Casual sex on spring break: intentions and behaviors of Canadian students.

Although numerous studies have been conducted on the sexual behavior of young people, there is little research on sexual behavior occurring in specific contexts. One context that has been ignored by researchers is the North American spring break holiday, a one-week break in the school calendar in late February or early March. Approximately one million U.S. students participate in some form of spring break vacation (Josiam, Clements, & Hobson, 1995). Spring break vacations are also popular among Canadians, with thousands of students heading to popular vacation spots (S. Cox, Inter-Campus Programs, personal communication, September, 1995).

Mewhinney, Herold, and Maticka-Tyndale (1995), using focus groups and interviews with Canadian students who had traveled to Florida for spring break, found the key elements of a spring break vacation to include a group holiday with friends traveling and rooming together, a perpetual party atmosphere, high alcohol consumption, sexually suggestive contests and displays, and the perception that casual sex is common. Overall, there is the perception that sexual norms are far more permissive on spring break vacation than at home, providing an atmosphere of greater sexual freedom and the opportunity for engaging in new sexual experiences. Smeaton and Josiam (1996) reported similar findings from their survey of U.S. students on spring break vacation in Panama City Beach, Florida.

The behavior patterns found on spring break have also been found among nonstudent samples of holiday travelers. In their review of the tourism and vacation literature, Herold and van Kerkwijk (1992) identified the characteristics of vacations that were conducive to casual sex activity as a sense of freedom from at-home restrictions, a relaxation of inhibitions, a focus on having a good time, and high alcohol consumption.

We examined factors that might influence university students to engage in coitus with a new partner while on spring break. For the purpose of this paper, we refer to this behavior as casual sex: This was the term used by students in a preliminary study (Mewhinney et al., 1995) to refer to the type of sex engaged in while on spring break. The students portrayed spring break sexual partnerships as initiated rapidly, often within hours of meeting, and as temporary, not lasting beyond the spring break period. Given our focus on new, casual partnerships, we excluded from our analysis those young adults who were sexually active on spring break with a relationship partner from home or with someone whom they had known before spring break.

Triandis's theory of interpersonal behavior (TIB; 1977, 1980, 1994) was selected to guide data collection and analysis because it includes peer influences and situational characteristics in explaining behavior, both of which were identified in preliminary research (Mewhinney et al., 1995) as important influences on spring break sexual activity. The TIB belongs to the school of cognitive models that includes the theory of reasoned action (TRA; Ajzen & Fishbein, 1980) and the theory of planned behavior (TPB; Ajzen, 1985). These theories explain the influence of attitudes and norms on intentions, with these intentions, in turn, directly influencing behaviors. Triandis's model goes beyond the others by examining how factors other than intentions influence behavior, and by more fully specifying the factors that influence intentions. The TIB has proven useful in understanding complex behaviors, particularly those that may be influenced by the social and/or physical environment (e.g., in sexuality research, see Boyd & Wandersman, 1991; Godin, et al., 1996).

Using the framework of the TIB, we predicted that whether a young adult on spring break in Daytona Beach would engage in casual sex would be influenced by an expectation or intention to engage in coitus with a new partner on spring break, prior experience with casual sex, and an environment conducive to new coital partnerships. The first factor, intention, has received considerable attention in the social psychological literature (e.g., Fishbein & Jaccard, 1973; Jorgensen & Sonstegard, 1984; White, Terry, & Hogg, 1994). Although factors that might intervene in the intention-behavior link have been discussed (e.g., Randall & Wolff, 1994), only Triandis has explicitly focused on variables such as prior experience and situational conditions that may facilitate, impede, or replace intentions as determinants of behavior.

Previous research supports the inclusion of these additional variables in explanations of sexual behavior. For example, researchers have found that for individuals with a greater number of past coital partners, erotic cues (e.g., arousal and situational cues) have a stronger influence on whether coitus occurs than does the quality of a relationship (Christopher & Cate, 1984; D'Augelli & D'Augelli, 1977). The TIB would make a similar prediction, although prior experience is conceptualized somewhat differently. In the TIB, prior experiences that closely approximate the current sexual situation, such as having coitus with someone within hours of meeting him or her, can replace intentions and independently influence a behavior. The influence of prior experience is strongest when the new event closely parallels the prior experiences (for this study, if the prior experience of casual sex occurred while on a previous spring break vacation in Daytona Beach) and when there are multiple instances of that prior experience.

Situational conditions work with intentions and prior experience to facilitate the likelihood of particular behaviors by providing opportunity and cues that contribute to an individual's erotic expectations (see also D'Augelli & D'Augelli, 1977; Rook & Hammen, 1977). In the TIB, situational conditions that are conducive to actions are specific to the action and situation of interest. Thus, for this study, we must consider conditions associated with being on vacation (Herold & van Kerkwijk, 1992).

Triandis suggests that intentions and prior experiences are the dominant influences on behavior, with situational conditions acting as mediators that facilitate or impede behaviors. Thus, if the situation prohibits or impedes casual sex, prior behavior and/or intentions will not result in the predicted behaviors. If conditions conducive to casual sex are present, then its realization will depend on prior experience with casual sex and/or intentions.

The TIB also provides an explanation of how intentions are formed. As in other cognitive models (e.g., TRA and TPB), personal attitudes and social norms are presented as influencing the formation of intentions or plans. However, the TIB more fully specifies the personal and social dimensions. Personal attitudes include affective and evaluative components. The affective component of attitudes toward casual sex involves the feelings that one anticipates as part of casual sex (e.g. pleasure, elation, fear, disgust). The evaluative component of attitudes is the cognitive evaluation of the probable consequences of casual sex (e.g., desirable, undesirable, good, bad). These two forms of attitudes are specified separately because each has a distinct influence on the formation of intentions. For example, a young adult may anticipate feelings of pleasure and excitement at the thought of casual sex on spring break but may judge the potential results of engaging in casual sex to be undesirable.

