Is It "Sex"?: College Students' Interpretations of Sexual Behavior Terminology.

Because sexual behavior is private and cannot easily be measured or verified, social scientists rely on individuals' self-reports of their sexual behaviors. This information can be obtained through face-to-face interviews, surveys, or the use of computer-assisted technologies (Wagstaff, Abramson, & Pinkerton, 2000). The accuracy and reliability of the data so obtained depend upon respondent and instrument-related characteristics (for reviews on measurement error in sexual behavior surveys see Berk, Abramson & Okami, 1995; Catania, Binson, van der Straten, & Stone, 1995; Catania, Gibson, Marin, Coates, & Greenblatt, 1990). For example, survey respondents may under- or over-estimate how often they engage in a given sexual activity because the survey items or terms are not clearly understood or have different meanings for individual respondents (Catania et al., 1996; Cecil & Zimet, 1998; Weir, Roddy, Zekeng, Ryan, & Wong, 1998).

Research conducted in the 1970s and 1980s provides evidence that men and women interpret and use sexual behavior terms differently. Robinson, Balkwell, and Ward (1980) found that men and women associated different meanings with the word intercourse. Men tended to associate words with intercourse that were body-centered (e.g., words like breast, kiss, and vagina), whereas women associated words with intercourse that were relationship-centered or associated with intimacy (e.g., love or marriage). Other researchers have found that men and women assign different connotative meanings and imagery to sexual words (Campos & Sueiro, 1991; McDermott, Drolet, & Fetro, 1989), and that men can list more slang expressions for terms associated with sexual activity than can women (Kutner & Brogan, 1974; Sanders & Robinson, 1979; Walsh & Leonard, 1974).

Recent research indicates that survey participants may hold different views on what constitutes a particular sexual behavior (Sanders & Reinisch, 1999; Sonenstein, Ku, & Pleck, 1997). For example, Sanders and Reinisch asked college students to indicate whether or not they would say that they had sex with someone if they had engaged in a particular activity with that person. The activities included manual-genital contact or oral contact with breasts or nipples, oral-genital contact, penile-vaginal intercourse, and penile-anal intercourse. Almost all participants considered engaging in penile-vaginal intercourse as having had sex. In contrast, approximately 60% and 20% reported that they would not consider oral-genital contact and penile-anal intercourse as having had sex.

The findings of Sanders and Reinisch (1999) confirm the qualitative field work conducted by Sonenstein et al. (1997). Their interviews with young men between the ages of 15 and 19 years indicated that participants differed in their interpretations of the term having sex. For instance, one youth believed that touching a woman's breast constituted having sex. In addition, these young men held different opinions as to whether fellatio was sexual intercourse.

In American culture, sex is associated closely with the male orgasm, which is sometimes viewed as the natural culmination of all sexual activity (Abramson & Pinkerton, 1995; Laumann, Gagnon, Michael, & Michaels, 1994; Masters, 1966). Because many sexual activities, such as vaginal intercourse, do not invariably lead to female orgasm (Kinsey, Pomeroy, Martin, & Gebhard, 1953; Maines, 1999), American women might be expected to be less likely to define sex solely in terms of orgasm (i.e., his or her orgasm). In line with this thinking, Berk et al. (1995) found that female college students reported that they had engaged in oral sex more frequently than did the male college students surveyed. This is an inconsistent finding if, as assumed by the authors, their participants represented an interdating population. The authors suggested that women might be more likely to remember and report oral sex experiences, because oral sex is more salient to women.

In sum, prior research indicates that misunderstandings of sexual behavior terminology exist; that the terms used in sexual behavior surveys are subject to individual interpretation; and that individuals' interpretations of survey terms can influence responses. More in-depth and systematic research is therefore needed on the meaning(s) that respondents may give to terms that identify or describe sexual behavior.

THE PRESENT STUDY

The present study explores the manner in which survey respondents interpret the term sex. In particular, we wanted to obtain a better understanding of the behaviors that young adults consider to be sex, and we wanted to determine if their judgments could be influenced by individual characteristics and situational factors. Thus, we asked participants to judge whether two hypothetical actors, Jim and Susie, would consider different sexual behaviors to be sex, and whether such factors as the gender of the participant, the gender of the actor, the type of act, and whether or not orgasm occurred influenced their labeling of behaviors. We used hypothetical scenarios because we believed that the use of scenarios would reduce anxiety in individuals who may not have wanted to divulge beliefs about their own sexual behavior, and would therefore increase item response rates and the validity of obtained responses.

We hypothesized that almost all study participants would believe that Jim and Susie would classify vaginal intercourse as sex and, following Sanders and Reinisch (1999), that Jim and Susie would be less likely to label anal intercourse and oral intercourse (fellatio/cunnilingus) as sex. In addition, we anticipated that participants would indicate that Jim and Susie would be more likely to label a behavior as sex if the act resulted in orgasm, and that this would be especially true when the scenario focused on whether Jim did or did not have an orgasm. Relatedly, we hypothesized that participants would attribute a broader definition of sex to Susie than to Jim, in that she might consider many activities to be sex that Jim might not (especially nonorgasmic activities). Thus, because we thought that women would identify with Susie and men would identify with Jim, we believed that men would place more importance on Jim having an orgasm than would women, and that women would place more importance on Susie having an orgasm than would men. Therefore, we expected to find significant participant gender by actor (Jim or Susie) interactions, such that, for example, the young men in our study would be more likely than their female counterparts to believe that Jim would consider a behavior to be sex if he alone had an orgasm.

