extramarital relationships, but women report more positive attitudes toward justifications for emotional relationships. Although these studies support the general prediction of sex differences in attitudes toward infidelity, we know of no previous study that has specifically examined attitudes toward partner infidelity.
In the present study we examined men and women's attitudes toward partner sexual and emotional infidelity. To put our results in context with previous studies, we also sought to replicate previous findings of relative sex differences in infidelity distress (Carpenter, 2012; Sagarin et al., 2012). Consistent with evolutionary predictions and the findings of recent meta-analyses, we expected the results from the attitude scales to be in line with those from the infidelity distress measures. Specifically, we hypothesized that relative to women, men's attitudes toward partner sexual infidelity would be significantly more negative than their attitudes toward partner emotional infidelity.
Materials and Methods
Participants
University students (N = 324) aged 18 years or older (212 females, 112 males) were recruited from an introductory psychology participant pool. Most were White (87.3%), heterosexual (93.2%), and of traditional college student age (M = 19.78 years, SD = 4.02). The Institutional Review Board at Ball State University approved this study.
Measures
Forced-choice infidelity distress. Following Buss et al. (1992), participants were asked to think of a serious committed relationship they have had in the past, are currently having, or would like to have and then to imagine that they discover their partner is involved with someone else. They were then asked to select which scenario they would find more distressing: Imagining their partner (a) has formed a deep emotional attachment with someone else, or (b) engaging in sexual intercourse with someone else.
Continuous infidelity distress. As in the forced-choice measure, participants were asked to think about a serious committed relationship and then to imagine that they discover their partner is involved with someone else. They rated on 5-point, endpoint-only labeled scales (1 = not distressing, 5 = very distressing) how distressed they would be if “You discover that your partner is engaging in sexual intercourse with someone else,” and if “You discover that your partner has formed a deep emotional attachment with someone else”.
Attitude scales. Participants rated “your romantic partner having sexual intercourse with another person” and “your romantic partner forming a deep emotional attachment to another person” separately on five 7-point semantic differential scales: good-bad, positive-negative, natural-unnatural, pleasant-unpleasant, and acceptable-unacceptable. Responses were averaged, with higher scores indicating more favorable attitudes. Internal reliability was good for both the sexual (α = .85) and emotional (α = .91) scales.
Procedure
Participants responded to the measures (presented in counterbalanced order) either in person in small, mixed gender sessions (n = 73) or online (n = 251). When included in the analyses reported below, no significant effects of in-person vs. online administration were found. The infidelity distress scales were always presented consecutively (with forced-choice and continuous formats being counterbalanced), and the attitude scales were always presented consecutively (with sexual and emotional infidelity counterbalanced). A demographic questionnaire was always presented last.
Results
Descriptive statistics and correlations are displayed in Table 1. Responses to the forced-choice distress measure were significantly correlated with scores on the continuous measure of distress to partner emotional infidelity, but not with scores on the continuous measure of distress to partner sexual infidelity. Attitudes toward partner infidelity were moderately and significantly correlated with continuous distress ratings, but not with the forced-choice measure.
Descriptive Statistics and Bivariate Correlations (N = 324)
p < .01. Participant Sex was coded 1 = female, 2 = male.
Forced-choice Distress was coded 1 = emotional infidelity, 2 = sexual infidelity.
Replication of gender differences in infidelity distress
Consistent with previous research employing forced-choice measures with college students, a 2 (Participant Gender) × 2 (Sexual or Emotional Infidelity selected as most distressing) chi-square test of independence was significant, ✗2 (1, N = 324) = 36.36, p < .001, ɸ = .34. The majority (80.7%) of women chose emotional infidelity as more distressing, but 51.8% of men chose sexual infidelity as more distressing.