Social norms emanating from an individual's reference group are the second component influencing the formation of intentions. Here the TIB's predictions are similar to those of reference group theory (Christopher & Cate, 1984; Winslow, Franzini, & Hwang, 1992). As with personal attitudes, Triandis specifies several dimensions of reference group influence: (a) agreements or promises that are formed between friends (pacts either to engage or not to engage in casual sex); (b) the perceived norms and expectations of one's immediate reference group (subjective social norms); (c) beliefs about what is appropriate for a member in one's position or status (role beliefs); and (d) the internalized personal standards or moral codes (personal normative beliefs). The first two dimensions represent the social group as the point of reference. The second two dimensions represent the individual's transformation of social norms into self-expectations and standards and take the self as the point of reference. As with affective and evaluative components of attitudes, each of these may exert different pressures on the individual. For example, there may be pressure to fulfill the pacts or agreements made with friends or to act in accordance with the norms and expectations set by the reference group. There may also be pressure to act in a way that is considered appropriate to one's age, gender, or relationship status, or that is in accord with one's personal standards or moral code.

The conceptualization of the components of the theory of interpersonal behavior, the relationships among them, and the inclusion of new components have evolved over time (Triandis, 1977, 1980, 1994). Figure 1 presents the full structural model for the theory of interpersonal behavior as used in this study. In accordance with Triandis's suggestion that additional factors may be tested for their potential contribution to the explanatory power of the model, we have added situational expectations to the prediction of intentions and two measures of peer influence to the prediction of behavior.

[Figure 1 ILLUSTRATION OMITTED]

Situational expectations are the expectational predecessors of situational conditions. These are conceptualized as the individual's expectations about whether spring break experiences will be conducive to casual sex. It is similar to the control belief concept in the TPB (Ajzen, 1991), a variable that is indicative of the degree to which a behavior is under the control of an individual. We suggest that when there are situational expectations conducive to casual sex, the individual expects that casual sex will occur. Triandis models participation in activities (situational conditions) as directly influencing the behavior. The addition of the expectation of participation as an influence on intentions permits us to examine more precisely the impact of the spring break environment, both as anticipated and as experienced, on casual sex.

We also added two peer influences to the explanation of casual sex activity: pacts and role modeling. These fit within Triandis's conceptualization of social norms and relate to our specific interest in the spring break environment. In preliminary research (Mewhinney et al., 1995), we found that a spring break vacation was a group activity. By including characteristics of peer influence that are part of the spring break experience, we are able to examine the direct effect of these aspects of the environment on engaging in casual sex.

We completed this study in two parts. First, in the prebreak study, we examined expectations and intentions before a spring break trip and tested the portion of the TIB that explains the formation of intentions. Second, in the Daytona study, we examined experiences on the trip and tested the portion of the TIB that explains behavior. The methodology, results and discussion specific to each study are presented separately. The discussion following these presentations addresses portions of the results that bridge both studies.

STUDY 1: PREBREAK STUDY

Methods

Recruitment of participants. In February, 1996, students from four universities in southern Ontario who were planning a one-week spring break vacation in Daytona Beach, Florida were recruited before their trip through travel agents, advertisements in campus papers, posters on campus, class announcements, and campus information booths. Self-administered questionnaires were given or mailed to students following telephone or face-to-face contact with a research assistant. In addition, questionnaires were distributed and collected by research assistants at the beginning of chartered bus rides to Daytona Beach. In this prebreak sample, 151 surveys were completed.

Questionnaire development. Triandis's (1977, 1980) theory includes a measurement model that operationalizes each concept in a manner that is specific to the time, context, and behavior of interest. Triandis provides the structure and form of measurement for each concept (e.g., personal attitudes are measured using semantic differential scales, social norms are measured using Likert scales), but the specific items and indicators must be established based on preliminary elicitation research. Elicitation research was conducted following Triandis's guidelines (Ajzen & Fishbein, 1980; Triandis, 1977, 1980). Students who had traveled to Florida for spring break before 1996 participated in semistructured, open-ended group and individual interviews. Concepts were operationalized based on content analysis of the interviews.

A draft of the prebreak questionnaire was tested and refined in a two-week test-retest procedure with a sample of 10 men and 10 women who anticipated going on a spring break vacation. Items that did not meet criteria set for test-retest reliability (kappa coefficient of at least 0.70) or that were more weakly correlated with other items measuring the same construct (evaluated using Cronbach's alpha) were eliminated from the final questionnaire. The specificity of items used in the scalar measures of constructs was tested using confirmatory factor analysis to ensure that items loaded only on their designated constructs. Construct validity was assessed by examining correlation matrices to verify that scalar measures correlated with criterion factors of gender, age, and prior coital experience in a manner consistent with prior research in the field.

The prebreak surveys included questions about sexual history, spring break plans and expectations, and the constructs used to predict intentions to engage in sexual activity. The final prebreak questionnaire could be completed in an average of 20 minutes.

Measurement of constructs. Seven-point semantic differential scales consisting of adjective pairs used by students in the elicitation research to describe the experience and consequences of casual sex on spring break measured affective and evaluative attitude components. Participants were asked to rate their feelings about and evaluation of the consequences of "having sex with someone you meet on spring break." To ensure that this was a measure of attitudes toward coitus and not toward condom use, we asked participants to make this assessment without consideration of whether condoms were used. The adjective pairs measuring feelings when thinking about casual sex (affective component) were fun-loving/serious, exciting/dull, pleasant/ unpleasant, adventuresome/ordinary. The adjective pairs measuring cognitive evaluations of the potential consequences of casual sex were good/bad, smart/stupid, responsible/irresponsible. Mean scores of the pairs for each attitude were used to measure the affective attitude and cognitive evaluations toward casual sex on spring break. Cronbach's alpha for the affective and cognitive measures was 0.76 and 0.82, respectively.

The four dimensions of social norms--personal norms, subjective social norms, role beliefs, and pacts--were measured. The strength of personal norms supporting casual sex was measured by the mean score of responses to three questions, each with a 7-point bipolar response: (1) "When on Spring Break ... you feet you should have sex with someone you meet there if you want to"; (2) "... you would feel guilty if you have sex with someone you meet there" (reverse-coded); (3) "... it would be against your values to have sex with someone you meet there" (reverse-coded). Cronbach's alpha for this measure was 0.83.

Subjective social norms, or the perception of how much one's reference group approves or disapproves of his or her casual sex activity, was measured using the mean score of 7-point bipolar responses to three questions. Respondents indicated how likely or unlikely it was that each of three groups of referent others (male friends, female friends, and friends who went with them to Daytona Beach) would think that they should have sex if they met someone who was appealing. Cronbach's alpha for subjective social norms was 0.85.

