The impact of social interaction on the sexual double standard

Abstract

It is widely held that a sexual double standard exists such that women are evaluated more harshly than men for engaging in sexual activity. Previous research, however, has failed to document this sexual double standard reliably. We argue that previous research has been unable to identify the double standard because it has focused on the individual rather than the interpersonal dynamics that take place in social settings. The present experiment examines the hypothesis that group dynamics give rise to the sexual double standard. Participants, both individually and in small collaborative groups, evaluated a male or female target that had 1, 7, or 19 sex partners. A double standard did not emerge when individual participants evaluated targets. However, when collaborative groups of participants evaluated the targets, a double standard emerged in some domains. The results highlight the value of studying interpersonal processes in a group context.

In contemporary Western society, it is widely believed that men and women are evaluated differently for engaging in identical levels of sexual activity (Milhausen & Herold, 2001). Specifically, men with many sexual partners are believed to be evaluated in a positive fashion, whereas women with many partners are believed to be evaluated in a negative fashion. The differential evaluation of sexually active men and women has been called the sexual double standard.

The sexual double standard has received a lot of attention from contemporary critics of Western culture (Lamb, 2002; Tanenbaum, 2000; White, 2002). Tanenbaum (2000), for example, has documented the harassment and distress experienced by teenage girls who have been branded with derogatory sexual terms by their peers. Other writers have critiqued the way the media helps to create and reinforce negative stereotypes of sexually active women (Waggett, 1989), and how these stereotypes may contribute to violence against women (Malamuth & Check, 1981). Given the potency of the sexual double standard in contemporary discourse, one would assume that behavioral scientists have documented the double standard extensively and have identified many of the mechanisms that generate and sustain it. Despite much systematic research, however, there is only weak evidence for the existence of this allegedly pervasive phenomenon.

In this article we argue that previous attempts to document and explain the sexual double standard have been missing a potentially important ingredient—genuine social interaction. We also offer a new framework for conceptualizing the double standard—one that emphasizes the fundamental role of interpersonal processes in generating the double standard—and report an experiment designed to test some hypotheses derived from this framework. We begin with a brief review of previous research on the double standard.

EMPIRICAL RESEARCH ON THE SEXUAL DOUBLE STANDARD

The sexual double standard seems to be a ubiquitous phenomenon in contemporary society. A recent survey revealed that 85% of people, regardless of their own personal values, believe that a double standard exists in Western culture (Marks, 2002). The double standard has also been a fairly popular subject for researchers over the last few decades. Since the mid 1970s there have been over 30 studies that in some way address the double standard.

Although the sexual double standard seems pervasive, empirical research seldom shows that people evaluate sexually active men and women differently. In fact, much of the literature reveals little or no evidence of a double standard as it has traditionally been conceptualized (Gentry, 1998; Jacoby & Williams, 1985; Marks & Fraley, 2005; Oliver & Sedikides, 1992; O'Sullivan, 1995; Sprecher, 1989; Sprecher, McKinney, Walsh, & Anderson, 1988; Sprecher, Regan, McKinney, Maxwell, & Wazienski, 1997). O'Sullivan (1995), for example, conducted a person perception experiment in which individual participants read vignettes of a male or female target who had a high or low number of past sexual partners. Targets were then evaluated in domains such as likeability and morality. Both men and women who had more partners were evaluated more negatively than those with fewer partners, but a double standard was not found (i.e., both genders were derogated equally for engaging in frequent sex). Gentry (1998) also employed a person perception task and found that raters judged both male and female targets who were presented as having an above average amount of sexual activity in a more negative fashion than targets presented as having a lower amount of sexual experience. Again, no evidence of a double standard was found. In another person perception experiment, Marks and Fraley (2005) found that, even in the absence of valenced sexual information (e.g., insinuating that having many partners is a negative characteristic), people tended to derogate both genders equally for having many sexual partners.

Although many studies provide little support for the double standard, there are some studies that have reported evidence for it. For instance, Garcia (1982) found that, compared to androgynous respondents, sex‐typed respondents (i.e., masculine males and feminine females) displayed a slight preference for female targets in the low‐sexual‐experience condition, whereas sexual experience did not affect preference for high‐ or low‐experience male targets. Sprecher and her colleagues (Sprecher & Hatfield, 1996; Sprecher, McKinney, & Orbuch, 1991) report that engaging in casual sex has more deleterious effects on evaluations of women than of men. However, other studies have not shown this effect (Mark & Miller, 1986; O'Sullivan, 1995). Sprecher et al. (1991) also reported a double standard conditional on target age, but again this effect has failed to manifest in other studies (Sprecher, 1989; Sprecher et al., 1988).

In summary, although it appears that people do evaluate a target according to the amount of sexual activity that target has experienced, research rarely shows that people evaluate men and women differently as a function of their sexual activity. Even when interactions between target sex and sexual activity have been found, they are found in narrowly defined situations and are not easily replicated. If the sexual double standard is as pervasive and powerful as many people believe, empirical evidence for its role in person perception research should be less elusive.

GROUP DYNAMICS AND SOCIAL INFLUENCE

One conclusion that can be drawn from the existing literature is that the sexual double standard does not exist. Another possibility, however, is that behavioral scientists have failed to tap it properly. Commonly used paradigms for studying the sexual double standard may not successfully recreate situations in which the double standard manifests itself. One factor that all of the previously reviewed person perception studies have in common is a focus on evaluations made by individuals as opposed to peer groups. This may be a fundamental limitation of existing research because anecdotal accounts of the sexual double standard, such as adolescent boys bragging about their sexual conquests in locker rooms or girls gossiping about the reputations of highly sexually active females, reveal an important social component to the phenomenon. The double standard appears to emerge in the context of interactions among groups of people, such as friends and peers. The potential power of the group was articulated well by White (2002), who wrote, “bring [a sexually active woman] up among people who don't know one another well, and instantly everyone finds themselves on common ground” (p. 13).