To test for sex differences on the continuous measures of infidelity distress, a 2 (Participant Sex) × 2 (Infidelity Type: Sexual vs. Emotional) ANOVA with repeated measures on the second factor was performed. A main effect was found such that women (M = 4.78, SD = 0.52) reported more distress than men (M = 4.50, SD = 0.83), F(1, 322) = 21.60, p < .001, d = 0.40, ηp2 = .06. A main effect was also found for infidelity type, with partner sexual infidelity (M = 4.76, SD = 0.58) rated more distressing than emotional infidelity (M = 4.61, SD = 0.74), F(1, 322) = 19.08, p < .001, d = 0.23, ηp2 = .06. However, these main effects were qualified by a significant Participant Sex × Infidelity Type interaction, F(1, 322) = 12.61, p < .001, ηp2 = .04. Importantly, this interaction was in the evolutionary theory supportive direction: Men rated sexual infidelity significantly more distressing (M = 4.69, SD = 0.74) than they rated emotional infidelity (M = 4.32, SD = 0.92), F(1, 322) = 23.96, p < .001, d = 0.44, 95% CI [0.33, 0.55], but there was little difference between women's ratings of sexual (M = 4.80, SD = 0.48) and emotional infidelity (M = 4.76, SD = 0.57), F(1, 322) = 0.48, p = .29, d = 0.08, 95% CI [0.03, 0.13].
Sex differences in attitudes
To examine attitudes toward partner infidelity, a 2 (Participant Sex) × 2 (Infidelity Type: Sexual vs. Emotional) ANOVA with repeated measures on the second factor was conducted. A main effect was found such that women (M = 1.75, SD = 0.86) reported significantly more negative attitudes toward partner infidelity than men (M = 2.05, SD = 1.03), F(1, 322) = 11.18, p < .001, d = 0.32, ηp2 = .03. A main effect was also found for Infidelity Type, with sexual infidelity (M = 1.41, SD = 0.65) rated significantly more negatively than emotional infidelity (M = 2.29, SD = 1.23), F(1, 322) = 191.85, d = 0.89, p < .001, ηp2 = .37. However, these main effects were qualified by a significant Participant Sex × Infidelity Type interaction, F (1, 322) = 14.99, p < .001, ηp2 = .04. As expected, men rated sexual infidelity (M = 1.44, SD = 0.70) more negatively than they rated emotional infidelity (M = 2.66, SD = 1.37), F(1, 322) = 120.00, p < .001, d = 1.12, 95% CI [0.99, 1.26]. Although women also rated sexual infidelity (M = 1.40, SD = 0.62) more negatively than they rated emotional infidelity (M = 2.09, SD = 1.10), this difference was not as large and thus in the evolutionary theory supportive direction, F(1, 322) = 72.03, p < .001, d = 0.77, 95% CI [0.69, 0.85].
Supplementary Analyses
As noted by a reviewer to an earlier version of this manuscript, examination of the means and standard deviations in the preceding analyses suggests the continuous measures are skewed. Indeed, the skew was most evident for the continuous distress scores (skew = −2.70, −2.09 for sexual and emotional infidelity distress, respectively), but also was observed for the attitude scores (skew = 2.49, 0.98 toward sexual and emotional infidelity, respectively). Given the concern that these deviations from normality violate the assumptions of the ANOVAs and thus affected the results (p values), we also performed non-parametric tests of our hypotheses. To do so we first computed the difference between participants' sexual and emotional infidelity distress ratings (also see Tagler and Gentry, 2011), and the corresponding difference between participants' attitudes toward sexual and emotional infidelity. These difference scores represent the within-subjects factor (infidelity type) in the previously reported ANOVAs. As expected, for distress to partner infidelity the sexual – emotional difference was more pronounced for men (M = 0.37, SD = 0.94) than for women (M = 0.03, SD = 0.70). Likewise, the sexual – emotional difference in attitudes was also more pronounced for men (M = −1.21, SD = 1.38) than for women (M = −0.71, SD = 1.05).