Role beliefs, or the strength of beliefs about whether casual sex is appropriate for someone given his or her status or position, was measured using the mean score of 7-point bipolar (strongly agree (1) to strongly disagree (7)) responses to three questions. Respondents were asked whether having sex with someone met on spring break was "OK for someone of my gender ... of my age ... of my relationship status." Cronbach's alpha for role beliefs was 0.79.

To measure pacts, we asked participants whether they had made promises, agreements, or pacts with the friends who would accompany them to Daytona either to engage or not to engage in casual sex. The presence and nature of these pacts or agreements was effect coded as -1 for an agreement not to have sex, 0 for no pact or agreement, and 1 for an agreement to have sex.

Because situational expectations closely parallel Ajzen's (1991) control belief, Ajzen's method for operationalizing control beliefs was used to assess situational expectations. During the elicitation phase, students identified specific situations or experiences that had occurred in Daytona Beach that they thought were conducive to or impeded engaging in coitus with a new partner while in Daytona. Ten experiences were identified and used in measuring the influence of anticipated situations on engaging in casual sex: partying, being in a break-loose mood, drinking alcohol, getting drunk, watching contests such as "hot body" and "wet t-shirt," dancing dirty, trying to "pick someone up," someone trying to "pick up" the respondent, the appearance that everyone was having sex, and someone wanting to have sex with the respondent. Participants used a 4-point scale ranging from never (1) to frequently (4), to rate the frequency with which they expected to be involved in each of the experiences. The perceived degree of influence of each experience on whether they engaged in casual sex was scored on a 7-point bipolar scale (strongly agree (1) to strongly disagree (7)) in response to the question "How likely is it that you would have sex with someone you meet on Spring Break if...." The score for each experience or situation was constructed from the product of the frequency rating and the degree-of-influence rating. Cronbach's alpha for the total situational expectation score consisting of the mean of the 10 items' scores was 0.96.

To measure the criterion variable, we asked participants how often they expected or intended to engage in casual sex on spring break (defined as sexual intercourse with someone they had met on spring break). Answers ranged from never (1) to frequently (4).

Data Analysis. In the TIB, personal and status characteristics, such as gender, are external to the model: Their influence on intentions and behaviors is mediated through their influence on attitudes, norms, and experiences. Research on sexuality, however, has documented interactions between gender and attitudes, norms, experiences, and behaviors, suggesting that gender may also influence intentions and behaviors from within the model. These interactions include the way peer influences affect attitudes and behaviors (e.g., Christopher & Cate, 1984), the way attitudes and behaviors are linked (e.g., Oliver & Hyde, 1993), the way prior experiences influence future behaviors (Christopher & Roosa, 1991), and the characteristics of relationships that are necessary for coital interaction (e.g., Christopher & Roosa, 1991). These findings necessitate a consideration of potential gender interactions when designing an analysis plan. In this study, men's and women's covariance matrices for all variables used in the multivariate procedures were tested for significant differences using LISREL VI (Joreskog & Sorbom, 1986). This procedure permitted us to identify the number and location of interactions. With the exception of the covariance of subjective social norms and intentions, there were no significant differences between the covariance matrices for men and women. The interaction of gender with subjective social norms was accommodated in this analysis by using gender-specific terms for men's and women's subjective social norms. Because there was only one interaction between gender and an independent variable, the remainder of the data for men and women were aggregated.

During exploratory research, students argued that the interpretation of spring break activities was influenced by one's peer group. If friends endorsed casual sex, for example, then a student expected and interpreted the activities on spring break as conducive to casual sex. If friends did not endorse casual sex, the same activities would not necessarily be seen as leading to or facilitating casual sex. The possible mediator role of peers with respect to situational expectations was tested using four regression analyses. In each, intention was regressed on situational expectations along with one measure of social norms and the interaction between expectations and social norms. The interaction between situational expectations and pacts was the only significant interaction. Therefore, it was included in the final analysis.

Including several dimensions of complex constructs such as personal attitudes and social norms in an analysis allows us to identify the contribution and relative strength of each dimension in explaining the dependent variable. However, because dimensions are expected to be closely related to one another, they present the possibility of multicollinearity. The presence of multivariate collinearity was tested using measures of tolerance and variance inflation, comparisons of the eigenvalues of the variables' uncentered cross-products matrix, and examination of tile proportion of variance of each independent variable associated with each eigenvalue (Smith & Sasaki, 1979). These supported the conclusion that each predictor variable represented a sufficiently independent portion of variance in intentions to conduct multivariate analyses. Because the inclusion of interaction and gender-specific terms in the proposed analysis further heightened the possibility of multicollinearity, all independent variables measured at the interval level were centered around their mean values as suggested by Smith and Sasaki (1979) and Aiken and West (1991). The measure for agreements with friends (pact) retained its original unweighted effect coding.

The model for intentions was tested using hierarchical ordinary least squares regressions. The predictor variables in the original TIB were entered as a block in the first regression followed by the addition of situational expectations and the interaction between pacts and situational expectations.

Results

Profile of the Sample. Participants were 66 males and 85 females aged 18-31 (77% were ages 21-23). All participants were White and identified themselves as heterosexual and single. Thirty-five percent of men and 38% of women reported they were in a long-term relationship.

Spring break expectations. As seen in Table 1, this was the first spring break trip and the first trip to Daytona Beach for most students. Consistent with descriptions in the elicitation research, spring break in Daytona is a social event, with 82% of men and 88% of women accompanied by friends. However, this is not necessarily a vacation for couples: Forty-eight percent of the men and 68% of the women who were in relationships reported that their partners would not accompany them to Daytona.

Table 1. Percentage of Men and Women With Different Sexual and Spring Break Experiences and Intention

Variable                                          Males      Females

Prior Sexual Experience
      Previous coital experience                   95%         92%
      Ever coitus within 24 hrs.
      of meeting partner                           65%(++)     34%
Spring Break Experience
      1st spring break trip                        73%(++)     48%
      1st Daytona Beach trip                       73%         87%
Plan to Go With Friends                            82%         88%
Plan to Go with Relationship Partner(a)            18%         12%
Pacts or Agreements Formed With Friends:
      Not to engage in casual sex                   8%         15%
      To engage in casual sex                      46%(+++)     1%
Coital Intentions for Spring Break
      Intend coitus with relationship partner      18%          9%
      Intend coitus with acquaintance(b)           62%(+++)    25%
      Intend coitus with someone met on spring
        break(c)                                   76%(+++)    19%
       Do not intend to engage in coitus            8%(+++)    61%

Note. Sample comprised of 66 males and 85 females.