Why should people behave in a way that is consistent with the double standard in social contexts but not in individual contexts (i.e., when evaluating a target in absence of peers, as in the case of standard person perception paradigms)? According to social psychological theory, people tend to attribute to others attitudes that are consistent with social norms (Allport, 1924; Miller & Prentice, 1994). In our society, women who frequently engage in sexual activity are not held in as high esteem as sexually active men (Marks, 2002; Milhausen & Herold, 2001). Thus, when people discuss a sexually active man while in a social context, they may assume that other group members hold the man in high regard. Similarly, when discussing a sexually active woman while in a social context, they may assume that other group members hold her in low regard. This may result in the individual using the perceived group norm as a reference point on which to base his or her evaluation (Allport, 1962).

Although this framework suggests that perceived norms may lead to behavior consistent with the sexual double standard, it does not specify the precise ways in which individual attitudes are affected by group processes. It may be the case that people will adopt the double standard in social contexts in order to maintain a façade of consistency with perceived group norms. In an attempt to adhere to perceived group norms (Asch, 1955, 1956; Reitan & Shaw, 1964), people may come to behave in ways consistent with the double standard, despite the fact that their evaluations do not necessarily reflect the kinds of judgments they would have made outside of the group context. When in a group, social norms (and the rewards and punishments for acting consistently or inconsistently with those norms) are likely to be more salient than when alone. Through observational learning, people learn that publicly violating norms can lead to severe social sanctions. For example, males who do not openly acknowledge or discuss their sexual activity may be teased about their sexual orientation, and females whose sexual activities are made public may be given derogatory labels. Conversely, those who publicly act in a manner consistent with norms may be rewarded. For instance, men who have a reputation for their many sexual “conquests” may be admired or be given high status in their peer groups, whereas women who refrain from discussing their sexual activity may be seen as “good girls.” Thus, people may conform to group norms to gain rewards or avoid punishments, and, in turn, hold others to those same norms.

Consistent with this interpretation, gender differences in self‐reported sexual behavior are greatest (i.e., men report high levels of sexual activity and women report low levels of sexual activity) when participants are made aware that their responses may be seen by a peer (Alexander & Fisher, 2003). This suggests that self‐reports of sexual behavior reflect responses influenced by normative expectations for males and females. If people act in a manner consistent with a double standard in order to conform to a perceived norm, the attitudes expressed in a group context will not necessarily be internalized (i.e., privately accepted as being correct).

It may also be the case, however, that the task of evaluating unfamiliar sexually active people is an inherently ambiguous task; one that leads people to look to other group members for relevant information. Because the way we evaluate people has important implications for our social lives (Gilbert, 1998), people want to be able to interpret ambiguous stimuli in the “right” way. In other words, person perception can be viewed as a judgmental task (McGrath, 1984), but one that people approach as an intellective one (Kaplan & Miller, 1987). While in a group, people may try to infer what others are thinking about the person being evaluated. People may then discuss possible attributes of the target consistent with what they perceive others are thinking, reinforcing other group members' inferences. In such situations, shared consensus may provide grounds for genuine belief, leading people to internalize the consensus reached by the group (Jacobs & Campbell, 1961; Sherif, 1935).

In summary, although individuals may not endorse a double standard, individuals in social groups may do so because they assume that others will use evaluative standards in accordance with the double standard and, thus, may feel compelled to use those standards themselves.

OVERVIEW OF THE PRESENT EXPERIMENT

Our primary objective in this research was to test the hypothesis that the sexual double standard is a function of group dynamics. In this experiment, we had participants both individually (i.e., by him or herself) and in groups evaluate a target that was either male or female and who reported a varying number of sexual partners. Targets were evaluated in each of four evaluative domains that, according to various theoretical perspectives, should be characterized by a sexual double standard (i.e., a domain in which sexually active men and women are evaluated differently). The first domain, values, concerns such qualities as honesty, morality, and trustworthiness. Girls and women have traditionally been severely restricted in their sexual opportunities in order to keep them “pure” and “innocent.” Boys and men, on the other hand, have usually been encouraged to engage in sex as for reasons unrelated to values (Mosher & Thompson, 1988). The second domain, dominance, concerns qualities such as assertiveness and leadership. According to social role theory (Eagly, 1987), women adopting an assertive or dominant role may be seen as violating role expectations, whereas men adopting an assertive or dominant role may be seen as acting appropriately. The third domain, success, concerns qualities such as status and monetary wealth. According to evolutionary theory, men with high status and plentiful resources are good mate choices due to their ability to provide resources for partners and offspring (see Buss, 1994). Thus, there may be evolved associations between male promiscuity and success. Promiscuous women, however, may be seen as engaging in sexual activity as to accrue resources from several men, perhaps signaling that they are incapable of accruing their own resources. Finally, the domain of intelligence concerns qualities such as mental ability and decision making. While sexual activity can have negative outcomes for men and women (e.g., sexually transmitted diseases), the consequences for women may be more numerous than those for men (namely unwanted pregnancies and a damaged reputation). Thus, women engaging in sex with multiple partners may be seen as less intelligent than men engaging in identical behavior.