METHOD

Participants

The sample consisted of 223 undergraduate students enrolled in a human sexuality course at a large university in the western United States. Participants ranged in age from 19 to 40 years (M = 22.2, SD = 2.2). Sixty-five percent (n = 145) were women. Thirty-seven percent self-identified as Asian/Pacific Islander, 34% as White, 12% as Hispanic, 4% as African American, 1% as Native American, 6% as Bi- or Multiracial, and 6% as Other. English was the first language for 70%. Most of the participants (96%) were college seniors and single (97%). Of the single participants, 51% had a steady partner, 12% had multiple dating partners, and 37% were not presently dating anyone. Ten percent reported that they had never engaged in vaginal, anal, or oral intercourse; 15% did not indicate whether they had ever engaged in these behaviors. Eight percent self-identified as gay, lesbian, or bisexual.

Procedure

This study was conducted in March, 1998. On January 26, 1998, President Clinton addressed the nation to quell rumors that he had had an affair with Monica Lewinsky, a White House intern. In a much-quoted statement, he said, "I did not have sexual relations with that woman" (Shogren, Willman, & Cooper, 1998). By August, the press had reported details of the President's relationship with Ms. Lewinsky (Baker & Hams, 1998), including the fact that the two had engaged in oral sex (Broder, 1998; Janofsky & Van Natta, 1998). The national debate that arose as to whether oral sex was sex, and whether the President had been telling the truth when he denied having a sexual relationship with Ms. Lewinsky (Fisher, 1998; Kuczynski, 1998), received media attention in late 1998, when the President's lawyers argued that he did not perjure himself when he denied having a sexual relationship with Ms. Lewinsky because having oral sex does not constitute a sexual relationship (Gerstenzang & Shogren, 1998). Thus, this study was conducted before details of the President's relationship with Monica Lewinsky (and the meaning of oral sex) were widely reported by the press and, more importantly, widely discussed on college campuses.

Participants completed an anonymous, self-report questionnaire during a regularly-scheduled class period in the middle of the semester. The questionnaire, which was administered by the instructor of the class, consisted of 133 items and took approximately 30 minutes to complete. To control for order effects, the sexual behavior scenario questions were randomly arranged in two different orders; participants were distributed one of the two different versions of the survey.

MEASURES

Demographics. Seven items were used to assess participants' age, gender, year in school, marital and dating status, and race/ethnic background. Three additional items measured participants' sexual experience for anal, vaginal, and oral intercourse (e.g., "Have you ever engaged in vaginal intercourse?").

Sexual behavior scenarios. Each participant was asked to consider 16 different sexual behavior scenarios involving two hypothetical actors, Jim and Susie, and to indicate for each scenario if he/she believed that Jim and Susie would "consider this sex." Separate judgments were made for Jim and Susie. The scenarios varied by type of act (vaginal intercourse, anal intercourse, and oral intercourse [fellatio/cunnilingus]) and whether or not the behavior led to orgasm for Susie, Jim, both, or neither. For example, participants read, "Jim and Susie meet at a bar. They go back to his apartment where they engage in [vaginal] intercourse. [Only Jim has an orgasm.] Would Jim consider this sex? Would Susie consider this sex?" (The words in brackets reflect the phrases that differed between scenarios.) Response choices were yes and no (which were subsequently coded as 1 and 0, respectively), making the dependent variable dichotomous.

To reflect the nature of the three different sexual acts, we varied the wording across scenarios. The wording of the oral intercourse scenarios indicated that one partner (the active partner) was stimulating the other partner (the recipient), who did or did not achieve orgasm. For example, "Jim and Susie meet at a bar. They go back to his apartment where Jim performs oral sex on Susie and she has an orgasm. Would Jim consider this sex? Would Susie consider this sex?" However, the wording of the vaginal and anal intercourse scenarios indicated that both partners were actively participating in the behavior. Hence, the oral intercourse scenarios reflected more asymmetrical behaviors, and the vaginal and anal intercourse scenarios reflected more symmetrical behaviors.

Other measures. Participants completed other measures, which were not the focus of the present investigation. These measures assessed perceived social norms for different sexual behaviors, responses to scenarios about sexual partners, perceived pleasure of different sexual behaviors, and social desirability.

Statistical Analyses

A generalized estimating equations (GEE) approach was used to model participants' yes/no responses within each scenario to the question, "Would Jim (Susie) consider this sex?". Unlike traditional methods that have been developed to analyze continuous outcomes, this regression-based technique is appropriate when participants' observations are binary and correlated, as they were in this study (Diggle, Liang, & Zeger, 1994; Liang & Zeger, 1986; Zeger & Liang, 1986).

To test the hypothesis that vaginal intercourse would be more likely to be rated as sex than the other types of sex acts, we calculated the grand mean for each type of behavior (vaginal, oral, and anal intercourse) by averaging over the appropriate eight items; the data were then arrayed as a one between-, one within-subjects design. Participant gender was the between-subjects variable; type of behavior was the within-subject variable. The logistic regression model included main effects for participant gender and type of behavior, and the gender x type of behavior interaction. Single degree-of-freedom contrasts were used to test pair-wise differences between the grand means of each type of sexual behavior.

To test our remaining hypotheses, we performed three separate logistic regression analyses that examined responses to the scenarios for each act (i.e., vaginal, anal, and oral intercourse). In each analysis, participant gender was a between-subject effect and actor (Jim, Susie) was one of the within-subject effects. For oral intercourse, another within-subject variable was included in the analyses to reflect whether Jim or Susie was the recipient of the stimulation. Because the vaginal and oral intercourse scenarios did not reflect this kind of asymmetry, this variable was not included in analyses of the responses to these scenarios.

For each analysis, we initially used a saturated model that modeled the log odds of a yes response as a function of all main effects and interactions. If one or more of the interaction effects were not significant in the saturated model, we fit a reduced model that excluded nonsignificant interaction effects. We included all of the main effects (and lower order interactions) when an analysis based on the saturated model indicated that an interaction effect was needed. The chi square statistic was used to test the overall fit of the reduced model; the z statistic was used to test single-degree-of-freedom hypotheses and to conduct simple effects tests associated with significant main effects and interactions. All GEE analyses were performed using STATA v5.0 (Stata Corporation, 1997).