(a) 48% of men and 68% of women are in a relationship and report their partners will not accompany them.

(b) Someone the participant knew before coming on spring break but with whom the participant had not engaged in coitus.

(c) Casual sex.

(++) Difference between percentages significant at p [is less than] .01 (chi-square test).

(+++) Difference between percentages significant at p < .001 (Chi-square test).

Almost all who traveled to Daytona Beach had prior coital experience. The median number of prior coital partners was six for men and three for women. Fewer women than men intended to engage in casual sex on spring break. Women were less likely than men to form pacts or agreements with friends about casual sex; but when they did, women made agreements with friends not to engage in sex while in Daytona (15%), whereas men formed agreements to have sex (46%). In the elicitation interviews, students described agreements among women to "pull your friend out of a situation" if it appeared she might become sexually involved; men described competitive pacts to see who could "get laid" the most. The gender pattern seen in Table 1 is replicated in the mean scores on the scalar measures of attitudes, norms, and expectations related to casual sex. Men's scores were higher than women's on each measure, indicating that men's attitudes and beliefs were more supportive of casual sex on spring break than were women's.

Intentions to engage in casual sex. The Pearson product moment correlations between all variables used in the analysis ranged from .14 to .71. Those between intentions and predictor variables were positive and ranged from. 18 (attitudes with intentions) to .71 (situational expectations with intentions). Table 2 reports the results of the ordinary least squares regression of intention to engage in casual sex on centered predictors. Both the TIB (Model 1) and the enhanced TIB (Model 2) explained sizable portions of the variance in intentions ([R.sup.2] = .70 and .74 respectively). The enhanced TIB explained a significantly larger portion of the variance than the original TIB, F(2,134) = 12.51, p [is less than] .001. The strongest influence on men's intentions was their perceptions of the degree of endorsement they would receive from their friends for participation in casual sex while on spring break. Women, however, were not influenced by the perceived endorsement of others. For both men and women, all indicators except personal standards and role beliefs had a significant influence on the formation of intentions. Thus, it was the social norms using the peer group as the point of reference rather than those using the self that influenced intentions. The results of the interaction between pacts and situational expectations coincided with the comments students made during the elicitation phase. The positive effect of situational expectations was enhanced when there was a pact to engage in coitus with a new partner while on spring break. However, when there was a pact not to engage in coitus with a new partner, the association disappeared--the pact counteracted the anticipated environmental influence.

Table 2. Ordinary Least Squares Regression Coefficients for Intentions to Engage in Casual Sex on Centered Predictors

                                           Model 1: TIB

Variable                            B              S.E.    B

Affective Component                 0.12(**)       0.04    0.15
Evaluative Component                0.18(***)      0.04    0.27
Personal Standard                   0.05           0.04    0.10
Role Belief                         0.04           0.03    0.08
Subjective Social Norm: Women       0.05           0.04    0.06
Subjective Social Norm: Men         0.38(***)      0.05    0.45
Pact(a)                             0.39(***)      0.11    0.20
Situational Expectation
Pact x Situational Expectation(b)
Constant                            1.55           0.06
[R.sup.2]                                           .70

                                           Model 2: Enhanced TIB

Variable                            B              S.E.    B

Affective Component                 0.12(**)        0.04    0.15
Evaluative Component                0.16(***)       0.04    0.25
Personal Standard                   0.01            0.03    0.03
Role Belief                         0.03            0.04    0.06
Subjective Social Norm: Women       0.02            0.04    0.02
Subjective Social Norm: Men         0.27(***)       0.05    0.32
Pact(a)                             0.26(**)        0.11    0.14
Situational Expectation             0.13(**)        0.05    0.20
Pact x Situational Expectation(b)   0.21(***)       0.06    0.20
Constant                            1.49            0.06
[R.sup.2]                            .74

Notes. N = 151. All independent variables except Pact are centered around their mean.

(a) Effect coded -1 = pact not to engage in coitus; 0 = no pact; 1 = pact to engage in coitus.

(b) Interaction between pacts and situational expectations.

(**) p < .05.

(***) p < .001.

STUDY 2: DAYTONA STUDY

Methods

Recruitment. One male and one female graduate student who had experience and training in research methods and who were in the age range of students on the spring break trip collected the data for the Daytona study during the 1996 spring break season. Three methods of recruitment were used: on-site at Daytona Beach, on buses returning home, and through the mail. On-site recruitment procedures were similar to those followed by Smeaton and Josiam (1996) in their study of spring break in Panama City Beach, Florida, and by Eiser and Ford (1995) in their study of a British beach resort. Participants were recruited on beaches and pool decks between 11 a.m. and 4 p.m. from Wednesday to Saturday of each of the two weeks when Canadian students were in Daytona. This timing ensured that participants had been in Daytona for at least 2 days before completing the questionnaire and that they were likely to be sober, as the heaviest alcohol consumption began in the late afternoon. Students were approached by the research assistants who explained the survey procedures and ethical guidelines and verified that potential participants were from Canada (all were from eastern Ontario). Those who agreed to participate were provided with a pencil and the questionnaire on a clipboard. Up to five questionnaires were distributed at one time. Research assistants ensured that each participant completed the questionnaire privately and out of the range of vision of others. Of the 494 students approached in this manner, 484 completed questionnaires, for a response rate of 98%. This closely matches Smeaton and Josiam's (1996) response rate of 99%.

In the second recruitment strategy, questionnaires were distributed to students for completion at the beginning of the return bus trip home. Research assistants explained the purpose and procedures of the study, and one person on the bus was delegated to collect and return completed surveys in a pre-addressed envelope. It was not possible to calculate a valid response rate: A number of students declined to complete a survey because they had already done so, but 119 questionnaires were completed in this manner.