In the present experiment, participants initially evaluated a target individually. Participants were then placed in groups of three and asked to complete evaluations of the same target collaboratively. Finally, to see if attitudes expressed in the group were internalized (i.e., accepted as correct), participants completed the evaluations again individually in a post‐test condition.

Based on previous research (e.g., Marks & Fraley, 2005; O'Sullivan, 1995), it was not expected that a sexual double standard would arise in the baseline condition (i.e., when the initial individual evaluations take place). However, based on the theoretical framework presented earlier, we predicted that a double standard would arise in the group portion of the experiment. Attitudes expressed in the group context, however, may or may not reflect people's genuine beliefs. If people's attitudes are truly being shaped by group processes, the attitudes expressed in the group should carry over to the post‐group condition. If people are simply conforming to perceived group norms, however, then we should not expect them to endorse attitudes consistent with the sexual double standard in the post‐group condition.

It is important to note that social motives for change can result in attitude change in private as well as in public (see Wood, 2000, for a review). Therefore, we administered several questions at the end of the experiment concerning the group interaction. These items were included to assess the general climate in the group interaction, including how influenced the participants felt and how pressured they felt to agree. Whereas publicly endorsing an evaluation that one does not privately agree with may be rather uncomfortable, publicly endorsing an evaluation one privately accepts should feel comparatively easy. Should attitudes expressed in the group carry over to the post group condition, these items should allow us to better gauge the kind of influence that took place during the group discussion.

METHOD

Design

The independent variables in this experiment were target gender, target sexual experience (1, 7, or 19 partners), and the social context of the evaluations (pre‐group, group, and post‐group). (See footnote 1 for information regarding participant sex as a predictor.) The dependent variables were participants' evaluations of the target in each of four domains: values, dominance, success, and intelligence.

Participants

Approximately 500 participants would be needed to have a power of .80 to detect the critical two‐way interactions between target sex and number of sexual partners that would indicate a sexual double standard. An interaction between target sex and partners would indicate that the ratings given to targets with many (as opposed to few) partners depend on whether the target is a man or woman. We were able to recruit 468 students (144 male, 324 female) from a large Midwestern university to participate for credit in their classes. The average participant age was 19.27 years (SD = 2.64, range 17–45). Approximately 32.3% of the participants were white, 23.1% Asian, 11.9% Hispanic, 9.2% African American, and 23.5% of other ethnicities or failing to specify ethnicity.

Procedure

The experiment was conducted in a classroom with one large main room and eight smaller rooms around the perimeter. Upon arrival for the experiment, participants were told that they would be participating in a person evaluation experiment where they would evaluate a person repeatedly in a variety of situations, in an effort to help researchers understand “evaluation processes and dynamics.” To avoid demand characteristics, no information regarding gender or sexual activity was mentioned. Participants were assured that their responses would be anonymous and would only be identified by an arbitrary identification number.

Participants were initially seated in the main part of the classroom, with at least two empty seats between each participant. To determine the participants' initial impressions of the target (what we will call the “baseline rating”), each participant read a vignette about the target and completed the evaluation questionnaire individually. Fifteen minutes were allotted to the baseline evaluation, which was sufficient for all participants. After each participant had filled out the evaluation, participants were randomly assigned to same‐sex groups of three, and each group entered one of the small perimeter rooms. The members of each group then collaboratively evaluated the target in a forced‐consensus task (i.e., participants had to agree on a single rating for each item in the questionnaire). Same‐sex groups were used to maintain the integrity of participant sex as a possible predictor.1 Twenty minutes were allotted to the collaborative evaluation, which was sufficient for all groups. As each group finished their evaluations, they exited the small room and returned to their original seat in the main room. When all groups had finished their evaluations, participants again completed the questionnaire individually.

After completing the final evaluation, participants were given a questionnaire containing questions about the group session, as well as several questions designed to be manipulation checks. Participants were then thanked and debriefed.

Materials

All materials were completed in pencil and paper format. The vignette consisted of five questions (counterbalanced for order) and handwritten answers to those questions purportedly written by an anonymous 19‐year old individual (see Appendix A). The age and gender of this target person were listed at the top of the vignette. All other demographic/identifying information about the target (ID number, ethnicity) was blacked out. Participants were told that the questions were from a survey issued to the general public, and the answers were written by an anonymous person who consented to having his or her answers used in another experiment. Targets were male or female and reported having 1, 7, or 19 sexual partners. Information concerning the target's sexual activity was conveyed in response to the question, “What is something someone may not know about you?”

After reading the target's answers, participants rated the target on 22 evaluative items (see Appendix B). Participants rated each item on a 5‐point scale (1 = Strongly Disagree, 5 = Strongly Agree). For our analyses, items were grouped into four subscales based on the domains discussed previously. The first subscale was labeled Values (α = .85, .83, and .86, for baseline, group, and post‐group sections, respectively), and consisted of nine items, such as “This person is trustworthy” and “This person is immoral.” The second subscale was labeled Dominance (α = .72, .73, and .81), and contained five items, such as “This person often takes control of situations” and “This person would make a good leader.” The third subscale was labeled Success (α = .64, .41, and .68), and contained three items, including “This person is successful” and “This person will make a lot of money.” The last subscale was labeled Intelligence (α = .69, .61, and .75), and consisted of five items such as “This person is intelligent” and “This person did well in school.”

Six questions concerning the group interaction were administered at the end of the post‐group session. Participants were asked how much they liked the group, how much they felt the group liked them, how influenced they felt, how much they felt they influenced the group, how pressured they felt to agree, and how much the group interacted. Participants rated each item on a 7‐point scale (1 = Not at all, 7 = A lot).