We provide regression coefficients and standard errors for the final model for each sex act (i.e., vaginal, anal, and oral intercourse). In presenting our findings, we report cell probabilities instead of log odds or logits, which may be a less familiar metric for some readers. The cell probabilities were obtained by a back-transformation of the mean of the log odds (see Hosmer & Lemeshow, 1989); they can be interpreted as the conditional probability of a yes response, given the study participant's gender and the other explanatory variables in the model (e.g., whether the rating was made for Jim or Susie, whether the recipient of oral sex was Jim or Susie, and/or whether orgasm occurred).

RESULTS

Proportion Indicating that Jim/Susie Would Consider the Acts to be Sex

Table 1 shows the proportion of participants who indicated that Jim/Susie would classify each behavior as sex by type of behavior, whether the target person(s) had an orgasm, and the participant's gender. Overall, more than 90% indicated that vaginal and anal intercourse would be considered sex by Jim (overall Ms = .96 and .92, respectively) and Susie (overall Ms = .98 and .93). In contrast, less than 50% indicated that oral intercourse would be considered sex by Jim (overall M = .41) and Susie (overall M = .46), especially when the recipient did not experience orgasm (overall Ms = .35 and .41 for Jim and Susie, respectively). Participants' responses indicated that they thought that Jim and Susie would be most likely to agree with one another and consider the behavior to be sex when the two had vaginal or anal intercourse and both of them had an orgasm, and would be most likely to disagree with one another when the two of them engaged in oral intercourse (fellatio or cunnilingus) and the recipient alone had an orgasm.

Table 1. Proportion of Participants Who Think Jim or Susie Would Label Various Sexual Behaviors as Sex

                                     Would Jim consider it sex?

                                         All        Men    Women
                                     participants   only   only

Vaginal             Jim orgasms           .99        .99     .99
intercourse         Susie orgasms         .92        .95     .90
                    Both orgasm          1.00       1.00    1.00
                    Neither orgasm        .90        .95     .88

Anal                Jim orgasms           .96        .96     .96
intercourse         Susie orgasms         .88        .96     .84
                    Both orgasm           .96        .95     .97
                    Neither orgasm        .89        .91     .88

Oral sex,           Jim orgasms           .54        .58     .52
recipient = Jim     No orgasm             .34        .40     .31

Oral sex,           Susie orgasms         .41        .49     .36
recipient = Susie   No orgasm             .36        .44     .32

                                     Would Susie consider it sex?

                                         All        Men    Women
                                     participants   only   only

Vaginal             Jim orgasms           .96        .94     .97
intercourse         Susie orgasms        1.00       1.00    1.00
                    Both orgasm          1.00       1.00    1.00
                    Neither orgasm        .97        .97     .97

Anal                Jim orgasms           .91        .92     .90
intercourse         Susie orgasms         .94        .94     .94
                    Both orgasm           .97        .97     .97
                    Neither orgasm        .91        .92     .90

Oral sex,           Jim orgasms           .43        .50     .39
recipient = Jim     No orgasm             .39        .42     .37

Oral sex,           Susie orgasms         .58        .62     .55
recipient = Susie   No orgasm             .44        .45     .43

Is Vaginal Sex More Likely to be Considered Sex?

Participants indicated that Jim and Susie would be more likely to classify vaginal (M = .97) and anal (M = .92) intercourse as sex than they would be to classify oral intercourse as sex (M = .44) (zs = 15.36 and 14.08, ps [is less than] .001, respectively). Participants also indicated that Jim and Susie would be more likely to classify vaginal intercourse as sex than they would be to classify anal intercourse as sex, although the magnitude of the effect was not large (z = 3.53, p [is less than] .000). Overall, male participants' judgments of what Jim and Susie would consider to be sex were comparable to the judgments made by female participants (z = 1.06, p = .29). Finally, we obtained a nonsignificant participant gender x behavior interaction, indicating that male participants' implicit ranking of the three behaviors was comparable to that of female participants ([chi square] = 0.05, df = 2, p = .98).(1)

Vaginal Intercourse

A one between-, two within-subjects design (2 x 2 x 4) was used to analyze the eight vaginal intercourse items. The between-subjects variable was participant gender; the within-subject variables were actor (Jim, Susie) and orgasm (Jim had an orgasm and Susie did not; Susie had an orgasm and Jim did not; both had an orgasm; or neither had an orgasm). This analysis allowed us to test the following three hypotheses. First, participants would be more likely to label a behavior as sex when Jim and/or Susie had an orgasm than when neither one had an orgasm; and participants would be more likely to do so when Jim had an orgasm. Second, if participants attributed a broader definition of sex to Susie than to Jim, a definition that does not depend on having an orgasm, they would indicate that Susie would be more likely than Jim to label as sex the contacts described by the eight scenarios. Third, there would be a significant participant gender by outcome interaction, as male participants, for example, would be more likely than female participants to believe that Jim would only consider vaginal intercourse to be sex if he had an orgasm.

Because all participants thought that Jim and Susie would label vaginal intercourse as sex if both had an orgasm, we could not fit any model that contained the participant gender x actor x orgasm interaction. We therefore fit a model with main effects for participant gender, actor, and orgasm, and the participant gender x orgasm, participant gender x actor, and actor x orgasm interactions. Two single-degree-of-freedom contrasts were used to partition the orgasm main effect--specifically, Jim (Susie) having an orgasm was contrasted with neither having an orgasm. When the initial fit indicated that there were no participant gender effects, we fit a reduced model that included effects for actor, orgasm, and their interaction.

With the reduced model, the two main effects and their interaction were statistically significant, although the differences between the means were small (see Table 2). The actor main effect indicated that participants thought Susie (M = .99) was slightly more likely than Jim (M = .96) to consider vaginal intercourse to be sex; the orgasm contrasts indicated that participants believed that Jim and Susie would be somewhat more likely to label vaginal intercourse to be sex if one of them had an orgasm than if neither of them had an orgasm (Jim: M = .99; Susie: M = .98; Neither: M = .95).