In the final recruitment strategy, questionnaires were mailed to the 151 students who had completed prebreak surveys (all had agreed to receive a post-break survey) within one week of returning home. These students were provided with an addressed, stamped return envelope and a separate stamped, addressed postcard. All who did not return a postcard received up to three phone reminders. Seventy-eight participants returned these questionnaires, for a response rate of 52%.

Participants' birthdates and university student numbers were used to match questionnaires and to eliminate duplicates. Three duplicates were found. In each case, the questionnaire completed after the longest time in Daytona was retained.

Comparison of surveys from different recruitment strategies. Responses on questionnaires using the three recruitment strategies were compared on 60 items, controlling for gender and duration of time in Daytona. Of the 60 comparisons, 3 were significant at p [is less than] .05. This number of significant comparisons would be expected by chance at this level of significance. Based on these results, the three sampling strategies were judged to produce similar results, and data were combined for the Daytona analyses.

Questionnaire development and measurement of constructs. The Daytona survey contained questions on sexual and spring break history and activities and experiences while in Daytona Beach. It could be completed in 10 minutes. Questions for this survey were developed using the same procedures described for the prebreak survey. Where questions addressed the same concepts in both surveys, wording was modified to accommodate the different time frame. Several constructs were unique to the Daytona survey: the spring break experience, prior casual sex experience, role modeling, and engaging in casual sex.

First, the degree of participation in spring break experiences conducive to casual sex was measured using mean scores of participants' ratings (on a 4-point scale ranging from never to frequently) of how often they participated in each of the ten experiences listed in the prebreak questionnaire as situational expectations. Cronbach's alpha was 0.82.

In the TIB, prior experience is measured as the number of times an individual has already engaged in the behavior of interest. Because few students had ever been to Daytona Beach for spring break vacation before this trip, the use of experiences specific to this environment was judged unfeasible. As an alternate indicator of similar prior experience, we examined whether an individual had ever engaged in coitus with someone they had known for less than 24 hours. This was scored dichotomously (0 = no, 1 = yes).

Participant reports of the proportion of friends who had participated in coitus while on spring break (0 = none; 1 = few; 2 = about half; 3 = most) were used as a measure of perceived role modeling of coital activity.

To measure the criterion variable we asked participants whether they engaged in coitus while on spring break and what their relationship was to the partner. For purposes of this analysis, a casual sex variable was created that was coded 0 if the respondent had not engaged in coitus while on spring break (N = 499) and 1 if the respondent had engaged in coitus with someone they met while on spring break (N = 94).

Data Analysis. The tests for equivalence of men's and women's covariance matrices and the collinearity diagnostics conducted for the prebreak analysis were replicated for the Daytona sample. Although men's and women's covariance matrices were not significantly different for the original TIB constructs, they were significantly different when the two peer influence variables were added. Consequently, separate regression analyses were conducted for each gender. No multicollinearity was identified, and there were no interaction terms in these regression models, Therefore, variables were not centered.

The models for coitus with someone who was met on spring break were tested separately for men and women using hierarchical logistic regressions with forced entry of blocks of variables. The predictor variables in Triandis's original model were included in the first model. Two variables were added in the second: proportion of friends who had engaged in coitus on spring break and the form of agreement or pact struck with friends about casual sex on spring break.

Results

Sample profile. Participants were 306 males and 375 females from six universities in Ontario, Canada. They ranged in age from 18 to 30 years with 69% clustering between ages 21 and 23. All participants were White and identified themselves as heterosexual and single. Twenty-eight percent of men and 35% of women reported that they were in a long-term relationship.

Spring break experiences. Table 3 reports responses to questions about sexual history and spring break intentions and experience. The gender differences identified in the prebreak sample were replicated here. Consistent with these differences, men reported participating in significantly more spring break activities conducive to casual sex (M = 2.96, SD = 0.58) than did women (M = 2.79, SD = 0.57; p [is less than] .05), and a significantly larger percentage of men's friends engaged in coitus on spring break (M = 0.22, SD = 0.24) than did women's friends (M = 0.12, SD = 0.21; p [is less than] .001). However, there were no significant differences in the coital activity that men and women reported on spring break. Seventy-four percent of students did not engage in coitus during the spring break trip. Those whose relationship partners accompanied them restricted coital activity to these partners. Most others engaged in casual sex.

Table 3. Percentage of Men and Women Reporting Sexual and Spring Break Experience and Intentions.

Variable                                           Males     Females

Prior Sexual Experience
  Previous coital experience                        91%         87%
  Ever coitus within 24 hrs of meeting partner      61%(++)     34%
Pacts or Agreements Formed With Friends
  Not to engage in casual sex                        5%(+++)    21%
  To engage in casual sex                           30%(++)      5%
Intend to Engage in Casual Sex                      55%(++)     11%
Coital Activity on Spring Break
  Coitus with relationship partner                   7%          8%
  Coitus with acquaintance(a)                        6%          4%
  Coitus with someone met on spring break(b)        15%         13%
  No coital activity                                72%         77%

Note. Sample comprises 306 males and 375 females.

(a) Someone the participant knew before coming on spring break but with whom the participant had not engaged in coitus, (b) Casual sex.

(+++) Difference between percentages significant at p [is less than] .001 (Chi-square test).

Fewer men engaged in coital activity with new partners than intended to (55% of the men in this sample intended to engage in coital activity, but only 15% did). For women, however, there was almost no difference in the percentages who intended and who engaged in coitus with a new partner. For those who engaged in coitus with a new partner, there was no significant gender difference in the number of partners. During spring break, 68% had one partner, 13% had two partners, and 19% had more than two partners.

Explaining participation in casual sex. When considered separately (Table 4), intentions to engage in casual sex, prior casual sex experience, participation in spring break activities, the proportion of friends who engaged in coitus while on spring break, and the formation of pacts about casual sex with friends were significantly different for those who engaged and did not engage in casual sex while on spring break. Having a relationship partner at home did not necessarily impede engaging in casual sex on spring break.