Six questions regarding the vignette were administered following the group interaction questions and served as manipulation checks. One item asked participants to report the gender of the target. The other five consisted of one question about each section on the vignette. Of these five questions, only the critical question, “How many sexual partners did the person have?”, was used along with the target gender question to ensure the participants had paid adequate attention to the vignette. All participants answered these two questions correctly.

RESULTS

Hierarchical regression analyses2 were performed separately for each of the four evaluative domains within each of the three conditions (i.e., baseline, group, post‐group). The first block contained target sex and the number of partners reported by the target. The second block contained the interaction between target sex and partners. An alpha level of .05 was used for all statistical tests.

Domain of values

In the baseline condition (i.e., the first individual session), females were rated as being more value oriented (i.e., trustworthy, moral, honest; M = 3.46) than males (M = 3.30; β = .11, p<.05), and targets were rated as being less value oriented as the number of sexual partners increased (M = 3.78, 3.30, & 3.08, for 1, 7, and 19 partners, respectively; β = −.41, p<.05). Introducing the interaction term did not result in an increase in the variance explained ΔR ² = .00, Fchange(1, 464) = 1.37, ns.

In the group condition, targets with more partners were rated as being less value oriented, (M = 3.70, 3.26, and 2.88, for 1, 7, and 19 partners, respectively; β = −.58, p<.05). There was no main effect of target sex. Introducing the interaction term did not result in an increase in the variance explained ΔR ² = .00, Fchange(1, 152) = 0, ns.

In the post‐group condition (i.e., the second individual session), females were rated as more value oriented (M = 3.34) than males, (M = 3.19; β = .11, p<.05), and targets with many partners were rated as being less value oriented, (M = 3.67, 3.23, and 2.91, for 1, 7, and 19 partners, respectively; β = −.47, p<.05). Introducing the interaction term did not result in an increase in the variance explained ΔR ² = .00, Fchange(1, 464) = 0.32, ns. Table 1 reports all coefficients and simple slopes. See Figure 1 for simple slopes in all three conditions.

Table 1. Summary of hierarchical regression analysis for primary variables predicting evaluations in the values domain

		B	SE B	β	ΔR^2
Condition	Predictor
Baseline	Block 1				.18*
	Partners	−.04	.00	−.41*
	Target Sex	.15	.05	.11*
	Block 2				.00
	Target Sex×Partners	.03	.03	.05
	Simple Slopes
	Male	−.04	.01	−.43*
	Female	−.03	.01	−.39*
Group	Block 1				.35*
	Partners	−.04	.01	−.58*
	Target Sex	.10	.07	.09
	Block 2				.00
	Target Sex×Partners	.00	.04	.00
	Simple Slopes
	Male	−.04	.01	−.56*
	Female	−.04	.01	−.62*
Post‐group	Block 1				.24*
	Partners	−.04	.00	−.47*
	Target Sex	.14	.05	.11*
	Block 2				.00
	Target Sex×Partners	−.01	.03	−.02
	Simple Slopes
	Male	−.04	.01	−.45*
	Female	−.04	.01	−.51*
*p<.05.

Figure 1. Evaluations of values as a function of target sex and the number of sexual partners reported by the target.

Domain of dominance

In the baseline condition, there were no effects of target sex or partners, nor an interaction between them. In the group condition, targets with more partners were rated as more dominant, (β = .19, p<.05). Introducing the interaction term led to an increase in the amount of variance explained, ΔR ² = .03, Fchange(1, 152) = 4.97, p<.05. The interaction (β = −.32, p<.05) was such that male targets were rated as more dominant as the number of sexual partners increased (M = 2.33, 2.45, and 2.79, for 1, 7, and 19 partners, respectively; β = .35, p<.05), whereas female targets were not (M = 2.59, 2.38, and 2.58, for 1, 7, and 19 partners, respectively; β = .02, ns).

In the post‐group condition, targets with more partners were rated as more dominant, (M = 2.56, 2.52, and 2.75, for 1, 7, and 19 partners, respectively; β = .17, p<.05). There was no main effect of target sex. Introducing the interaction term led to an increase in the amount of variance explained, ΔR ² = .03, Fchange(1, 464) = 14.39, p<.05. The interaction (β = −.17, p<.05) was such that male targets were rated as more dominant as the number of sexual partners increased (M = 2.44, 2.53, and 2.87, for 1, 7, and 19 partners, respectively; β = .35, p<.05), whereas female targets were not (M = 2.66, 2.52, and 2.63, for 1, 7, and 19 partners, respectively; β = .02, ns). Table 2 reports all coefficients and simple slopes. See Figure 2 for simple slopes in all three conditions.

Table 2. Summary of hierarchical regression analysis for primary variables predicting evaluations in the dominance domain

		B	SE B	β	ΔR^2
Condition	Predictor
Baseline	Block 1				.00
	Partners	−.00	.00	−.02
	Target Sex	.02	.05	.02
	Block 2				.01
	Target Sex×Partners	−.04	.03	−.07
	Simple Slopes
	Male	.03	.04	.05
	Female	−.05	.04	−.09
Group	Block 1				.04
	Partners	.01	.01	.19*
	Target Sex	−.01	.08	−.01
	Block 2				.03*
	Target Sex×Partners	−.09	.04	−.18*
	Simple Slopes
	Male	.20	.06	.35*
	Female	.01	.06	.02
Post‐group	Block 1				.03*
	Partners	.01	.00	.16*
	Target Sex	−.01	.05	−.01
	Block 2				.03*
	Target Sex×Partners	−.10	.03	−.17*
	Simple Slopes
	Male	.19	.03	.34*
	Female	−.01	.03	−.01
*p<.05.