First, we performed the Mann-Whitney procedure on the difference scores. Consistent with evolutionary theory and our ANOVA results, the test revealed a significant relative sex difference in partner infidelity distress ratings, U = 9575.50, z = 3.42, p = .001, r = -.19, and also in attitudes toward partner infidelity, U = 9408.00, z = −3.09, p = .002, r = -.17. In addition, we performed permutation tests using Howell's (2007) resampling program. Specifically, we determined the probability of obtaining relative sex differences more extreme than what we obtained under the null hypothesis that participant sex has no relation to reactions to partner infidelity. To do so, 100,000 resamples of the data were performed. Consistent with ANOVA results, the probability of obtaining our infidelity distress results (t = 3.55) was unlikely, with only 43 out of 100,000 (p = .00043) resampled results as extreme. Likewise, the probability of obtaining our attitude results (t = −3.61) was extremely rare, with just 36 out of 100,000 (p = .00036) resampled results as extreme.
Discussion
Across all measures, we found relative sex differences in reactions to partner infidelity. On both forced choice and continuous measures of distress, men were significantly more likely, compared to women, to perceive partner sexual infidelity as more distressing than emotional infidelity. Furthermore, and the unique aspect of the present study, men and women reported differences in their attitudes toward partner infidelity. Men, to a significantly larger degree than women, evaluated partner sexual infidelity more negatively than emotional infidelity. In combination, these results are consistent with the evolutionary perspective (Sagarin et al., 2012).
As has been the case with research examining sex differences in jealousy and distress, the present attitude results may be interpreted differently depending on one's theoretical allegiance, and on what one subsequently views as the critical comparisons. If past trends hold, social-cognition theorists are likely to exclusively focus on simple effects within participant sex (e.g., Carpenter, 2012), ignore the significant Participant Sex × Infidelity type interaction, and conclude that there are no sex differences in attitudes merely because both men and women evaluated sexual infidelity more negatively than emotional infidelity. This interpretation, in effect, demands a disordinal (i.e., crossover) interaction as evidence for the evolutionary perspective and ignores the actually obtained, statistically significant result: Men reported a significantly larger difference in their attitudes toward the two forms of infidelity. To date, social-cognitive theorists have not yet addressed the empirically robust Participant Sex × Infidelity Type interaction (Sagarin et al., 2012). At the very least, the social-cognitive perspective is incomplete without an acknowledgement of the robust, albeit ordinal, Participant Sex × Infidelity Type interaction.
It is important to acknowledge that the pattern of relative sex differences differed across our three outcome measures. On the forced-choice distress item, the large majority of women chose emotional infidelity as the more distressing form of partner infidelity, whereas men were approximately evenly split in their choice. In contrast, on the continuous distress scales, men reported significantly more distress to sexual infidelity than they did to emotional infidelity, but women's ratings of the two forms of infidelity were similar. On the attitude scales, both men and women evaluated sexual infidelity more negatively than emotional infidelity, but the difference was larger for men. Nonetheless, it is important to again emphasize that each of the outcomes was in the evolutionary theory supportive direction: Men, relative to women, reacted more negatively to partner sexual infidelity than to emotional infidelity. Critically, these results are thus fully consistent with both the Carpenter (2012) and Sagarin et al. (2012) meta-analyses. As such, we believe the present results provide strong evidence for the evolutionary perspective.
But, why the different patterns? Although the specific additional influences are unclear and should be investigated in future studies, it is logical to conclude that there are also situational and individual difference factors that influence reaction to partner infidelity. In any given study, some of these factors may increase or decrease the degree to which men's reactions differ from women's. But as clearly argued by Sagarin et al. (2012), selection pressures were influenced by reproductive competitions that were intrasexual, and as a result the evolutionary perspective does not imply that men's evolved sensitivities to sexual infidelity are necessarily always greater than women's responses to sexual infidelity. Likewise, women's evolved sensitivities to emotional infidelity are not necessarily always greater than men's. Rather, selection pressures altered the relative patterns of responses to both forms of infidelity, produced in different situations. Sagarin et al. (2012) note that “it would be foolish to think that sexually dimorphic selection pressures comprised the only factor impacting responses” (p. 597). Much like the now cliché nature-nurture debate, entrenched “either-or” debates concerning the veracity of evolutionary versus social-cognitive explanations are disruptive to progress. Instead, more research attention should be devoted to understanding the factors that sometimes lead to men and women reacting with more similarity to a given form of partner infidelity. Based on previous findings, possible moderators include participant age, socioeconomic status, household size, and the finding that sex differences are largest among U.S. college student samples (Carpenter, 2012; Sagarin et al., 2012; Zengel, Edlund, and Sagarin, 2013).