Table 4. Percentages and Mean Scores on Selected Predictor Variables, by Casual Sex Activity

                                     Coital Activity on Spring Break

                                           No Coital
Variable                                   Activity    Casual Sex

Women
 Intend to engage in casual sex             8%(+++)           31%
 Prior casual sex experience               23%(+++)           55%
 Spring break activities                  2.75(***)          3.28
 Proportion of friends
  engaged in coitus                       0.07(***)          0.38
 Pacts formed with friends
  No casual sex                            24%(+++)            8%
  Casual sex                                2%(+++)           19%
 Currently in a relationship(a)            31%                26%
Men
 Intend to engage in casual sex            50%(+++)           88%
 Prior casual sex experience               51%(+)             71%
 Spring break activities                  2.93(***)          3.41
 Proportion of friends
  engaged in coitus                       0.19(***)          0.38
 Pacts formed with friends
  No casual sex                             2%(++)             7%
  Casual sex                               27%(++)            48%
Currently in a relationship(a)             26%                18%

Notes. This table excludes data for students who engaged in coitus with someone they knew. Sample comprises 268 women and 205 men who did not engage in coital activity, and 48 women and 46 men who engaged in casual sex.

(a) None of these relationship partners were in Daytona Beach.

(***) p < .001 (t-test).

(+) p < .05 (Chi-square test).

(++) p < .01 (Chi-square test).

(+++)p < .001 (Chi-square test).

Table 5 reports hierarchical logistic regressions of the dichotomous measure of participation in casual sex first on the variables in the TIB, and then with the addition of two measures of peer influence. Regressions were conducted separately for women and men. In the first regression, using variables from the TIB, results were similar for men and women. Intentions and spring break experience significantly differentiated between those who engaged and did not engage in casual sex, with spring break experiences the stronger of the two predictors. The absence of a significant effect for prior casual sex experience could be related to the measurement of this construct. Because of the novelty of the Daytona Beach experience for the majority of students, prior experience was measured using any casual sex experience rather than the more specific measure of casual sex on spring break recommended by Triandis.

Table 5. Hierarchical Logistic Regression Coefficients for Casual Sex(a) on Predictors From Two Models, Women and Men Regressed Separately

                                     Model 1: TIB
                              Odds        Log         Partial
Variable                    (Antilog)     Odds        Correlation

Women

 Prior experience(b)          1.85        0.62        0.03
 Intentions(c)                2.94        1.08(*)     0.11
 Spring break experience      5.57        1.72(**)    0.25
 Role model(d)
 Pact(e)
 Constant                                -7.30
Men
 Prior experience(b)          0.74       -0.30        0.00
 Intentions(c)                3.68        1.30(**)    0.14
 Spring break experience      6.14        1.82(***)   0.24
 Role model(d)
 Pact(e)
 Constant                                -8.04

                                     Model 2: Enhanced TIB

Variable                      Odds          Log           Partial
                            (Antilog)       Odds        Correlation
Women

 Prior experience(b)          1.40         0.34           0.00
 Intentions(c)                2.41         0.88           0.00
 Spring break experience      2.81         1.03           0.09
 Role model(d)                3.59         1.28(***)      0.28
 Pact(e)                      11.92        2.48(**)       0.16
 Constant                                 -5.79
Men
 Prior experience(b)          0.81        -0.23           0.00
 Intentions(c)                4.83         1.58(**)       0.15
 Spring break experience     12.27         2.51(***)      0.28
 Role model(d)                1.22         0.20           0.00
 Pact(e)                      0.51        -0.67           0.00
 Constant                                -10.49

Notes. N = 306 women; N = 251 men. All coefficients are maximum-likelihood estimates. Models 1 and 2 both correctly classified 80% of the cases. For women, 1 over baseline, Chi-square improvement was 38.236 (df = 3; p < .01); and for 2 over 1, Chi-square improvement was 35.747 (df = 2; p < .01). For men, 1 over baseline, Chi-square improvement was 33.300 (df = 3; p < .01), and for 2 over 1, Chi-square improvement was 5.441 (df = 2; p = .06).

(a) 0 = no coital activity; 1 = coitus with a partner met on spring break,

(b) 0 = no prior casual sex experience; 1 = prior casual sex experience,

(c) 0 = no intentions to engage in casual sex on spring break; 1 = intend to engage in casual sex on spring break,

(d) Measured as proportion of friends who engaged in coitus on spring break,

(e) 1 = pact to engage in casual sex; 0 = no pact; -1 = pact not to engage in casual sex.

(*) p < .05.

(**) p < .01.

(***) p < .001.

Results of the second regression were different for men and women. For men, the addition of two measures of peer influence did not improve the predictive power of the model, as seen in the absence of a significant increase in the improvement in chi-square or in the percentage of cases that were correctly classified. Intentions and prior casual sex experience remained the significant predictors of engaging in casual sex. For women, however, there was a significant improvement in the predictive power of the model. The two peer influences, perceived role modeling and pacts, were the significant factors in classifying students correctly. Although the frequency of participation in spring break experiences conducive to forming new coital partnerships had the strongest effect in Model 1, it was just below statistical significance (p = .06) in Model 2. In Model 2, the proportion of friends engaging in sex had the strongest effect on whether women engaged in casual sex.

DISCUSSION

In their review of research on premarital sexual decisionmaking Christopher and Roosa (1991) noted that the contexts and settings in which coital partnerships are formed have received little research attention. The research reported in this paper focused on the North American spring break vacation and its influence on a young adult to engage in casual sex. It built on the review of characteristics of vacations that contribute to engaging in sex by Herold and van Kerkwijk (1992), and on the research on "vacation sex" conducted by Eiser and Ford (1995) and Smeaton and Josiam (1996). In this study, 21% of men and 17% of women reported engaging in coitus with a new partner, 15% and 13%, respectively, with someone they met on spring break.

Three factors stand out about these percentages. First, there is greater similarity in the percentages of men and women who engaged in casual sex than would be expected considering the degree of dissimilarity between men and women on all other variables examined in this research. On variables used to predict either the formation of intentions or engaging in casual sex, men's scores clustered in the direction of acceptance, endorsement, and expectation of situations conducive to casual sex; women's clustered in the opposite direction. This finding is consistent with previous research (for reviews, see Christopher & Roosa, 1991; Oliver & Hyde, 1993). Despite this clustering, casual sex did not occur for most men. This suggests that women's attitudes, norms, expectations, and intentions, which were less supportive of engaging in casual sex, determined whether coitus occurred, consistent with the traditional image of women as "gatekeepers" (Herold, 1984).

