Quick viewing(Text Mode)

Sexual Infidelity Among Married and Cohabiting Americans

Sexual Infidelity Among Married and Cohabiting Americans

JUDITH TREAS University of CaliforniaÐIrvine

DEIRDRE GIESEN Utrecht University*

ࡗ

Sexual In®delity Among Married and Cohabiting Americans

Virtually all American couples, married or cohab- partner (Smith, 1994). About half the states in the iting, expect sexual exclusivity of one another. U.S. retain laws against adultery that, although This article asks why some people are sexually they are rarely enforced, would deny married per- exclusive while others have sex with someone be- sons who have the right to vote, sides their mate. Previous research has linked serve alcohol, practice law, adopt children, or personal values, sexual opportunities, and quality raise their own children (Constitutional barriers, of the marital relationship to extramartial sex. 1992; Siegel, 1992). American couples, whether This paper integrates these ®ndings in a multi- married or cohabiting, agree that it is important to variate model that incorporates factors informing be monogamous (Blumstein & Schwartz, 1983; sexual decision making as well as demographic Greeley, 1991). ``risk factors.'' Nationally representative survey Couples' agreements about sexual exclusivity data show higher likelihood of sexual in®delity are a contractual condition of their unions. As among those with stronger sexual interests, more with all contracts, bargains are sometimes broken. permissive sexual values, lower subjective satis- Although sexual ®delity is the dominant practice, faction with their union, weaker network ties to recent surveys show that between 1.5 and 3.6% partner, and greater sexual opportunities. With of married persons had a secondary sex partner in these factors controlled, gender differences are the past year (Smith, 1991; Choi, Catania, & Dol- substantially reduced or eliminated, although ra- cini, 1994; Leigh, Temple, & Trocki, 1993). This cial effects persist. paper asks why some people are sexually exclu- sive while others have sex with someone besides their mate. Americans disapprove of sexual in®delity. More than 90% of the general public say it is ``always'' PREVIOUS RESEARCH or ``almost always'' wrong for a married person to have sex with someone besides the Research on sexual in®delity has focused on three domainsÐthe personal values of the individual, Department of Sociology, Social Science Plaza A3151, the opportunities for extramarital sex, and the cou- University of California, Irvine, CA 92697-5100. ple's relationship. Permissive sexual values are associated with *Department of Sociology/ICS, Utrecht University, Hei- extramarital sex. Among Americans who believe delberglaan 1, 3584 CS Utrecht, The . extramarital relations are ``not at all wrong,'' 76% Key Words: , extramarital sex, marriage, sex- report having had extramarital sex compared to ual behavior. only 10% of those who think extramarital sex is