Figure 2. Evaluations of dominance as a function of target sex and the number of sexual partners reported by the target.

Domain of success

In the baseline condition, targets were rated as less successful as the number of partners increased (M = 3.24, 3.12, and 3.09, for 1, 7, and 19 partners, respectively; β = −.10, p<.05). There was no main effect of target sex. Introducing the interaction term did not result in an increase in the variance explained ΔR ² = .00, Fchange(1, 464) = 0.93, ns.

In the group condition, there were no main effects present. Introducing the interaction term led to an increase in the amount of variance explained, ΔR ² = .04, Fchange(1, 152) = 5.81, p<.05. The interaction (β = −.19, p<.05) was such that success ratings of male targets were not significantly affected by sexual experience (M = 3.05, 2.96, and 3.17, for 1, 7, and 19 partners, respectively; β = .13, ns), whereas those for females were negatively affected as the number of partners increased, (M = 3.29, 3.08, and 3.00, for 1, 7, and 19 partners, respectively; β = −.26, p<.05).

In the post‐group condition, targets with more sexual experience were rated as less successful (M = 3.17, 3.06, and 2.96, for 1, 7, and 19 partners, respectively; β = −.15, p<.05). There was no main effect or target sex. Introducing the interaction term led to an increase in the amount of variance explained, ΔR ² = .03, Fchange(1, 464) = 13.32, p<.05. The interaction (β = −.17, p<.05) was such that sexual experience had no effect on the success ratings of male targets (M = 3.09, 2.96, and 3.07, for 1, 7, and 19 partners, respectively; β = .01, ns), whereas female targets were rated as less successful as the number of partners increased, (M = 3.24, 3.15, and 2.85, for 1, 7, and 19 partners, respectively; β = −.33, p<.05). Table 3 reports all coefficients and simple slopes. See Figure 3 for simple slopes in all three conditions.

Table 3. Summary of hierarchical regression analysis for primary variables predicting evaluations in the success domain

		B	SE B	β	ΔR^2
Condition	Predictor
Baseline	Block 1				.01*
	Partners	−.01	.00	−.10*
	Target Sex	.07	.05	.06
	Block 2				.00
	Target Sex×Partners	−.03	.03	−.05
	Simple Slopes
	Male	−.03	.04	−.05
	Female	−.08	.04	−.15*
Group	Block 1				.01
	Partners	−.00	.01	−.06
	Target Sex	.06	.07	.07
	Block 2				.04*
	Target Sex×Partners	−.09	.04	−.19*
	Simple Slopes
	Male	.06	.05	.13
	Female	−.11	.05	−.26*
Post‐group	Block 1				.03*
	Partners	−.01	.00	−.15*
	Target Sex	.04	.05	.04
	Block 2				.03*
	Target Sex×Partners	−.09	.02	−.17*
	Simple Slopes
	Male	.01	.04	.01
	Female	−.17	.03	−.33*
*p<.05.

Figure 3. Evaluations of success as a function of target sex and the number of sexual partners reported by the target.

Domain of intelligence

In the baseline condition, female targets were rated as being more intelligent (M = 3.55) than male targets (M = 3.40; β = .12, p<.05), and targets were rated as being less intelligent as the number of sexual partners increased, (M = 3.71, 3.42, and 3.32, for 1, 7, and 19 partners, respectively; β = −.27, p<.05). Introducing the interaction term did not result in an increase in the variance explained ΔR ² = .00, Fchange(1, 464) = 0.21, ns.

In the group condition, targets were rated less intelligent as the number of partners increased (M = 3.70, 3.46, and 3.28, for 1, 7, and 19 partners, respectively; β = −.34, p<.05). There was no main effect of target sex. Introducing the interaction term resulted in a slight increase in the variance explained ΔR ² = .02, Fchange(1, 152) = 3.11, p = .08. The interaction (β = −.13, p = .08) was such that both males and females were rated as less intelligent as sexual partners increased, but the derogation was more severe for females (M = 3.79, 3.46, and 3.23, for 1, 7, and 19 partners, respectively; β = −.49, p<.05) than for males (M = 3.60, 3.41, and 3.32, for 1, 7, and 19 partners, respectively; β = −.23, p<.05).3

In the post‐group condition, targets were rated less intelligent as the number of partners increased (M = 3.74, 3.43, and 3.25, for 1, 7, and 19 partners, respectively; β = −.36, p<.05). There was no main effect of target sex. Introducing the interaction term led to an increase in the amount of variance explained, ΔR ² = .01, Fchange(1, 464) = 4.68, p<.05. The interaction (β = −.09, p<.05) was such that both males and females were rated as less intelligent as sexual partners increased, but the derogation was more severe for females (M = 3.77, 3.54, and 3.20, for 1, 7, and 19 partners, respectively; β = −.46, p<.05) than for males (M = 3.69, 3.32, and 3.30, for 1, 7, and 19 partners, respectively; β = −.26, p<.05). Table 4 reports all coefficients and simple slopes. See Figure 4 for simple slopes in all three conditions.

