Sexual Fantasies About One's Partner Versus Someone Else: Gender Differences in Incidence and Frequency.

Sexual fantasies can provide important insight into some of the different scripts that underlie the sexual arousal and sexual behavior of men and women in our culture (Gagnon & Simon, 1973). Because sexual fantasies are private and do not depend on the participation of a partner, they may be more revealing of gender differences in sexuality than actual behavior (Ellis & Symons, 1990). If men and women do differ, at least to some extent, in what they find sexually exciting or in their attitudes toward sexuality, these differences may be more apparent if they are not diluted by the inevitable compromises that people make in the interest of maintaining a relationship. In addition, in fantasy one can imagine whatever one likes unconstrained by social convention, practical and legal barriers, or by fears of embarrassment, criticism, or rejection (Wilson, 1997). Therefore, fantasies may provide a clearer picture than behavior of what is truly erotic to men and women.

Whether gender differences in sexuality should be attributed to distal evolutionary factors or proximal sociocultural factors remains a controversial and contentious topic (Angier, 1999; Buss & Schmitt, 1993; Eagley & Wood, 1999). However, as Oliver and Hyde (1993) noted in their meta-analytic review of gender differences in sexuality, evolutionary psychology and sociocultural theories actually agree on a number of predictions. For example, both theories generally predict that, on average, women will be more cautious than men in choosing sex partners and less interested in sex for its own sake outside of any romantic or relationship context. Evolutionary psychology perspectives on sexuality suggest that, compared to women, men may be more primed to respond to opportunities for having sex with multiple partners outside the context of a relationship. Sociocultural theorists similarly argue that due to differences in socialization, social roles, and sexual scripts, women are more prone to inhibit sexual interest unless it occurs within the context of a committed relationship.

Prior research has revealed a number of consistent gender differences in sexual fantasy content that are congruent with both evolutionary psychology and sociocultural perspectives (see review by Leitenberg & Henning, 1995). Surprisingly, however, no previous studies have examined whether men and women differ in their likelihood of having sexual fantasies about their current partner as compared to fantasies about someone else (extradyadic fantasies). If men are indeed more likely than women to be aroused by sexual stimuli outside the context of a relationship, one would also predict that despite considerable within-gender variation, men on average would be more likely than women to have sexual fantasies that involve someone other than their current partner. In addition, it could be hypothesized that when women do fantasize about someone other than their current partner, they would be more likely than men to fantasize about a former partner with whom they have shared a close emotional connection in the past.

Because prior research has consistently shown that men report having more frequent sexual fantasies in total than women (Leitenberg & Henning, 1995), the present study will control for this response pattern in data analysis. Otherwise, any difference in the frequency of extradyadic fantasies could simply reflect the fact that men report more frequent sexual fantasies in general. We predict that even after controlling for the total frequency of sexual fantasies, men as compared to women will report that a larger proportion of their sexual fantasies involve someone other than their current partner.

We also controlled for two other behaviors that men engage in on average more often than women that might be related to having a higher proportion of extradyadic fantasies. Men tend to report that they have had a greater number of sex partners than women (e.g., Laumann, Gagnon, Michael, & Michaels, 1994), and more men than women report that they have engaged in actual extradyadic sexual behavior or "cheated" on their partners (e.g., Laumann et al., 1994; Thompson, 1983). Since prior experience can influence fantasy and vice versa (e.g., Eisenman, 1982; Malamuth, 1981), both of these variables were expected to be associated with an increased likelihood of having extradyadic sexual fantasies. The present study, therefore, both controls for these variables when examining gender differences in the proportion of extradyadic fantasies, and examines whether the relationship between these variables and the proportion of extradyadic fantasies is similar, as expected, for both men and women.