48 Journal of Marriage and the 62 (February 2000): 48±60 Sexual In®delity 49

``always wrong'' (Smith, 1994). Being male, Af- Compared to married couples, cohabitors are not rican-American, and well educated are all asso- as sexually exclusive (Forste & Tanfer, 1996)Ð ciated with permissive sexual values (Smith, consistent with their less conventional values 1994). So is living in a big city. Extramarital per- (Clarkberg, Stolzenberg, & Waite, 1995), with the missiveness is linked to liberal political and relig- lower levels of commitment in cohabiting unions ious ideologies (Smith, 1994). It is also related to (Bumpass, Sweet, & Cherlin, 1991), and with dif- gender egalitarianism and premarital permissive- ferences in the sorts of partners chosen for cohab- ness (Reiss, Anderson, & Sponaugle, 1980). itation as opposed to marriage (Forste & Tanfer, Opportunities, namely potential partners and 1996). circumstances assuring secrecy, facilitate extra- Findings have accumulated in a piecemeal marital sex. Some Americans admit they would fashion. Since no study has integrated value pref- have extramarital sex if their mate would not ®nd erences, sexual opportunities, relationship con- out (Greeley, 1991). Couples who lead separate straints, and demographic risk factors into a single lives, for example, have more opportunities and multivariate model, we cannot discount the pos- are more likely to have secondary sex partners sibility of spurious associations between these fac- (Blumstein & Schwartz, 1983). Married people tors and in®delity. who perceive alternative partners to be available are more likely to have had extramarital sex CONCEPTUAL FRAMEWORK (Johnson, 1970; Maykovich, 1976). Of course, those predisposed to extramarital sex might be Everyday accounts of extramarital sex often stress more likely to recognize opportunities that arise. irrational causes like alcohol-impaired judgment Dissatisfaction with the marital relationship it- or sexual addiction (Giddens, 1992). Although self is associated with extramarital sex (Brown, cultural scripts focus on and , 1991; Vaughn, 1986). Those who engage in adul- people contemplating in®delity describe consid- tery are less likely to report happy ered decisions. The self-conscious evaluation of (Greeley, 1991; Bell, Turner, & Rosen, 1975). In- extramarital options has been called ``thinking'' ®delity has been linked to men's sexual dissatis- (Atwater, 1982) or ``the debate'' (Lawson, 1988). faction (Maykovich, 1976) and to women's per- A reports making ``a quick sort of negative ception of inequity in the marriage (Prins, Buunk, and positive checklist'' (Lawson, pp. 134±136). A & VanYperen, 1983). Causal direction is unclear, con®des, ``(I)t's a question you have to however, and other studies fail to ®nd a signi®cant ask yourself before.... `Why am I doing this? association for marital happiness (Maykovich, What will I get out of it? How does this affect the 1976), marital adjustment (Johnson, 1970), seeing status quo?''' (Lawson, p. 147). a mate as less affectionate (Edwards & Booth, Given social norms and strong dyadic expec- 1976), or, for Whites, quality of marital sex (Choi tations for sexual exclusivity, sexual in®delity de- et al., 1994). National surveys identify demo- mands calculated behavior. Theorizing about sex graphic risk factors for multiple sex partners. Ed- in terms of anticipated costs and gains yields use- ucation is positively related not only to permissive ful insights, as Reiss and Miller (1979) suggested sexual values, but also to sexual in®delity (Smith, when hypothesizing a ``reward-cost balance'' for 1991; Leigh et al., 1993). Being African-Ameri- premarital permissiveness. A decision-making can is associated with greater likelihood of mul- framework also serves to integrate piecemeal re- tiple sexual relationships than being White (Smith, sults of prior studies on extramarital sex. 1991; Dolcini et al., 1993). Men engage in more extramarital sex than women (Choi et al., 1994; Tastes and Values Smith, 1991), perhaps because of male-female dif- ferences in reproductive strategies (Lancaster, A review of clinical and research studies identi®es 1994), the gendered nature of learned sexual 31 reasons for extramarital relations; Most, falling scripts (Gagnon & Simon, 1973), or a double stan- under the categories of sex, , dard that judges men's sexual permissiveness less , and ego bolstering, pertain to personal grat- harshly than women's. The number of sex partners i®cation (Glass & Wright, 1992). Some people's declines with age (Dolcini et al., 1993; Smith, tastes and values increase the likelihood that they 1991), which might re¯ect biological effects of will engage in extramarital sex. People highly in- aging (Edwards & Booth, 1994) or recent cohorts' terested in sex might eschew sexual exclusivity more permissive sexual values (Smith, 1994). because they anticipate greater pleasure from ex- 50 Journal of Marriage and the Family tramarital relations. On the other hand, nonper- Social networks composed of people who are missive values are known to be negatively asso- apt to disapprove of adultery discourage extra- ciated with sexual in®delity, perhaps because marital relations, if only because one must go to people who hold these values anticipate discom- greater lengths to keep sexual in®delity secret. In- fort reconciling dissonant beliefs and behavior terestingly, married couples who became nonmon- (Lawson, 1988). ogamous ``swingers'' were insulated from social networks monitoring behavior and imposing costs Hypothesis 1a: Greater interest in sex is as- on nonconformists: Swingers knew fewer neigh- sociated with a greater likelihood of in®- bors, visited relatives less often, and joined fewer delity. religious groups (Gilmartin, 1974).

Hypothesis 1b: Nonpermissive sexual val- H2d. When partners enjoy one another's ues are associated with a lower likelihood and networks, the likeli- of in®delity. hood of in®delity is lower.