Table 2. Final GEE Model for the Vaginal Intercourse Items

                                                      Standard
Effect                                  Coefficient    error

Orgasm 1
  (Jim orgasms vs. neither orgasms)         0.54        .23
Orgasm 2
  (Susie orgasms vs. neither orgasms)       2.43        .68
Actor                                      -2.27        .75
Actor x Orgasm 1                            0.72        .18
Actor x Orgasm 2                           -2.32        .68
Constant                                    5.83        .69

Effect                                      z

Orgasm 1
  (Jim orgasms vs. neither orgasms)      2.30(*)
Orgasm 2
  (Susie orgasms vs. neither orgasms)    3.58(***)
Actor                                   -3.04(**)
Actor x Orgasm 1                         4.08(***)
Actor x Orgasm 2                        -3.41(**)
Constant                                 8.42(***)

Note. Model [chi square] (5) = 54.62, p < .001.

(*) p < .05.

(**) p < 01.

(***) p < .001.

The significant actor x orgasm interaction was consistent with our hypotheses, although the magnitude of the effects was small. Participants believed that Jim would be more willing to consider vaginal intercourse to be sex if he had an orgasm than if neither he nor Susie had an orgasm (M = .99 vs. M = .90, respectively; z = 3.66, p [is less than] .001). Likewise, participants believed that Susie would be somewhat more willing to consider vaginal intercourse to be sex if she had an orgasm than if neither of them had an orgasm (M = 1.00 vs. M = .97, respectively; z = 3.50, p [is less than] .001). On the other hand, respondents thought that whether or not Jim had an orgasm would not affect whether Susie considered vaginal intercourse to be sex (M = .96 if Jim had an orgasm; M = .97 if neither did; z = .78, p = .44), and whether or not Susie had an orgasm would not affect Jim's labeling of the act as sex (M = .92 if Susie had an orgasm; M = .90 if neither did; z = 1.04, p = .30). Finally, when neither partner had an orgasm, participants thought that Susie would be more likely than Jim to consider vaginal intercourse to be sex (M = .97 vs. M = .90; z = 3.50, p [is less than] .001).

To summarize, we received support for our hypotheses that participants believe that vaginal intercourse is more likely to be considered sex if it results in orgasm, particularly by the person who experiences the orgasm, and that Susie would be expected to have a broader definition of sex than Jim. It is important to note, however, that the associated effect sizes were small--almost all participants labeled vaginal intercourse as sex, regardless of the gender of the actor or whether orgasm was experienced. Moreover, the hypothesized participant gender x actor interactions were not obtained and there was no evidence that Jim's orgasm was judged to be more important than Susie's.

Anal Intercourse

Our hypotheses for anal intercourse were similar to those presented above for vaginal intercourse, and a similar one between-, two within-subjects design (2 x 2 x 4) was employed to analyze the data from the eight anal intercourse items. The within-subjects variables were actor and orgasm. Three single-degree-of-freedom contrasts were used to partition the orgasm main effect: (a) Jim had an orgasm, (b) Susie had an orgasm, and (c) Jim and Susie both had orgasms were contrasted with neither having an orgasm.

In the saturated model, the gender and actor main effects and the gender x actor and gender x orgasm interactions were not significant. (Thus, participants believed that Jim and Susie were equally likely to consider anal intercourse sex, and male participants were as likely as female participants to believe that an actor would label anal intercourse as sex.) Therefore, a reduced model was fit that included participant gender, actor, and orgasm main effects and the actor x orgasm and participant gender x actor x orgasm interactions.

Table 3 shows the regression coefficients (and standard errors) for the effects in the reduced model. Despite the significant effects, the differences between the means were not large. The significant contrasts associated with the orgasm main effect indicate that participants believed that Jim and Susie would be slightly more likely to consider anal intercourse to be sex if Jim had an orgasm (M = .94) or if both of them had an orgasm (M = .97), as opposed to neither having an orgasm (M = .90). In contrast, participants believed that Jim and Susie would be as likely to label anal intercourse to be sex when Susie had an orgasm (M = .92) as when neither of them had an orgasm.

Table 3. Final GEE Model for the Anal Intercourse Items

                                                      Standard
Effect                                  Coefficient    error

Participant gender                          .34         .21
Orgasm 1
  (Jim orgasms vs. neither orgasms)         .30         .10
Orgasm 2
  (Susie orgasms vs. neither orgasms)       .13         .10
Orgasm 3
  (both orgasm vs. neither orgasms)         .58         .16
Actor                                       .09         .11
Actor x Orgasm 1                            .27         .10
Actor x Orgasm 2                           -.05         .07
Actor x Orgasm 3                           -.04         .11
Participant gender x Actor x Orgasm 1      -.06         .07
Participant gender x Actor x Orgasm 2       .15         .07
Participant gender x Actor x Orgasm 3      -.13         .10
Constant                                   3.32         .35

Effect                                      z

Participant gender                       1.66(+)
Orgasm 1
  (Jim orgasms vs. neither orgasms)      3.05(**)
Orgasm 2
  (Susie orgasms vs. neither orgasms)    1.32
Orgasm 3
  (both orgasm vs. neither orgasms)      3.70(***)
Actor                                    0.84
Actor x Orgasm 1                         2.79(**)
Actor x Orgasm 2                        -0.73
Actor x Orgasm 3                        -0.39
Participant gender x Actor x Orgasm 1   -0.84
Participant gender x Actor x Orgasm 2    2.33(*)
Participant gender x Actor x Orgasm 3   -1.34
Constant                                 9.49(***)

Note. Model [chi square] (11) = 25.36, p < .008.

(+) p < .10.

(*) p < .05.

(**) p < .01.