Another variable that might influence the proportion of extradyadic fantasies is the duration of the participant's current relationship. On the one hand, it could be predicted that participants who have been in their relationship for many years would tend to have more fantasies that involve someone other than their current partner. This hypothesis would be based on the notion that boredom, the desire for sexual variety, or both would increase as a function of relationship length. However, in contrast, if it were the case that frequent extradyadic fantasies are associated with infidelity or low relationship satisfaction, the opposite association might be expected between relationship length and the frequency of these fantasies, since frequent extradyadic fantasies might be associated with relationships that are not destined to last.

Regardless of whether there is any direct association between relationship length and the proportion of extradyadic sexual fantasies, it is also important to investigate the possibility of a gender by relationship length interaction. It is possible that as people mature, they may be more confident in their sexuality and less influenced by culturally based gender roles and sexual scripts. In a similar fashion, evolutionary psychology theorists have proposed that as people enter middle age they may be less affected by evolutionary-based influences on sexuality due to decreasing hormone levels (Wilson, 1997). Both of these perspectives imply that gender differences in the proportion of extradyadic fantasies may exist only among younger men and women in relatively recent relationships.

METHOD

Participants

Anonymous questionnaires were distributed to 336 students enrolled in an introductory psychology course at a mid-sized public university in New England. Ninety-four percent of the students (N = 317) returned completed surveys and received course credit in exchange for their participation. Anonymous questionnaires were also mailed to 1,136 faculty and staff employees of the university. Twenty-four percent of the university employees (N = 273) returned completed surveys using preaddressed envelopes that were included with the questionnaire. Data obtained from students and university employees was combined to form a total sample of 590 participants.

Because the present study is focused on the sexual fantasies of individuals who are currently in heterosexual relationships, 26 participants were excluded from the sample because they reported that their sexual orientation was primarily or exclusively gay, lesbian, or bisexual. All participants reporting that they were not currently in a relationship (N = 188) were also removed from the sample. Twenty-one cases were removed due to missing data on key variables, and an additional 6 cases were removed as univariate or multivariate outliers based on extremely high z scores or Mahalanobis distances with p [is less than] .001. All results and descriptive statistics are based on the 349 participants who remained after these initial data screening procedures.

The final sample of 349 participants (215 females, 134 males) was comprised of 147 students and 202 university employees. The mean age for students included in the sample was 19.0 years (SD = 1.2), as compared to 42.8 years (SD = 10.1) for the university employees (mean age for combined sample = 32.8, SD = 14.1, range = 18-70). Socioeconomic status (SES) was computed using Hollingshead's (1975) four factor index. For students in the sample, SES was calculated based on data regarding the participant's parents. Although there was a small difference in mean SES between students and university employees in the sample (47.0 vs. 53.7, t(273) = -6.08, p [is less than] .001), both groups were in Social Class II, medium business owners, minor professionals, and technical workers. Consistent with the demographic characteristics of this area of New England, 95% of participants were Caucasian.

All 349 participants reported that they were currently involved in a heterosexual relationship, with a mean relationship length of 9.4 years (SD = 10.8). Participants were asked to rate the level of commitment in their current relationship on a 7-point Likert scale ranging from 1 (no commitment) to 7 (very, strong commitment). The mean rating for the sample was 6.2 (SD = 1.3). Forty-eight percent of the sample (41% of the women, 59% of the men) reported that they were currently married and 4% (6% of the women, 2% of the men) reported that they were not married, but were currently cohabiting with their partner. Ninety-two percent of the sample reported that they had been in relationships prior to their current one. The mean number of prior relationships for the sample was 4.8 (SD = 4.8), and the mean number of prior sex partners was 6.8 (SD = 7.7).

Measures

An anonymous questionnaire was used for data collection. The wording of certain items in the demographics section was altered slightly for the university employee mailing so as to be age appropriate for this older sample. All other questions were identical for both samples. Participants first completed a 16-item demographics section. This section included questions regarding age, gender, ethnicity, sexual orientation, socioeconomic status, and religion. To determine whether participants were currently in a relationship, and for how long, we asked: "Are you currently in a relationship? (Do you have a significant other, boyfriend, girlfriend, sexual partner, spouse, etc.?)" and "If you answered `YES', how long have you been in your current relationship? (please be specific, e.g. 5 years or 6 months)". To assess participants' number of prior sex partners, we asked: "Regardless of how one defines a `relationship,' how many different people have you had sexual intercourse or oral sex with in your life?"