The second factor of interest is that the percentages reporting coital activity with a new partner were similar to the British beach resort vacationers in Eiser and Ford's (1995) study (22% for men and 20% for women). However, the percentages for women in our study and in the British study were considerably higher than that reported by Smeaton and Josiam (1996) in their Panama City Beach spring break study where only 4% of women reported sex with a new partner. In our study and the British study, data were collected by male and female research assistants who were close in age to the vacationers, whereas Smeaton and Josiam's data were collected by two male professors. Our procedure may have produced greater comfort and honesty among the female participants, resulting in greater consistency in women's and men's reports.

The third factor to note is that the percentage of men and women reporting casual sex on spring break was considerably lower than the percentage of students reporting prior casual sex. Thus, although spring break is a vacation conducive to new coital partnerships for some, most students on spring break did not participate in coitus.

Since Reiss' (1967) work on personal standards for sexual behavior, consideration of the relationship between standards and behavior has had a prominent place in sexuality research. In this research it was personal attitudes rather than personal standards (as differentiated by Triandis) that influenced casual sex, with attitudes influencing intentions to engage in casual sex. Personal standards and role beliefs--that is, the variables that measured social norms as expressed in personal moral codes or standards--did not have a significant influence on intentions or expectations about casual sex. These results contrast with those of other studies using the Triandis model in which personal standards or role beliefs were found to have an effect that was stronger than that of attitudes (Boyd & Wandersman, 1991; Godin et al., 1996). This highlights a characteristic of the spring break environment and the behavior under consideration that was more fully described by students in the elicitation phase: Activities on spring break, including coitus, were described as exceptions to the everyday experience--as outside of usual expectations, standards, or norms. Students used phrases such as "what happens in Daytona, stays in Daytona," "nothing that happens there comes home," and "nothing counts." They portrayed an atmosphere in which the usual rules and moral codes did not apply. Students provided detailed descriptions of how some had behaved "totally out of character" or in ways that "they never would at home." These illustrations and the results of the statistical analysis support the picture of spring break as an environment in which personal codes are temporarily suspended.

Although personal codes did not influence spring break coital behaviors, perceptions of peer expectations and promises or pacts made with peers did. This also conformed to the image of spring break described in the elicitation phase of the research: Friends travel and "hang" together in Daytona Beach. In statistical analyses, peers had a significant influence on both the formation of intentions to engage in casual sex and in casual sex activity when in Daytona Beach. However, peer group influence took different forms for men and women. For men, peer influence operated through pacts formed with friends about casual sex and perceptions of friends' approval or disapproval of casual sex. In both cases this influence was on the formation of intentions to engage in casual sex. Peer influences did not have a direct effect on engaging in casual sex. For women, peer influence operated through the formation of pacts between friends about casual sex and through role modeling (i.e., the percentage of friends on spring break who participated in coitus). Pacts influenced intentions and, along with the percentage of friends who engaged in coitus on spring break, had a direct effect on engaging in casual sex in the Daytona sample. Thus, men's intentions were more strongly influenced by the effect of peers than were women's. However, peers had an additional direct influence on women's coital activity on spring break. A common finding in other studies is that peers have a stronger influence on men than on women (for a review, see Christopher and Roosa, 1991), though some have found that the influence is experienced equally by men and women (e.g. Winslow et al., 1992). However, in most research only one form of peer influence is tested. The effect of different forms of peer influence found in this study leads us to concur with Wilcox and Udry's (1986) suggestion that it is important to examine more fully the different types of peer influence in future work.

The activities and experiences that students considered conducive to casual sex paralleled those identified by Herold and van Kerkwijk (1992): a perpetual party atmosphere, high rates of alcohol consumption, sexual contests, dancing dirty, frequent attempts made to "pick up" a sexual partner, and being in a break-loose mood. Along with positive attitudes toward casual sex while on spring break and perceived peer support for casual sex, anticipation of this spring break experience led to the formation of intentions or expectations of engaging in casual sex. Once in Daytona Beach, participation in casual sex depended not only on prior intentions but also on the degree of participation in spring break activities. For women, however, peer influences in Daytona Beach were more important in determining casual sex activity than were intentions or participation in spring break activities. For men, the greater consistency in the importance of the spring break environment on both expectations of casual sex and casual sex activity suggests that they are more susceptible to influences from the external environment and potentially are more likely to interpret the environment as providing sexual cues.

There are limitations to the generalizability of the results of this research. First, because of the voluntary nature of the samples and the restriction to one research site it is difficult to know whether the results can be generalized to other vacations. The high response rate for recruitment at Daytona Beach, the consistency in results across the different sampling designs in the Daytona study, the similarity in the reports of prior sexual activity with reports obtained in Canadian studies using national probability samples (Maticka-Tyndale, 1997), and the similarities of our results to those of other studies of vacation sex in other locations (Eiser & Ford, 1995; Herold & van Kerkwijk, 1992; Smeaton & Josiam, 1996), support confidence in the findings. However, the generalizability of findings of a study of this nature must always be treated with caution until more research, ideally with improved sampling strategies, has been conducted.

A second limitation is the absence of a longitudinal design to test the full Triandis model. Initially, one of our goals was to obtain such a sample. However, because of the last-minute nature of spring break holidays and the intense academic schedule of midterm exams during the weeks just before spring break it was difficult to recruit a sufficient number of participants before spring break. Although the lack of a significant difference between the responses of students recruited in Daytona and those recruited before spring break gave us confidence that results in both studies were reflective of the same population, a lull test of the Triandis model awaits a longitudinal sample.

Some researchers may feel that the absence of standardized measures of constructs introduced an additional limitation to the generalizability of our results. Triandis's theory stresses the importance of measurements specific to the time, place, event, and population, and provides a methodology for creating and testing such measures. We followed these recommendations for creating measures and testing their reliability and validity. However, Triandis's rejection of standardized scales for their lack of specificity to the topic of study may be viewed with skepticism by some researchers. Whether one sees the types of measures used in this study as enhancing or limiting the quality of the results depends on whether the measures are considered a strength or a weakness of the design.

These limitations point to two methodological directions for future work. The first is the exploration of more representative sampling designs either for studies of contexts such as vacations or for sampling diverse contexts and comparing them for their influence on different forms of sexual partnering. Second, an empirical comparison of Triandis's measurement model to the use of standardized scales could help resolve some of the questions raised about potential limitations of Triandis's approach.