(***) p < .001.

The orgasm main effect was qualified by an actor x orgasm interaction and by a participant gender x actor x orgasm interaction. With respect to the former, participants believed that Jim would be somewhat more likely to consider anal intercourse to be sex when he had an orgasm, as opposed to when neither he nor Susie had an orgasm (M = .96 vs. .89, respectively; z = 3.41, p = .001), but that Susie would be equally likely to rate these outcomes to be sex (Ms = .91 and .91, respectively; z = .30, p = .77). Further, they believed that Jim would be slightly more likely than Susie to label anal intercourse to be sex when he had an orgasm (M = .96 vs. .91, respectively; z = 2.40, p = .016), and that Jim and Susie would provide comparable judgments when neither of them had an orgasm (z = -.82, p = .41).

Finally, the significant contrast associated with the participant gender x actor x orgasm interaction suggested that male and female participants tended to disagree about how Jim and Susie would react if only Susie had an orgasm. Female participants indicated that Susie would be more likely to label anal intercourse to be sex when she had an orgasm than when neither she nor Jim had an orgasm (M = .94 vs. M = .89, respectively; z = 2.20, p = .03), and that Jim would be equally likely to consider both outcomes as sex (Ms = .84 and .86; z = .64, p = .52). Female participants also indicated that Susie would be more likely than Jim to consider anal intercourse as sex when she alone had an orgasm (M = .94 vs. .84, respectively; z = 3.26, p [is less than] .001). In contrast, male participants indicated that Jim and Susie would hold comparable views regardless of whether Susie had an orgasm (Ms = .96 and .94) or neither partner had an orgasm (Ms = .93 and .94).

In short, the pattern of results obtained for anal intercourse suggests that participants believed that Jim's orgasm was a determining factor in whether or not Jim viewed anal intercourse to be sex. Moreover, women, but not men, believed that Susie's having or not having an orgasm during anal intercourse would affect her classification of this behavior, but would not influence whether Jim considered it sex. None of these effects were large, however, and overall, the majority of participants believed that the actors would consider anal intercourse to be sex.

Oral Intercourse (Fellatio/Cunnilingus)

Data from the eight oral sex items were arrayed and analyzed as a one between-, three within-subjects design. The between-subjects variable was participant gender; the within-subject variables were actor (Jim, Susie), gender of the recipient of oral sex (Jim, Susie), and whether or not the recipient achieved orgasm. Our analysis allowed us to test the following pair of hypotheses about participants' perceptions. First, oral intercourse would be more likely to be considered sex if the recipient experienced orgasm, especially if the recipient was male. Second, Jim would be less likely than Susie to consider oral intercourse sex, especially if it did not result in orgasm.

Because none of the effects associated with participant gender were significant in the initial model, a reduced model was fit that did not include any of the participant gender effects. As shown in Table 4, the orgasm and actor main effects were significant. The significant orgasm main effect indicated that, as predicted, participants believed that Jim and Susie would be more likely to classify oral intercourse to be sex when the recipient achieved orgasm than when the recipient did not (M = .49 vs. .38, respectively). The significant actor main effect indicated that oral intercourse was more likely to be considered sex when the actor was Susie rather than Jim (M = .46 vs. .41, respectively).

Table 4. Final GEE Model for the Oral Intercourse Items

                                                  Standard
Effect                              Coefficient    error

Recipient gender                       -.04         .03
Orgasm                                  .21         .03
Actor                                  -.09         .04
Recipient gender x Orgasm               .03         .03
Actor x Recipient gender                .16         .04
Orgasm x Actor                          .04         .02
Actor x Orgasm x Recipient gender       .13         .03
Constant                               -.27         .11

Effect                                  z

Recipient gender                    -1.37
Orgasm                               6.58(**)
Actor                               -2.24(*)
Recipient gender x Orgasm             .99
Actor x Recipient gender             4.35(**)
Orgasm x Actor                       1.57
Actor x Orgasm x Recipient gender    5.35(**)
Constant                            -2.43(*)

Note. Model [chi square] (7) = 90.94, p < .001.

(*) p < .05.

(**) p < .01.

The significant actor x recipient gender interaction suggested that participants believed that Jim and Susie would label the behavior differently, depending on who was the recipient. Participants indicated that Jim would be more likely to consider oral intercourse to be sex when he was the recipient than when Susie was the recipient (M = .44 vs. .38, respectively; z = 2.50, p = .01). Similarly, participants indicated that Susie would be more likely to label the behavior as sex when she was the recipient than when Jim was the recipient (M = .51 vs. .41, respectively; z = -4.18, p [is less than] .001).

Finally, participants' responses suggested that Jim would be more likely to label oral intercourse to be sex when he was the recipient and he had an orgasm than when Susie was the recipient and she had an orgasm (M = .54 vs. .41 respectively; z = 4.64, p [is less than] .001). Similarly, participants indicated that Susie would be more likely to consider oral intercourse to be sex if she was the recipient and she had an orgasm than if Jim was the recipient and he had an orgasm (M = .58 vs. .43, respectively; z = 5.06, p [is less than] .001). When the act did not result in orgasm, participants indicated that Jim was equally likely to label the behavior to be sex regardless of whether Susie performed oral sex on him or he performed oral sex on her (M = .34 vs. .36, respectively; z = -.77, p = .44), and that Susie was equally likely to label the behavior as sex whether Jim performed oral sex on her or she performed oral sex on him (M = .43 vs. .39, respectively; z = 1.51, p =. 13).

In sum, less that half of participants overall believed that Jim and Susie would rate oral intercourse to be sex. Further, oral intercourse was more likely to be considered sex by the female actor and by the recipient of the behavior, especially if the recipient achieved orgasm. If the recipient did not have an orgasm, participants believed that the recipient would be as likely to consider oral intercourse to be sex as the partner who performed the act. Contrary to predictions, male and female participants did not judge the oral intercourse scenarios differently.