Participants were also asked whether they had ever engaged in extradyadic sexual behavior, or "cheated" on their current partner. Cheating was defined as "any type of sexual or romantic encounter with someone other than your current partner ranging from kissing to intercourse." Our definition of extradyadic sexual behavior was intentionally not restricted to intercourse. We felt that if actual extradyadic sexual experiences influenced fantasy, even the most tentative sexual encounters could exert such an effect.

The next section of the questionnaire ascertained the frequency of particular types of sexual fantasies. Prior to rating the frequency of their fantasies, participants were provided with the following broad definition of sexual fantasy adapted from Leitenberg and Henning (1995):

   In answering the following questions, consider the term sexual fantasy to
   be defined as any erotic or sexually arousing mental imagery that a person
   has while awake. It can be an elaborate story, or it can be a fleeting
   thought of some sexual activity. It can involve bizarre imagery, or it can
   be quite realistic. It can involve memories of past events, or it can be a
   completely imaginary experience. It can occur spontaneously or be
   intentionally imagined, or it can be provoked by other thoughts, feelings,
   or sensory cues. Sexual fantasies can take place outside of sexual
   activity, or they can occur during solitary masturbation or during sexual
   activity with a partner.

Half of the sample received surveys that referred to "sexual/romantic fantasies" in this definition and throughout the questionnaire. The other half of the sample received surveys that only referred to "sexual fantasies." We wanted to determine if this simple inclusion of the word "romantic" would increase the frequency of fantasies reported by women. However, comparisons of the groups receiving these two different versions revealed that this manipulation made no difference in any of the results that follow for either women or men.

Participants were asked to rate the frequency of their sexual fantasies over the past 2 months involving (a) their current partner and (b) someone who is not their current partner. This relatively short time interval was chosen in order to increase the accuracy of participants' estimates. A 2-month interval was thought to be long enough to insure that participants were reporting on all but the most infrequent of their fantasies. Within the category of someone who is not their current partner, we further measured the frequency of fantasies that involved a former partner versus fantasies that involved someone real or imaginary who had never been their partner.

Participants responded by circling a number on a 9-point Likert scale. Each point of the scale included a specific frequency label ranging from 1 (never) to 9 (9 or more times a day). Successive points on the scale represented a level of frequency approximately double the frequency represented by the previous point (e.g., 4 [about twice a week] indicates having fantasies approximately twice as frequently as 3 [about once a week]). The scale anchors used in the present study were designed to include the typical range of fantasy frequencies reported by both men and women in past research (Ellis & Symons, 1990; Jones & Barlow, 1990).

RESULTS

We first determined if there was a gender difference in the percentage of men and women who reported having any sexual fantasies about someone other than their current partner in the past 2 months. Ninety-eight percent of men compared to 80% of women reported having extradyadic fantasies, [chi square] (1, N = 349) = 23.53, p [is less than] .001. The odds ratio indicated that the men in our sample were 11 times more likely to have extradyadic fantasies than the women.

There was no difference in incidence as a function of marital status. We divided the sample into two groups: participants who were either married or cohabiting with their partner, and participants living apart from their partner. The percentage of participants who reported having extradyadic fantasies in the past 2 months was virtually identical for both groups: 86.6% for the married/cohabiting group (males = 98%, females = 78%) and 86.5% for the living-apart group (males = 98%, females = 81%).

We next examined whether there was a gender difference in the relative frequency of extradyadic fantasies. Before analyzing the frequency data, participants' responses on the fantasy frequency items were converted from the 9-point Likert scale used in the questionnaire to an actual frequency rate expressed as number of fantasies per month. This conversion was based on the mean of the frequency range anchored to each scale point. For example, respondents who circled a 2 (1-3 times per month) were scored as reporting two fantasies per month.