Table 4. Summary of hierarchical regression analysis for primary variables predicting evaluations in the intelligence domain

Variable		B	SE B	β	ΔR^2
Condition	Predictor
Baseline	Block 1				.09*
	Partners	−.02	.00	−.27*
	Target Sex	.13	.05	.12*
	Block 2				.00
	Target Sex×Partners	−.01	.02	−.02
	Simple Slopes
	Male	−.02	.01	−.24*
	Female	−.02	.00	−.30*
Group	Block 1				.14*
	Partners	−.02	.01	−.37*
	Target Sex	.06	.07	.07
	Block 2				.02
	Target Sex×Partners	−.06	.03	−.13
	Simple Slopes
	Male	−.01	.01	−.24*
	Female	−.03	.01	−.49*
Post‐group	Block 1				.13*
	Partners	−.03	.00	−.36*
	Target Sex	.07	.05	.06
	Block 2				.01*
	Target Sex×Partners	−.05	.02	−.09*
	Simple Slopes
	Male	−.02	.01	−.26*
	Female	−.03	.00	−.46*
*p<.05.

Figure 4. Evaluations of intelligence as a function of target sex and the number of sexual partners reported by the target.

To summarize, evaluations in the baseline condition were not consistent with a sexual double standard. In the group condition, a double standard emerged in the domains of dominance, success, and intelligence. No evidence of a double standard appeared in the domain of values during the group condition. In the post‐group conditions, the patterns of evaluations were very similar to those in the group conditions.

Measures of group influence

In order to tap the dynamics taking place in the group condition, participants were asked six questions about their perception of the group's interaction. Participants reported a high degree of interaction within the group (M = 5.55, SD = 1.25), feeling liked by other group members (M = 5.01, SD = 1.07), liking their group members (M = 5.32, SD = 1.13), feeling they moderately influenced their group (M = 4.71, SD = 1.16), feeling moderately influenced by the group (M = 4.37, SD = 1.26, modal response = 5), and feeling little pressure to agree with the group's consensus (M = 2.85, SD = 1.76, modal response = 1).

Responses to the six measures of perceived group influence were highly intercorrelated. Participants felt more influenced when they thought the group interacted more, (r = .43, p<.05), when they thought their fellow group members liked them (r = .15, p<.05), when they liked their fellow group members (r = .17, p<.05), and when they felt more pressure to agree (r = .22, p<.05). Feeling pressured to agree was negatively correlated with the amount of perceived interaction in the group (r = −.12, p<.05), how much participants felt liked (r = −.20, p<.05), and how much they liked their fellow group members (r = −.18, p<.05). Table 5 reports all means and correlations.

Table 5. Correlations between post‐group items concerning the interaction with other group members

Item	M	SD	1	2	3	4	5	6
1. We interacted a lot	5.55	1.25	1
2. They liked me	5.01	1.07	.43*	1
3. I liked them	5.32	1.13	.50*	.70*	1
4. I influenced them	4.71	1.56	.29*	.39*	.28*	1
5. They influenced me	4.37	1.26	.24*	.15*	.17*	.26*	1
6. I felt pressure to agree	2.85	1.76	−.12*	−.20*	−.18*	−.04	.22*	1
*p<.05

We also explored the possibility that one or more of these group processes (liking, felt influence, etc.) mediated the findings of the sexual double standard in groups. Although a traditional mediational analysis (e.g., that described by Baron & Kenny, 1986) was not a feasible option due to unequal degrees of freedom between the group process data and the ratings in the group condition, we did examine the correlation between the group process data and group and post‐group measures. Significant correlations between these data would indicate that a mediational relationship exists. The only significant correlations occurred in the values domain, where pressure to agree was correlated with group and post‐group evaluations (rs = −.16 and −.10, respectively, ps<.05). As the values domain was the only one in which we found no evidence of a double standard, it is unlikely that group processes (or at least the ones we measured) mediate the findings reported here.

DISCUSSION

The purpose of this research was to test the hypothesis that the sexual double standard is due, in part, to social interaction. By testing people both individually and in groups, we were able to examine the effect that being in a small group had on evaluations of sexually active male and female targets. If the double standard is a product of social influence, then people in groups should have evaluated highly sexually active female targets more harshly than sexually active male targets. Whether or not these evaluations were internalized was examined by (a) studying target evaluations once the influence of the group was removed, and (b) studying perceptions of the climate in the group discussion.

Consistent with previous research (e.g., Gentry, 1998; Marks & Fraley, 2005; O'Sullivan, 1995), the double standard did not emerge when participants evaluated targets in a non‐group context. However, when evaluating targets in a group context, people did exhibit a double standard in the domains of dominance, success, and intelligence. In the domain of dominance, groups exhibited a double standard such that evaluations of male targets were positively related to sexual experience, whereas evaluations of female targets were unrelated to sexual experience. This effect contrasts with the result obtained in the baseline condition, where evaluations of neither male nor female targets were related to sexual experience. This finding may be related to pervasive gender norms. Men are expected to be assertive, controlling, and powerful, whereas women generally are not expected to behave in these ways (Cejka & Eagly, 1999; Glick, 1991; Lips, 1994). In the group condition, these gender stereotypes may have become more salient, resulting in the emergence of a positive relationship between sexual experience and dominance for male targets.

In the domain of success, groups exhibited a double standard such that evaluations of male targets were positively (albeit not significantly) related to sexual experience, whereas evaluations of female targets were negatively related to sexual experience. This finding may also be related to gender stereotypes. Being in a social context may have resulted in evocations of a woman who pursues sexual relationships for material gain. For instance, upon their arrival to a professional‐level sports league, male athletes are warned of women seeking sexual relationships for monetary gain (Guschov, 1998).

In the intelligence domain, both male and female targets were seen as less intelligent as the number of sexual partners increased. This is not surprising, considering the focus on sexually transmitted diseases in contemporary Western society—engaging in frequent casual sex may not be considered a “smart” thing to do. However, in the group interaction, participants penalized female targets slightly more than male targets for having many sexual partners. This may be because the consequences of sex for females, such as pregnancy or a poor reputation, became more salient in the group discussion.