Impact of Sexual Experience on Responses to the Scenarios

In order to assess the impact of participants' sexual experience on their responses to the sexual behavior scenarios, sexual experience was included as a covariate in the final GEE models for vaginal, oral, and anal intercourse. For each act, sexual experience was defined as ever having engaged in that act. Thus, for the vaginal intercourse scenario, participants who had ever engaged in vaginal intercourse were contrasted with those who had never engaged in vaginal intercourse.

In the models for vaginal and anal intercourse, the coefficient for sexual experience was not significantly different from zero. Moreover, the values of the remaining coefficients did not differ appreciably from the values obtained when the sexual experience covariate was not included. In the model for oral sex, the value of the coefficient for sexual experience was significantly different from zero; however, the values of the other coefficients in the model were comparable to the values observed when the covariate was not in the model. The vaginal (anal) intercourse findings indicated that the responses of the participants who had had vaginal (anal) intercourse were comparable to the responses of the participants who had never engaged in the behavior. Closer examination of the oral intercourse findings revealed that the participants who had engaged in oral sex were less likely to indicate that Jim (Susie) would label the behavior as sex. For all three sexual behaviors, participants' prior sexual experiences (as assessed by a single binary covariate) did not seem to play a role in their interpretation of the word sex.

DISCUSSION

The present study examined how respondents interpret the word sex and how various factors impact respondents' interpretations. The results suggest that young adults may subscribe to broad cultural definitions of sexuality when deciding whether a given act is considered to be sex. As such, the results are consistent with the results of prior studies (Sanders & Reinisch, 1999; Sonenstein et al., 1997) that also investigated individuals' interpretations of the meaning of sex and related terms. This study extends past research by exploring some of the contextual factors that can influence individuals' definitions of what constitutes sex. In particular, whether or not a given behavior is considered to be sex may depend on factors such as the nature of the behavior, whether or not orgasm occurs, and who experiences orgasm.

The majority of respondents labeled vaginal and anal intercourse as sex. Our results suggest that there may be more consensus in society vis-a-vis the meaning of sex as vaginal or anal intercourse, and there may be less consensus vis-a-vis the meaning of sex as oral intercourse. Thus, although the results suggest that there is potential for measurement error if one is not careful about the wording of sexual behavior surveys, the overall magnitude of that error is not likely to be large in the case of vaginal and anal intercourse.

Not surprisingly, vaginal and anal intercourse were rated as more likely than oral intercourse to be considered sex. This finding is consistent with the results of Sanders and Reinisch (1999), and with the results of a similar study that we conducted with younger undergraduates at a mid-Atlantic university who were enrolled in an introductory psychology class (Bogart et al., 1999). However, a higher percentage of individuals in the present study than individuals in Sanders and Reinisch's (1999) study considered anal intercourse to be sex. Overall, 97%, 93%, and 44% of the participants in our study thought that vaginal, anal, and oral intercourse, respectively, would be considered sex. In comparison, 99%, 81%, and 40% of the students in the Sanders and Reinisch study classified engaging in these behaviors as having sex. This difference may due to our use of a sample of undergraduates enrolled in a human sexuality course at a large western U.S. university. In general, students enrolled in a human sexuality class probably have many of the same attitudes that other college students do about vaginal and oral intercourse, which occur more frequently than anal intercourse. Moreover, attitudes towards vaginal and oral intercourse are less likely to be influenced by the content of a human sexuality course. On the other hand, students enrolled in a human sexuality class may have more liberal attitudes toward anal intercourse, which occurs more infrequently, because they are more liberal in general than students who would not choose to enroll in a sexuality class and/or because attitudes toward anal intercourse are more likely to be influenced by course content.

Our findings for vaginal and oral intercourse were quite similar. For both of these activities, participants believed that Susie would have a broader definition of sex than Jim, and that an act would be more likely to be considered sex by Jim and Susie if it culminated in orgasm. Moreover, the person who experienced the orgasm would be more likely to consider the behavior to be sex. In the particular case of vaginal intercourse, the participants indicated that Jim would be more likely to consider it sex if he orgasmed; however, Susie's outcome would not affect his labeling of the act as sex. Parallel results were obtained for Susie. (If neither partner had an orgasm, participants indicated that it was more likely that Susie would label the behavior as sex than would Jim, lending support to the cultural belief that orgasm is less important to women.) Likewise, with oral intercourse the recipient would be more likely than the partner who was performing the act to consider the behavior as sex, especially if the recipient had an orgasm. This finding--that participants believed that the person receiving oral sex would be more likely to consider the behavior as sex than would the partner performing the act--may be related to the finding that orgasm matters to people's definitions of sex. Individuals who stimulate their partners may not consider the activity to be sex because they are unlikely to experience orgasm themselves (unless reciprocal stimulation is provided).

The findings for anal intercourse deviated slightly from this pattern. Overall, participants indicated that Jim and Susie would not differ in whether or not they considered this behavior to be sex. Participants also indicated that whereas Jim would be more likely to consider anal intercourse to be sex if he had an orgasm, Susie would be as likely to consider the behavior sex whether she did or did not have an orgasm. However, the latter finding only held for male participants: female participants indicated that having an orgasm would influence how Susie classified anal intercourse. The overall results suggest that participants believed that it was Jim's orgasm, or lack thereof, that was the main factor in labeling anal intercourse as sex.

Contrary to our hypotheses, there were very few interaction effects involving the gender of the study participants. We had expected that some cultural beliefs would influence male and female participants differently. If orgasm is more important to men, for instance, we would expect to find having (or not having) an orgasm differentially reflected in male and female participants' views of how Jim would classify acts that did or did not result in orgasm. With the exceptions noted above, however, we observed essentially no interactions that involved participant gender. We speculate that the lack of interactions may be due to the convergence of men's and women's views of sexual behavior during the latter 1980s and early 1990s (Oliver & Hyde, 1993). Future studies in the same vein could test the effect of the media on sexual attitudes by comparing the results of a study that uses both an adolescent sample that was exposed to the media throughout their childhood, and an older sample that came of age prior to the 1980s.