A two-tailed t test was used to first examine gender differences in the number of total fantasies per month reported by participants. The degrees of freedom were adjusted to account for heterogeneity of variance (Levene, 1960). As predicted, men reported a significantly higher number of fantasies per month as compared to women (76.7 vs. 34.1, t(192) = -4.77, p [is less than] .001). In order to control for this gender difference in the overall frequency of sexual fantasies, percentage scores were created by dividing the number of extradyadic fantasies reported by the total number of fantasies reported. These percentage scores allowed us to examine gender differences in the percentage of total fantasies that involve someone other than the participant's current partner.

A two-tailed t test with adjusted degrees of freedom was used to determine the significance of gender differences in the percentage of sexual fantasies that were extradyadic in nature. As indicated in Table 1, men reported that a higher percentage of their sexual fantasies involved someone other than their current partner as compared to women (54% vs. 36%, t(311) = -5.1, p [is less than] .001). Furthermore, men were more likely to fantasize about someone else (54% of all fantasies), rather than about their current partner. whereas the reverse was true for women, who were more likely to fantasize about a current partner (64% of all fantasies) than about someone else.

Table 1. Gender Differences in the Relative Frequency
of Particular Categories of Sexual Fantasies

Sexual fantasy category                      Males   Females

All fantasies

Involving participant's current partner      46%       64%
Involving someone other than participant's
 current partner                             54%       36%
(Extradyadic fantasies)
SD                                           30%       34%

Extradyadic fantasies

Involving a former partner                   22%       34%
Involving someone who has never been
 the participant's partner                   78%       66%
SD                                           29%       37%

An additional t test was performed to examine gender differences in the percentage of extradyadic fantasies that involved a former partner as opposed to someone who had never been the participant's partner. This t test was based on the 297 participants in the sample who reported that they (a) had prior relationships or sex partners and (b) had experienced extradyadic fantasies. Table 1 shows that women reported that a significantly higher percentage of their extradyadic sexual fantasies involved a former partner as compared to men, t(295) = -2.9, p = .004. However, Table 1 also shows that only a small percentage of extradyadic fantasies were about former partners: only 34% among women and 22% among men.

As expected, men (M = 8.9, SD = 10.0) reported having had a greater number of prior sex partners than women (M = 5.5, SD = 5.3), t(181) = -3.65, p [is less than] .001. Current relationship length was also somewhat longer for men (M = 11.8 years, SD = 11.9) than for women (M = 7.8 years, SD = 9.7), t(240) = -3.25, p [is less than] .001, and was probably related to the difference in average age between men (M = 36.3, SD = 14.3) and women (M = 30.6, SD = 13.6) in the sample, t(347) = -3.74, p [is less than] .001. Although socioeconomic status was also slightly higher for men as compared to women (52.6 vs. 49.8, t(347) = -2.51, p = .01), both genders were in Social Class II, characterized as including medium business owners, minor professionals and technical workers. Contrary to expectations, a nearly identical percentage of men and women (28% vs. 29%) had engaged in extradyadic sexual behavior or cheated on their current partner.

A hierarchical multiple regression was conducted to determine if gender predicted the percentage of extradyadic fantasies when controlling for the effects of length of current relationship, number of prior sex partners, prior incidents of extradyadic sexual behavior while with the current partner, and socioeconomic status. Age was also considered as a possible independent variable, but it was too highly correlated with length of current relationship (r = .83) to include both variables as predictors of extradyadic fantasies. The hierarchical multiple regression methodology also permitted an examination of length of relationship, number of prior sex partners, prior extradyadic sexual behavior, and SES as possible moderators of the gender difference in the percentage of extradyadic sexual fantasies. In other words, it was possible to examine whether there were significant interactions between gender and these variables.