Focusing more on the unique contribution of the present study with an eye toward future research, the results indicate that an attitude approach may prove useful. As overall evaluative tendencies, attitudes are reflective of not only affective, but also cognitive, and behavioral experiences. Arguably then, measuring attitudes may be both a more thorough and efficient approach to the study of sex differences in reactions to partner infidelity. A limitation at this point is the inability to determine the basis of sex differences in attitudes. It is possible, for example, that the attitude differences reported here are mostly the result of emotional responses to partner infidelity. However, as shown in Table 1, we found attitudes toward sexual and emotional infidelity to be only moderately correlated with responses to the corresponding continuous measures of distress (rs = -.37, -.40), and very weakly correlated with choices on the forced choice distress item (rs = -.04, .10). Thus, we speculate that attitudes toward partner infidelity would be better predicted by a combination of emotions, cognitions, and behaviors. Moreover, perhaps it is the cognitive and behavioral components of attitudes that account for the finding that men and women both report more negative attitudes toward partner sexual infidelity than toward emotional infidelity. Specifically, attributions for infidelity (cognition) are likely correlated with attitudes toward partner infidelity, but men and women may differ in the circumstances that lead to given attributions. Given that previous research has shown older, infidelity-experienced adults respond differently to infidelity distress measures (Sagarin et al., 2012; Tagler, 2010), it would be interesting to examine how previous relationship experiences relate to men and women's attitudes. Particularly powerful would be longitudinal research that examines reactions to partner infidelity over time (Kimeldorf, 2009), to identify the degree to which evolutionary-predicted sex differences in attitudes persist and the affective, cognitive, and behavioral factors that predict change.
A final finding of the present study that deserves comment is that responses to the continuous distress scales were rather weakly correlated with choices on the forced-choice distress item (rs = .10, -.20). We sought to compare these results with those of previous studies, but unfortunately few have reported the correlation between forced-choice and continuous infidelity distress measures. One exception is Tagler (2010) who, in contrast, found moderate-sized correlations (rs = .47, -.48). However, a more recent study by Tagler and Gentry (2011) found relationships more similar to the present results (rs = .18, -.16). We do not at the present have an explanation for these discrepant findings, but to put them in the proper context it is important to understand that the continuous and forced-choice infidelity distress measures differ in an important way beyond mere question format. Whereas the forced-choice question requires participants to make a single, relative judgment regarding sexual versus emotional partner infidelity, the continuous scales are designed to separately assess reactions to the two forms of infidelity, in effect allowing participants to report similar or even identical levels of distress to both. It is only in the context of a data analytic procedure that the researcher formally assesses the relative difference in reactions to sexual versus emotional infidelity on continuous scales. Given this logic, arguably a more appropriate approach is to first quantify the differences in responses to the continuous distress scales (sexual – emotional), and then correlate these difference scores with choices on the forced-choice question. When doing so in the present study, we found a larger, statistically significant relationship, r(323) = .25, p < .001. This result is similar to that recently reported (r = .33) by Tagler and Gentry (2011). More data is needed to understand the unique properties of continuous and forced-choice measures of reactions to partner infidelity, and their relationship.
In conclusion, the present research was the first to document relative sex differences in attitudes toward partner infidelity. The pattern of this overall evaluative sex difference is consistent with the evolutionary account of sexually dimorphic selection pressures, and also with previous research focusing on emotional reactions to partner sexual versus emotional infidelity. Continued research on attitudes toward partner infidelity has the potential to increase our understanding of sex differences by focusing on the affective, cognitive, and behavioral differences in men and women's reactions to partner infidelity.