The results of this study have obvious implications for the wording of questionnaire items that are used in sexual behavior surveys. Our findings suggest that most people include vaginal and anal intercourse in their definition of sex, but are less likely to count oral intercourse as sex. Thus, survey researchers who aim to measure sexual behavior may over- or underestimate the sexual activity level of their sample if respondents are not questioned about vaginal, anal, and oral sex separately. For example, suppose a survey asks respondents to report their overall level of sexual activity. A man who engages in oral intercourse, but not in vaginal or anal intercourse, may report that he is not sexually active if his definition of sex does not include oral intercourse. In contrast, a woman who includes oral intercourse in her definition of sex may overestimate the frequency of her recent sexual activity, possibly leading researchers to believe that she is riskier than she actually is.

The present study has several limitations. First, about one third of the college sample identified themselves as Asian American or Pacific Islander and another third identified themselves as White, and all were college students. We did not include race/ethnicity as a variable in our models, because we simply did not have a sufficient number of cases to fill each cell of the design. The results of the study, therefore, should be generalized with caution, particularly to individuals of different ethnic backgrounds and to individuals of lower educational levels. Our data are generalizable, however, to the extent that the questionnaire tapped participants' knowledge of broad cultural norms about the meaning of the word sex. Second, the names "Susie" and "Jim" are not necessarily equivalent, in that Jim might be considered a more neutral target name than Susie. Participants may have perceived someone who calls herself "Susie" differently than someone who, for example, calls herself "Sue" and this could have influenced their responses. Third, our sample consisted of undergraduates taking a human sexuality class at a large western U.S. university. As noted previously, participants in our sample may have had more liberal attitudes about anal sex than the general population.

Our use of scenarios instead of asking participants directly what they thought of the different sexual behavior terms is both a strength and a weakness. We assumed that the use of this methodology would minimize self-presentation bias: Because Jim and Susie served as proxies for participants, participants did not have to worry about how they would be perceived or about how to present themselves in the most favorable light. However, as a result of this approach, we can only discuss what participants believed about the sexual behavior of the two individuals in the scenarios, the fictional Jim and Susie. Because this situation may not reflect participants' own sexual scripts, the results may not be generalizable to the beliefs that participants have about their own behaviors.

Given the paucity of research on the meaning of sexual behavior terms, there are many avenues for future research. Because our sample size precluded us from examining ethnic differences in the definition of the word sex, our results should be replicated with larger samples of ethnically diverse individuals. In particular, a sample should be drawn from those age groups and ethnic groups at greatest risk for sexually transmitted diseases. Because the meaning of the word sex is likely to vary for different ethnic groups (Mays & Cochran, 1990), research is needed that compares the responses of different ethnic groups. In particular, future work should compare sexual behavior definitions among African American and Hispanic individuals, because HIV infection continues to increase within these groups (Centers for Disease Control and Prevention, 1998; Wortley & Fleming, 1997). Moreover, the meaning of other commonly used sexual behavior terms, such as sexual partner, needs to be examined. For example, many researchers use questionnaire measures to assess numbers of sexual partners, but respondents may not consider individuals with whom they "only" engaged in oral sex to be sex partners. In addition, some respondents may consider as sex partners only those with whom they have had a lengthy intimate relationship, and may not consider as partners those with whom they have engaged in sexual activity only one time.

In summary, the results presented here suggest that the meaning of sex may be more complex than typically is reflected in sexual behavior surveys. Sexuality researchers are therefore advised to choose questionnaire terms carefully, to be specific when defining terms, and to investigate the meaning of sex within the participants' culture(s). Accurate measurement of human sexual behavior is difficult, but the challenges it presents are not insurmountable if sufficient effort is put into careful questionnaire design (Abramson & Pinkerton, 1996).

(1) We also tested for differences between participants who did and did not report English as a first language. Overall, participants who did not speak English as their first language (M = .79) made judgments for Jim and Susie that were comparable to the judgments made by those for whom English was their first language (M = .77; z = .32, p = .75). Moreover, the pattern of responses associated with the three behaviors (i.e., vaginal, oral, and anal intercourse) was the same across both groups of English speakers (z = 4.66, p =. 10).

REFERENCES

Abramson, P. R., & Pinkerton, S. D. (1995). With pleasure: Thoughts on the nature of human sexuality. New York: Oxford University Press.

Abramson, P. R., & Pinkerton, S. D. (1996). Electrons and sex. Science, 273, 1155-1156.

Baker, P., & Harris, J. F. (1998, August 18). Clinton admits to Lewinsky relationship, challenges Starr to end personal `prying.' The Washington Post, p. 1A.

Berk, R., Abramson, P. R., & Okami, P. (1995). Sexual activities as told in surveys. In P. R. Abramson & S. D. Pinkerton (Eds.), Sexual nature/sexual culture (pp. 371-386). Chicago: University of Chicago Press.

Bogart, L. M., Pinkerton, S. D., Cecil, H., Myaskovsky, L., Wagstaff, D. A., & Abramson, P. R. (1999). Attitudes toward and definitions of having sex [Letter]. Journal of the American Medical Association, 282, 1917-1918.

Broder, J. M. (1998, March 7). Sifting through divergent sources for the truth about a relationship. New York Times, p. 8A.

Campos, A., & Sueiro, E. (1991). Stability in imagery scores. Perceptual and Motor Skills, 72, 411-414.

Catania, J. A., Binson, D., Canchola, J., Pollack, L. M., Hauck, W., & Coates, T. J. (1996). Effects of interviewer gender, interviewer choice, and item wording on responses to questions concerning sexual behavior. Public Opinion Quarterly; 60, 345-375.