The first step in the hierarchical multiple regression involved entering the five main effect predictor variables simultaneously into the model: (a) gender, (b) length of relationship (in months), (c) number of prior sex partners, (d) cheating on current partner, and (e) SES. Following the procedure described by Cohen and Cohen (1983), prior to analysis the relationship length, number of prior sex partners, and SES variables were centered around their means, while gender and cheating were dummy coded (0 = female, 1 = male; 0 = did not cheat, 1 = cheated). These centered and dummy coded variables were then used to create four interaction terms: (a) gender by relationship length, (b) gender by number of prior sex partners, (c) gender by cheating, and (d) gender by SES. These four interaction terms were entered simultaneously as the second step in the hierarchical multiple regression after the main effects had already been entered.

Table 2 displays the results of the hierarchical multiple regression. The overall [R.sup.2] for the full model that included all main effects and interaction terms was .25, F(9, 339) = 12.39, p [is less than] .001. There was a significantly higher percentage of extradyadic fantasies reported by men compared to women even when controlling for length of relationship, number of prior partners, past extradyadic sexual behavior while with the current partner, and SES. Significant positive regression coefficients were also found for relationship length, number of prior sex partners, and prior extradyadic sexual behavior. These results indicate that, regardless of gender, participants who have been in their current relationship for a longer period of time and participants who have had more prior sex partners have a higher percentage of extradyadic sexual fantasies. The results also indicate that, regardless of gender, participants who reported that they had cheated on their current partner in the past have a higher percentage of fantasies that involve someone other than their current partner. Socioeconomic status did not emerge as a significant predictor of the extradyadic percentage scores, nor did the SES by gender interaction.

Table 2. Hierarchical Regression Analysis for Variables
Predicting Percentage of Fantasies That are Extradyadic

Variables                             B     SE B   [Beta]

                                   Step 1: Main predictor variables

Gender                              11.99   3.47    .18
Relationship length                  0.06   0.01    .22
Number of sex partners               0.76   0.22    .17
Cheating on current partner         12.56   3.64    .17
SES                                   .15    .16    .05

                                   Step 2: Two-way interactions
                                   involving gender

Gender by relationship length       -0.03   0.03   -.08
Gender by number of sex partners    -1.41   0.47   -.27
Gender by cheating on partner      -22.47   7.33   -.21
Gender by SES                         .01    .33    .00
R = .50(**)
[R.sup.2] = .25(**)

Variables                          [sr.sup.2]       t

                                   Step 1: Main predictor variables

Gender                                .028       3.46(**)
Relationship length                   .045       4.37(**)
Number of sex partners                .027       3.38(**)
Cheating on current partner           .028       3.45(**)
SES                                   .002        .93

                                   Step 2: Two-way interactions
                                   involving gender

Gender by relationship length         .003      -1.18
Gender by number of sex partners      .020      -2.99(*)
Gender by cheating on partner         .021      -3.07(*)
Gender by SES                         .000        .03
R = .50(**)
[R.sup.2] = .25(**)

Note. N = 349.

(*) p < .01.

(**) p < .001.

The gender by length of relationship interaction did not emerge as a significant predictor in the multiple regression. This indicates that both male and female participants who have been in their current relationship for a longer period of time tend to report that a higher percentage of their sexual fantasies involve someone other than their current partner. It also indicates that, regardless of length of relationship, men tend to report that a higher percentage of their fantasies are extradyadic as compared to women.

The interaction of gender and number of prior sex partners, however, was a significant predictor of the extradyadic fantasy percentage scores. Figure 1 displays this interaction graphically as recommended by Aiken and West (1991), with separate simple regression lines fit to the data for men and women. The steep slope of the regression line for female participants indicates a strong positive correlation between number of prior sex partners and the percentage of extradyadic sexual fantasies. In contrast, the comparatively flat regression line for male participants indicates a relatively weak correlation between number of prior sex partners and the percentage of extradyadic fantasies. This interaction implies that the significant main effect for number of sex partners in the multiple regression is primarily due to the strong connection between number of prior partners and extradyadic fantasies found among female participants in the present study. In other words, women are more likely to fantasize about someone other than their current partner if they have had more sex partners in the past, whereas the likelihood of fantasizing about someone else is relatively high for men regardless of number of prior sexual partners.