The only domain to show no effect of social interaction on the evaluation of sexually active men and women was that of values. It may be the case that participants treated promiscuous sex primarily as a moral issue, resulting in the evaluation of excessively sexually active targets as immoral regardless of gender. It may also be the case that different values (albeit negative ones) are applied to sexually active men and women (i.e., sexually active men are manipulative, sexually active women are deceptive or easy). In any case, this finding suggests that the double standard may not be applicable to all areas in which people may be evaluated.

When the pressure to conform was removed, double standards that emerged in the group condition persisted into the post‐group condition. In other words, when participants had the opportunity to evaluate the target free from the observation of peers, they evaluated targets in accordance with a sexual double standard in most domains. There are several possible explanations for this finding. The first is that informational influence (Jacobs & Campbell, 1961) was occurring in the group condition. Participants may have been ambivalent about the accuracy of their initial evaluations and, during the group session, looked to others for cues about how to evaluate the target. People may have taken those cues as evidence about the target's personality and, when in the post‐group condition, provided evaluations consistent with the group's. This interpretation is strengthened by the fact that participants who felt liked by the group were more likely to report feeling influenced by the group's input, and less pressured to agree with the group. Feeling liked without feeling pressured indicates that the participants were naturally accepting the group's input, not that they were going along with the group's evaluations despite not agreeing with them.

A second possibility is that normative influence (Asch, 1955) was occurring in the group condition. As noted previously, enduring attitude change does not necessarily suggest that informational influence has occurred. Bond and Smith (1996) suggest that social motives for agreement can affect private as well as public attitudes. It is possible that pressure to conform to social norms was the driving force behind the double standard's emergence in most domains of the group condition. However, the fact that participants in the present experiment indicated feeling little pressure to agree with other group members makes the normative influence explanation less feasible.

A third possibility is that a kind of carry‐over effect occurred, in which participants simply anchored their post‐group responses with respect to those given most recently (i.e., those in the group session). Since the time lag between the group and post‐group condition was very short, it is possible that the group's evaluation was readily accessible, and participants evaluated the target accordingly. This possibility can be tested in future research by varying the time lag between the group and post‐group sessions. Allowing more time (e.g., a day or a week) between conditions would allow for a more rigorous test of the recency hypothesis. If the post‐group evaluations are still similar to the group's after a longer time lag between sessions, the informational influence interpretation would be strengthened. However, since participants generally reported feeling influenced by the group discussion, the carry‐over explanation is questionable.

In any case, the results of the present experiment exemplify the power of a social context on judgments made about sexually active others. The within‐subjects design employed here required participants to evaluate a specific target before joining a group—that initial evaluation should have made it harder for the group context to have an effect relative to a between subjects design. Thus, the effect sizes reported here might be even larger when the target is unfamiliar (i.e., has not yet been evaluated) by anyone in the group.

Limitations of the present experiment

There are some limitations of this research that should be kept in mind while evaluating the findings. First, only groups consisting of same‐sex participants were used. This was done to maintain the integrity of participant sex as a possible predictor variable. Although no major effects of participant sex were found (see footnote 1), any effect caused by different male‐to‐female ratios in the groups would have been missed. Once the effects of group dynamics on the sexual double standard are more substantiated, using groups of mixed‐sex individuals will prove valuable for further understanding of this phenomenon.

Second, the extent to which the setting used in this experiment mirrors person perception and evaluation in social situations is ambiguous. In this experiment, participants interacted with strangers during the group condition. However, judgments have shown to be more susceptible to the influence of a stranger than an acquaintance (Ex, 1960). Thus, it is not clear as to whether the findings of the present experiment generalize to interactions with friends or acquaintances. Moreover, the task of evaluating a stranger may be quite different than evaluating an acquaintance. Because less is known about strangers, it may be quite easy to attribute stereotypical characteristics to them. Thus, while the present experiment provides an interesting perspective on the effects of group dynamics on the sexual double standard, research using more familiar subjects or targets is required before stronger conclusions can be drawn.

Third, the dependent variables used in this experiment represent only a small percentage of evaluative domains in which perceptions of males and females may vary as a function of sexual experience. Future research may involve additional sets of dependent variables in order to efficiently target the effect that sexual experience has on evaluations. Domains relevant to the evolution of mating strategies (e.g., ambition, industriousness, and dependability; Buss, 1994) and gender norms (e.g., emotionality and intimacy; Zeman & Garber, 1996) may also be sensitive to group dynamics. Examining these domains in interpersonal settings may assist in substantiating the interpretations of the current experiment.

Implications and future directions for research

The present research sheds light on the contexts in which the double standard emerges. Before we can effectively address the problems stemming from the manner in which sexually active women are perceived, we must better understand the way in which the double standard works. The main implication of the present experiment is that social interaction is a primary ingredient in the sexual double standard. Participants in past research studies may very well have been aware of social norms or standards regarding sexually active people by virtue of having been involved in numerous group interactions of the kind presented in the present experiment. However, these participants may not have been motivated to apply those standards in a standard person perception research setting. In fact, when someone is asked to evaluate a person in a standard research setting, he or she may explicitly try to avoid applying stereotypical attributes to the target. In other words, participants' motivation may be to apply a fair and unprejudiced evaluation of the target person in question. By placing people in a group, their motivation may shift from one of fairness to one of accuracy or consensus with norms. Because social interaction is a pervasive part of daily life, studying the double standard in groups may provide an externally valid view of evaluations concerning sexually active men and women.