The areas of gender difference and similarity also suggest areas for future inquiry. The gender differences in all but coital activity on spring break remind us that coitus occurs between partners. A full understanding of how sexual activity and partnerships occur requires future research at the level of the couple. In addition, further inquiry is necessary to understand the variations in peer influence on men and women.

The sexual activity on spring break was described by students as a temporary, rapidly progressing partnership. This raises questions about how different forms of sexual partnerships are formed, what meaning they hold, and how they fit within the overall sexual scripts of young adults. Further investigation of environments conducive to rapid and casual sexual encounters, the specific characteristics of these environments that encourage sexual partnering, and the differences between those who are influenced and those who resist the influence of the environment is another area of inquiry suggested by this research.

Finally, the TIB provided a powerful multivariate model that integrated personal, social, and situational factors in explaining sexual planning and activity. This suggests that this model could be helpful in more fully explaining sexual partnering in other contexts.

REFERENCES

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

Ajzen, I. (1985). From intentions to actions: A theory of planned behavior. In J. Kuhl and J. Beckmann (Eds.), Action-control: From cognition to behavior, (pp. 11-39). Heidelberg, Germany: Springer.

Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50, 179-211.

Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, NJ: Prentice-Hall.

Boyd, B., & Wandersman, A. (1991). Predicting undergraduate condom use with the Fishbein and Ajzen and the Triandis attitude-behavior models: Implications for public health interventions. Journal of Applied Social Psychology, 21, 1810-1830.

Christopher, F. S., & Cate, R. M. (1984). Factors involved in premarital sexual decision-making. The Journal of Sex Research, 25, 255-266.

Christopher, F. S., & Roosa, M. W. (1991). Factors affecting sexual decisions in the premarital relationships of adolescents and young adults. In K. McKinney & S. Sprecher (Eds.), Sexuality in close relationships. (pp. 111-134). Hillsdale, NJ: Lawrence Erlbaum Associates.

D'Augelli, J. F., & D'Augelli, A. R. (1977). Moral reasoning and premarital sexual behavior: Toward reasoning about relationships. Journal of Social Issues. 33, 46-66.

Eiser, J. R., & Ford, N. (1995). Sexual relationships on holiday: A case of situational disinhibition? Journal of Social and Personal Relationships, 12, 323-339.

Fishbein, M., & Jaccard, J. (1973). Theoretical and methodological considerations in the prediction of family planning intentions and behavior. Representative Research in Social Psychology, 4, 37-51.

Godin, G., Maticka-Tyndale, E., Adrien, A., Manson-Singer, S., Willms, D., & Cappon, P. (1996). Cross-cultural testing of three social cognitive theories: An application to condom use. Journal of Applied Social Psychology, 26, 1556-1586.

Herold, E. S. (1984). Sexual behavior of Canadian young people. Toronto: Fitzhenry & Whiteside.

Herold, E. S., & van Kerkwijk, C. (1992). AIDS and sex tourism. AIDS and Society 4, 1-8.

Joreskog, K. G., & Sorbom, D. (1986). LISREL: Analysis of linear structural relationships by the method of maximum likelihood, Version VI. Mooresville, IN: Scientific Software, Inc.

Jorgensen, S. R., & Sonstegard, J. (1984). Predicting adolescent sexual and contraceptive behavior: An application and test of the Fishbein model. Journal of Marriage and the Family, 46, 43-55.

Josiam, B., Clements, J. S., & Hobson, P. (1995). Spring break student travel: A longitudinal study. Unpublished manuscript.

Maticka-Tyndale, E. (1997). Reducing the incidence of sexually transmitted disease through behavioral and social change. The Canadian Journal of Human Sexuality, 6, 89-104.

Mewhinney, D. M., Herold, E. S., & Maticka-Tyndale, E. (1995). Sexual scripts and risk-taking of Canadian university students on spring break in Daytona Beach, Florida. The Canadian Journal of Human Sexuality, 4, 273-288.

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

Randall, D. M., & Wolff, J. A. (1994). The time interval in the intention-behavior relationship: Meta-analysis. British Journal of Social Psychology, 33, 405-418.

Reiss, I. L. (1967). The social context of premarital sexual permissiveness. New York: Holt, Rinehart, & Winston.

Rook, K. S., & Hammen, C. L. (1977). A cognitive perspective on the experience of sexual arousal. Journal of Social Issues, 33, 7-29.

Smeaton, G., & Josiam, B. (1996). Sex, drugs and alcohol on the beach: "Where the boys are in the age of AIDS." Paper presented at the meeting of the Society for the Scientific Study of Sex, Houston, TX.

Smith, K. W., & Sasaki, M. S. (1979). Decreasing multicollinearity: A method for models with multiplicative functions. Sociological Methods and Research, 8, 35-56.

Triandis, H. C. (1977). Interpersonal behavior. Monterey, CA: Brooks/Cole.

Triandis, H. C. (1980). Values, attitudes, and interpersonal behavior. In H. Howe & M. Page (Eds.), Nebraska Symposium on Motivation 1979 (pp. 195-295). Lincoln, NE: University of Nebraska Press.

Triandis, H. C. (1994). Culture and social behavior New York: McGraw-Hill Inc.

White, K. M., Terry, D. J., & Hogg, M. A. (1994). Safer sex behavior: The role of attitudes, norms, and control factors. Journal of Applied Social Psychology 24, 2164-2192.

Wilcox, S. & Dury, W. (1986). Autism and accuracy in adolescent perceptions of friends' sexual attitudes and behaviors. Journal of Applied Social Psychology, 16, 361-374.

Winslow, R. W., Franzini, k. R., & Hwang, J. (1992). Perceived peer norms, casual sex and AIDS risk prevention. Journal of Applied Social Psychology, 22, 1809-1827.

Manuscript accepted October 16, 1997

This research was funded by Health Canada through the National Health Research and Development Programme, Grant #6606-5523-AIDS.

Address correspondence to Dr. E. Maticka-Tyndale, Department of Sociology and Anthropology, University of Windsor, 401 Sunset Avenue. Windsor, Ontario N9B 3P4; e-mail: [e-mail]