[ILLUSTRATION OMITTED]

The interaction of gender and prior extradyadic sexual behavior or cheating on the current partner also emerged as a significant predictor in the multiple regression, and is displayed graphically in Figure 2. The percentages of fantasies that involved someone other than their current partner were nearly identical for men (55%) and women (53%) who had cheated on their partner in the past. In contrast, among participants who had not cheated on their current partner, men reported that a significantly larger percentage of their fantasies (54%) were extradyadic in nature as compared to women (30%), t(247) = -5.9, p [is less than] .001.

[ILLUSTRATION OMITTED]

DISCUSSION

A recent New York Times poll ("The Way We Live," 2000) indicated that 48% of respondents did not think it was "okay" to fantasize about having sex with someone else even if they were faithful to their partner. In our sample 87% of participants had such fantasies in the past two months (98% of men and 80% of women). Clearly, extradyadic fantasies are quite common for both men and women, regardless of whether or not people think it is acceptable to have them. Perhaps if people realized how common these fantasies are they would find it more acceptable both for themselves and their partner.

Men and women differed in the percentage of sexual fantasies they had about their partners as compared to someone else. As predicted by both sociocultural and evolutionary psychology perspectives, a larger proportion of men's sexual fantasies involved someone other than their current partner as compared to the fantasies of women. In fact, men had a smaller percentage of fantasies about their current partner than about someone else, whereas the reverse was true for women. In addition, although the percentage of extradyadic fantasies involving former sex partners was quite small for both genders, this type of fantasy was more common for women than for men. Although none of these gender differences have been demonstrated in the past, they are consistent with past research which suggests that on average women's sexual fantasies are more likely to involve an emotional connection with a particular partner, whereas men's fantasies are more focused on explicit sexual imagery, often without any romantic or emotional context (Leitenberg & Henning, 1995).

Although both evolutionary psychology and various sociocultural theories were used to generate hypotheses in the present research, this study was not designed to test whether one of these theories better accounts for gender differences in sexual fantasies. While it may be impossible to separate research findings on gender differences from the cultural context and political implications that surround them (Marecek, 1995), these results do not favor one approach over another.

We would argue that regardless of whether gender differences in extradyadic sexual fantasies originated in natural selection or are simply outgrowths of cultural socialization, the fact that men and women on average differ in the likelihood of having these fantasies, even after controlling for length of current relationship, number of prior sex partners, prior extradyadic sexual behavior, and SES is worth knowing for its own sake. It may help men and women to better understand and relate to each other if there is an awareness of differences as well as similarities between them in what is sexually arousing in fantasy. Information regarding the incidence and frequency of extradyadic fantasies could also prove useful to therapists working with individuals or couples who feel guilt related to their own extradyadic fantasies or jealousy related to the extradyadic fantasies reported by a partner.

In addition to the gender difference, a significant main effect for length of relationship emerged in the multiple regression. Individuals who had been in their current relationship for a longer period of time reported that a higher proportion of their fantasies involved someone other than their current partner. This finding implies that longer periods of time in a relationship may lead to a decrease in the excitement associated with fantasizing about one's partner, an increase in the excitement associated with fantasizing about someone else, or both. A positive association between length of relationship and extradyadic fantasies also implies that such fantasies do not interfere with the ability to maintain a relationship over time. It should be noted that the interaction of gender and relationship length was not significant. Therefore, it can be concluded that relationship length has the same effect on the proportion of extradyadic fantasies for both men and women. The lack of a significant interaction also means that regardless of length of relationship men are more likely to have a higher percentage of extradyadic sexual fantasies than are women.