Catania, J. A., Binson, D., van der Straten, A., & Stone, V. (1995). Methodological research on sexual behavior in the AIDS era. Annual Review of Sex Research, 6, 77-125.

Catania, J. A., Gibson, D. R., Marin, B., Coates, T. J., & Greenblatt, R. M. (1990). Response bias in assessing sexual behaviors relevant to HIV transmission. Evaluation and Program Planning, 13, 19-29.

Cecil, H., & Zimet, G. (1998). Meanings assigned by undergraduates to frequency statements of condom use. Archives of Sexual Behavior, 27, 493-505.

Centers for Disease Control and Prevention. (1998). HIV/AIDS surveillance report, 9. Atlanta: U.S. Department of Health and Human Services.

Diggle, P. J., Liang, K.-Y., & Zeger, S. L. (1994). Analysis of longitudinal data. Oxford, England: Clarendon Press.

Fisher, M. (1998, February 18). The hot talk of radio: White House crisis breathes new life into today's conservative mainstays. The Washington Post, p. 1B.

Gerstenzang, J., & Shogren, E. (1998, December 9). The impeachment hearings: White House releases broad challenge to allegations. The Los Angeles Times, p. 23.

Hosmer, D. W., Jr., & Lemeshow, S. (1989). Applied logistic regression. New York: John Wiley & Sons.

Janofsky, M., & Van Natta, D., Jr. (1998, February 14). Intern's friend said to recount talk of affair. New York Times, p. 1A.

Kinsey, A. C., Pomeroy, W. B., Martin, C. E., & Gebhard, P. H. (1953). Sexual behavior in the human female. Philadelphia: W.B. Saunders.

Kuczynski, A. (1998, February 1). In offices, an excuse to mention s*x. New York Times, Sec. 9, p. 1.

Kutner, N. G., & Brogan, D. (1974). An investigation of sex-related slang vocabulary and sex-role orientation among male and female university students. Journal of Marriage and Family, 36, 474-484.

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

Liang, K.-Y., & Zeger, S. L. (1986). Longitudinal data analysis using generalized linear models. Biometrika, 73, 13-22.

Maines, R. P. (1999). The technology of orgasm: "Hysteria," the vibrator, and women's sexual satisfaction. Baltimore: Johns Hopkins University Press.

Masters, W. H. (1966). Human sexual response. Boston: Little, Brown.

Mays, V. M., & Cochran, S. D. (1990). Methodological issues in the assessment and prediction of AIDS risk-related sexual behaviors among black Americans. In B. Voeller, J. M. Reinisch, & M. Gottleib (Eds.), AIDS and sex: An integrated and biobehavioral approach (pp. 97-120). New York: Oxford University Press.

McDermott, R. J., Drolet, J. C., & Fetro, J. V. (1989). Connotative meanings of sexuality-related terms: Implications for educators and other practitioners. Journal of Sex Education and Therapy, 15, 103-113.

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

Robinson, I. F., Balkwell, J. W., & Ward, D. M. (1980). Meaning and behavior: An empirical study in sociolinguistics. Social Psychology Quarterly, 43, 253-258.

Sanders, J. S., & Robinson, W. L. (1979). Talking and not talking about sex: Male and female vocabularies. Journal of Communication, 29, 22-30.

Sanders, S. A., & Reinisch, J. M. (1999). Would you say you `had sex' if ...? Journal of the American Medical Association, 281, 275-277.

Shogren, E., Willman, D., & Cooper, R. T. (1998, January 27). Clinton strongly denies having sex with intern. The Los Angeles Times, p. 1.

Sonenstein, F. L., Ku, L., & Pleck, J. H. (1997). Measuring sexual behavior among teenage males in the United States. In J. Bancroft (Ed.), Researching sexual behavior. Methodological issues (pp. 87-105). Indianapolis, IN: Indiana University Press.

Stata Corporation (1997). Stata Statistical Software: Release 5.0. College Station, TX: Stata Corporation.

Wagstaff, D. A., Abramson, P. R., & Pinkerton, S. D. (2000). Research in human sexuality. In L. T. Szuchman & F. Muscarella (Eds.), The psychological science of sexuality: A research-based approach to understanding human sexuality, (pp. 3-59). New York, NY: John Wiley & Son.

Walsh, R. H., & Leonard, W. M. (1974). Usage of terms for sexual intercourse by men and women. Archives of Sexual Behavior, 3, 373-376.

Weir, S. S., Roddy, R. E., Zekeng, L., Ryan, K. A., & Wong, E. L. (1998). Measuring condom use: Asking "Do you or don't you" isn't enough. AIDS Education & Prevention, 10, 293-302.

Wortley, P. M., & Fleming, P. L. (1997). AIDS in women in the United States: Recent trends. Journal of the American Medical Association, 278, 911-916.

Zeger, S. L., & Liang, K.-Y. (1986). Longitudinal data analysis for discrete and continuous outcomes. Biometrics, 42, 121-130.

Manuscript accepted November 22, 1999

Laura M. Bogart Kent State University

Heather Cecil University of Alabama at Birmingham

David A. Wagstaff Pennsylvania State University

Steven D. Pinkerton Medical College of Wisconsin

Paul R. Abramson University of California-Los Angeles

This research was supported by center grant P30-MH52776 and postdoctoral training grant T32-MH 19985 from the National Institute of Mental Health, center grant P50-D10075 from the National Institute of Drug Abuse, and training grant MCJ-9040 from the Maternal and Child Health Bureau, U.S. Department of Health and Human Services. The authors extend their appreciation to Larissa Myaskovsky, Christi Charnock, Michelle Gray-Bernhardt, Mike DeMaster, and Mary Alunkal for help with data collection and data entry.

Address correspondence to Laura M. Bogart, Ph.D., Department of Psychology, 118 Kent Hall, Kent State University, Kent, OH 44242; e-mail: [e-mail].