Second, the present experiment suggests that a process akin to informational influence may be taking place when sexually active people are evaluated in a social context. This may have broad implications for the sexual double standard because people tend to accept information from others as being factual when informational influence occurs. For example, witnessing women being derogated for sexual activity may serve to reinforce the concept of a double standard in an observer's mind, thus perpetuating it. In addition, the framework addressed here suggests that it would be valuable to study the sexual double standard in social settings. For example, observing “hot spots” where social interactions are possibly centered on sex (e.g., bars, locker rooms) may offer insight to the kinds of stereotypes commonly applied, social norms followed, language used, and attitudes expressed concerning the sexual activity of males and females.

Third, the double standard emerged most strongly in the dominance and success domains, two areas strongly associated with gender stereotypes. This suggests that the double standard is associated with pervasive gender stereotypes. Constructs like “sexual man” or “sexual woman” may not have evoked sexual stereotypes for participants in a non‐social context. However, in a social context, contemplating what other group members were thinking might have led these stereotypes to become salient for participants, making them more likely to be applied to the target if appropriate. Specifically, targets seen as adhering to or violating the “proper” sexual stereotype for his or her gender may have been attributed other stereotypical attributes as well (e.g., a sexual man is a dominant one; a sexual woman is unable to procure her own resources and must sleep with men to get them).

Another implication of this experiment is that the double standard may be more pervasive than previous research indicates. Thus far, the double standard has emerged in somewhat narrow contexts, delimited by specific characteristics of the target (e.g., Sprecher et al., 1991), context of sexual behavior (e.g., Sprecher & Hatfield, 1996), or characteristics of the respondent (e.g., Garcia, 1982). The results of this experiment suggest that being in a social situation is sufficient to override specific subject and respondent characteristics in producing a double standard. Moreover, these results suggest that, at least in the interim, social interaction may still exert its influence even when the individual is removed from the social situation.

This research also has implications for research on socially important phenomena more generally. Two decades ago, Carlson (1984) discussed the failure of social psychologists to study human behavior in genuine social contexts. However, the continued use of imagined and anticipated interactions is still very prevalent in contemporary social psychological research, whereas studies involving truly social contexts are comparatively rare. Although asocial methods can and do produce interesting findings, it is imperative to study socially important phenomena in interpersonal contexts as well. The importance of doing so is exemplified in the present experiment. When we studied people individually, we did not find evidence for the double standard. When we placed people in groups and had them collaboratively evaluate others, a double standard did emerge in some domains. Studying meaningful social issues in a genuine social interaction context may help reveal important effects not otherwise found by examining isolated individuals (Snyder, 1981).

In closing, we believe that it may be beneficial to further explore the double standard in social contexts. By doing so, future researchers should be able to explicate the disparity between popular intuitions and the research literature, while opening the doors to novel avenues for advancing our understanding of attitudes towards sexuality.

Notes

1. Preliminary analyses were conducted using participant sex as a predictor. In the post‐group condition of the success domain, women rated female targets more harshly than male targets as the number of partners increased. As there were no other meaningful effects of participant sex, we collapsed the data across participant sex.

2. We used multiple regression for two reasons. First, the unit of analysis in the individual conditions was different from the unit of analysis in the group condition (individual ratings vs collaborative group ratings, respectively), rendering a mixed‐model ANOVA an inappropriate means of analysis. Second, we were conceptualizing the number of sexual partners as coming from a continuous distribution rather than a categorical one. Exploratory analyses revealed no violations of the linearity assumption.

3. The marginally significant result is likely the result of diminished statistical power in the group condition, which had one‐third as many degrees of freedom as the individual conditions. As can be seen in Table 4, the pattern of results and effect sizes in the group condition of the intelligence domain are comparable to those in the post‐group condition, where the interaction between target sex and partners was significant.

APPENDIX A

ID No. [blacked out]

Ethnicity [blacked out]

Age 19

Sex [male or female]

1. Please describe some of your hobbies.

I enjoy watching movies. I go to the theater a couple times a week, and also rent lots of videos and DVDs. I also like to ski. I don't get to go very often because it's hard to find free weekends anymore, but I take every chance I get to go skiing.

2. What is something someone may not know about you?

I've had sex with [1/7/19 guy(s)/girl(s)]. I don't have much to say about it. It's just kind of the way I've lived my life. I don't talk about it much with others, so most people are not aware of this.

3. Describe some of your fears.

Well, as far as tangible things, I'm scared of snakes and reptiles like snakes. As far as intangible things, I fear failing at my goals and not being a success. The thought of not being able to do the things I want scares me.

4. Tell us about your career aspirations.

I would like to become a writer someday. I feel I have pretty good ideas, and I have a talent for putting those ideas on paper. I'm not sure if I'd like to be a journalist or a freelance writer, but I know I want to cover current events and how they affect things.

5. How do you see yourself?

I see myself as a theorist more than a sensor. I internalize a lot of things, and thing abstractly about them. I don't have to try and think abstractly about things, it just happened. That's about all.

APPENDIX B

Values:

This person would make someone a good husband/wife.

I could be friends with this person.

This person is dishonest.

This person is respectful.

I would not like to know this person.

This person would make someone a good boyfriend/girlfriend.

This person is careless.

This person is trustworthy.

This person is immoral.

Dominance:

This person would make a good leader.

This person often takes control of situations.

This person will hold a powerful position.

This person influences others.

This person will be in charge of many people.

Success:

This person is successful.

This person will make a lot of money.

This person will have a good job.

Intelligence:

This person is intelligent.

This person did well in school.

This person is a failure.

This person makes a lot of mistakes.

This person performs well in everything he/she does.