There was also a significant main effect for number of prior sex partners. Individuals who had more sex partners in the past reported that a larger proportion of their fantasies were extradyadic, even though only one fifth to one third of these extradyadic fantasies actually were about former partners. Because the present study is correlational in nature, there is no way to determine the direction of causality in this association. In other words, it is possible that past experience is affecting fantasy, and individuals who have had more partners are drawing on memories of past experiences for their extradyadic sexual fantasies. However, it is also possible that fantasy is affecting behavior, and individuals who tend to fantasize about someone other than their current partner are more likely to choose to have sex with a variety of people. A final possibility is that a common underlying variable, such as a desire for varied sexual experiences, partly accounts for both a greater number of sex partners and a greater tendency to fantasize about someone other than one's current partner.

There was a significant interaction between gender and number of prior sex partners. This interaction revealed that number of prior sex partners was more important in predicting extradyadic fantasies among women than among men. For women, having more numerous prior sex partners was linked to a substantially higher percentage of extradyadic fantasies. This correlation was not nearly as strong for men. In fact, regardless of the number of prior partners they had, men were more likely to fantasize about someone else rather than their current partner. This suggests that the relative frequency of men's extradyadic fantasies is less likely to either predict future behavior or be influenced by past behavior as compared to women. For men, the proportion of sexual fantasies that involve someone other than their current partner will tend to be relatively high regardless of how few or how many prior sex partners they have had. This interaction also means that gender differences in extradyadic fantasies are likely to be less apparent among individuals who report very high numbers of prior sex partners. Based on the simple regression lines shown in Figure 1, among participants who had 15 or more prior partners (14% of the total sample, 9% of women, 20% of men), women were actually predicted to have a higher percentage of extradyadic fantasies than men.

The present study also examined the connection between having actually engaged in extradyadic sexual behavior or cheated on a current partner and fantasizing about someone other than one's current partner. A significant main effect was found showing that individuals who have cheated on their current partner have a significantly higher proportion of extradyadic fantasies. Once again, it is important to note that this correlational finding does not imply a particular causal direction in the relationship between extradyadic fantasies and actual extradyadic sexual behavior. It is possible that experiencing a higher proportion of extradyadic fantasies may increase a person's chances of cheating on their current partner. However, it is also possible that extradyadic sexual experiences may provide material for future sexual fantasies, thereby increasing the percentage of extradyadic fantasies. If both processes are taking place, it is possible to envision a positive feedback loop or vicious circle in which extradyadic sexual experiences tend to increase the frequency of extradyadic fantasies, which in turn increase the likelihood of future extradyadic sexual behavior.

The significant gender by cheating interaction found in the present study showed that the gender difference in extradyadic fantasies was only evident for people who had never cheated on their current partner. This interaction also showed that the main effect for prior extradyadic sexual behavior was primarily due to the strong connection between this type of behavior and extradyadic fantasies found among women in the sample. This connection was not apparent among men, who reported that a relatively high percentage of their fantasies involved someone other than their partner regardless of whether or not they had engaged in extradyadic sexual behavior in the past. Once again, it seems that the connection between behavior and the proportion of extradyadic fantasies is more tenuous for men as compared to women. Regardless of what they do in real life, men are prone to have sexual fantasies about someone other than their current partner.

Neither socioeconomic status nor the gender by SES interaction were significant predictors of the extradyadic fantasy percentage scores. The present study was based on a sample that was relatively homogenous in terms of SES. Therefore, these findings do not imply that SES would not emerge as a significant predictor in a similar study utilizing a more diverse sample.

Some limitations of the present study should be noted. Participants were asked to estimate the frequency of fantasies that occurred in the last two months, and retrospective data, even for such a recent period, may not be completely accurate. Nevertheless, even if these estimates are not exact, the robust gender differences in the proportion of extradyadic fantasies and their association to relationship length, number of prior partners and past extradyadic sexual behavior do not seem to be seriously compromised.

Another concern is that although the questionnaire used in the present study was anonymous, women may have been less likely than men to report that they were having sexual fantasies about someone other than their current partner even if they were having such fantasies to the same extent as men. Admitting to having extradyadic fantasies may represent more of a violation of a cultural taboo for women than for men. Unfortunately, all research on the frequency of sexual fantasies must rely on self-report data. One can only hope that the anonymous nature of the study served to counter such possible biases. The fact that an equal percentage of women and men admitted to actual cheating behavior suggests that this gender bias may not have been present.

The present study also is based on a convenience sample of university students and university employees. Obviously this sample is not representative of the U.S. population in terms of race, SES, geographic location, or education, and is limited to individuals in heterosexual relationships. In addition, although the response rate was quite high among the students in the sample, it was quite low (24%) among university employees in the sample. Future studies should further explore gender differences in the percentage of extradyadic fantasies in more demographically diverse and representative samples and among gay, lesbian, and bisexual individuals.

Despite these sampling limitations, this study represents the first empirical examination of gender differences in the proportion of extradyadic fantasies. The sample used in the present research showed considerable variation on a number of important demographic variables including age, length of relationship, marital status, and prior sexual experience. Though far from ideal, this is still an improvement over the majority of prior studies of gender differences in sexual fantasy, which have used samples comprised entirely of undergraduate students.

REFERENCES

Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage Publications.

Angier, N. (1999). Woman: An intimate geography. New York: Anchor Books.

Buss, D., & Schmitt, D. (1993). Sexual strategies theory: An evolutionary perspective on human mating. Psychological Review, 100, 204-232.

Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum Associates.

Eagly, A. H., & Wood, W. (1999). The origins of sex differences in human behavior: Evolved dispositions versus social roles. American Psychologist, 54, 408-423.

Eisenman, R. (1982). Sexual behavior as related to sex fantasies and experimental manipulation of authoritarianism and creativity. Journal of Personality and Social Psychology, 43, 853-860.

Ellis, B. J., & Symons, D. (1990). Sex differences in sexual fantasy: An evolutionary psychological approach. The Journal of Sex Research, 27, 527-555.

Gagnon, J. H., & Simon, W. (1973). Sexual conduct. Chicago: Aldine.

Hollingshead, A. B. (1975). Four factor index of social position. Unpublished manuscript, Yale University, New Haven, CT

Jones, J. C., & Barlow, D. H. (1990). Self-reported frequency of sexual urges, fantasies, and masturbatory fantasies in heterosexual males and females. Archives of Sexual Behavior, 19, 269-279.

Laumann, E. O., Gagnon, J. H., Michael, R. T., & Michaels, S. (1994). The social organization of sexuality: Sexual practices in the United States. Chicago: The University of Chicago Press.

Leitenberg, H., & Henning, K. (1995). Sexual fantasy. Psychological Bulletin, 117, 469-496.

Levene, H. (1960). Robust tests for the equality of variance. In I. Olkin (Ed.), Contributions to probability and statistics: Essays in honor of Harold Hotelling (pp. 278-292). Palo Alto, CA: Stanford University Press.

Malamuth, N. M. (1981). Rape fantasies as a function of exposure to violent sexual stimuli. Archives of Sexual Behavior, 10, 33-47.

Marecek, J. (1995). Gender, politics, and psychology's ways of knowing. American Psychologist, 50, 162-163.

Oliver, M. B., & Hyde, J. S. (1993). Gender differences in sexuality: A meta-analysis. Psychological Bulletin, 114, 29-51.

Thompson, A. P. (1983). Extramarital sex: A review of the research literature. The Journal of Sex Research, 19, 1-22.

The way we live now poll. (2000, May 7). The New York Times Magazine, p. 76.

Wilson, G. D. (1997). Gender differences in sexual fantasy: An evolutionary analysis. Personality and Individual Differences, 22, 27-31.

Manuscript accepted November 20, 2000

Thomas V. Hicks and Harold Leitenberg University of Vermont

Address correspondence to Harold Leitenberg, Department of Psychology, University of Vermont, Burlington, Vermont 05405; e-mail: [e-mail].