Opportunities H2e. Controlling for sexual values, attend- ing religious services more frequently is as- People with fewer opportunities for undetected sociated with lower likelihood of in®delity. sex must go to greater lengths to have extramarital sex. Individual endowments and learned skills af- fect how many sexual opportunities come one's Primary Relationship way. People with more sexual relationships in the Because partners expect ®delity, potential costs to past are more likely to have a secondary sex part- the primary relationship loom large in the face of ner (Bozon, 1996). The sexually experienced in®delity. A mate who learns of a partner's in®- might be more attractive; or they might have a delity might respond with emotionally-draining ``learned advantage'' if they are more ef®cient , tit-for-tat in®delities, physical than novices at recognizing sexual opportunities, , the withholding of couple services (e.g., recruiting sex partners, and managing sexual en- sex, companionship, monetary support), and even counters. (Pittman, 1989). H2a. Having had more sexual partners pre- Marital quality mediates costs. If a marriage is viously is associated with a greater likeli- judged to be unrewarding, one has less to lose hood of in®delity. from extramarital sex. One can afford to be indif- ferent, both to costs to the marital relation and to Social context also determines opportunities. sanctions a mate might offer. An extreme example As a place to socialize outside the company of a is the ``out-the-door'' where one partner mate, the workplace offers access to potential pursues an extramarital relationship to force a partners (Lawson, 1988). Some work presents mate to end an unhappy marriage (Brown, 1991). greater opportunities than other work. For in- Like subjective marital dissatisfaction, mates' so- stance, people whose jobs require overnight travel cial dissimilarity or heterogamy might prompt in- are more likely to have multiple sex partners ®delity because social differences imply lower (Wellings, Field, Johnson, & Wadsworth, 1994). marital returns as a result of fewer stabilizing Compared to small towns, big cities offer more commonalities in the relationship (Lehrer & Chis- opportunities for undetected sexÐmore potential wick, 1993). partners, greater anonymity, and more permissive sexual values (Smith, 1994). In fact, big city res- H3a. Greater dissatisfaction with the union idents do average more sex partners (Smith, is associated with greater likelihood of in- 1991). ®delity.

H2b. A job requiring personal with H3b. Greater disparity in partners' social potential sex partners is associated with characteristics is associated with greater greater likelihood of in®delity. likelihood of in®delity.

H2c. Big city residence is associated with People get locked into a union, however un- greater likelihood of in®delity. ful®lling, by investments that they cannot recoup Sexual In®delity 51 outside the relationship. Married people have were interviewed by NORC about sexual attitudes more invested in their unions than do cohabitors. and behavior (Laumann, Gagnon, Michael, & Mi- Besides a public commitment, the married are chaels, 1994, pp. 42±73). Respondents included more likely to have children and to own a home those who were legally married and those who jointly. They face higher exit costs should the re- were cohabiting. In a face-to-face survey, inter- lationship end. Because cohabitors risk less by an viewers asked about social background, health, affair, it is not surprising that cohabitors are more fertility, sexual activities, attitudes, and fantasies. likely to have secondary sex partners (Dolcini, et After answering demographic questions at the al., 1993). start of the interview, respondents ®lled out a short, self-administered questionnaire inquiring, H3c. Cohabiting is associated with a greater among other things, whether they had ever had likelihood of in®delity. extramarital sex and whether they had had sex with someone besides their regular partner in the The likelihood of ever having been unfaithful last 12 months. Interviewers then collected de- increases with the duration of the union due to tailed marital, cohabitation, and sexual histories. longer exposure to the risk of in®delity. At any This analysis focuses on 2,870 respondents for given time, however, the likelihood of in®delity whom sexual in®delity had been possible because might vary with union duration. There are two they had married or lived in a sexual relationship competing arguments. If couples who have been at one time or another. We included African- together longer have made more stabilizing in- American and Hispanic oversamples; but because vestments in their relationship, what they stand to of small sample size,we excluded 45 same-sex co- lose will discourage in®delity. Yet, declines in co- habitors and 4 other cases for whom partner's gen- ital frequency (Wellings et al., 1994) suggest that der was unknown. We also eliminated 223 re- some marital bene®ts wane with time. If bene®ts spondents for whom we could not ascertain length jeopardized by in®delity decline over time, the of exposure to the risk of in®delity. This left 2,598 likelihood of in®delity will increase at longer usable cases. union durations. Data quality is a concern with sensitive matters like extramarital sex. NHSLS self-reports of ex- H3d. (investment hypothesis) Longer union tramarital sex are consistent with those from the duration is associated with lower likelihood General Social Survey (Laumann et al., 1994), but of in®delity at a given time. Or: we examined how results were affected by differ- ent operational de®nitions of sexual in®delity. H3e. (habituation hypothesis) Longer du- Con®dence in our ®ndings increased when we ration is associated with greater likelihood could reconcile results regardless of: (a) the meth- of in®delity at a given time. od of eliciting the report of in®delity; (b) a focus on whether in®delity had ever occurred over the Integrating prior ®ndings on sexual in®delity, course of a union (cumulative incidence) or dur- a decision-making framework generates hypothe- ing a given time period (prevalence); and (c) the ses to be tested with superior survey data now respondents considered (in marriages or in all het- available. We estimate a multivariate model of erosexual unions). We constructed three measures sexual in®delity incorporating personal tastes and of in®delity. Given a dichotomous dependent vari- values, the sexual opportunity structure, and fea- able, we used logistic regression to estimate the tures of the primary (i.e., marital or cohabiting) multivariate models. relationship. We control for demographic ``risk factors'' that might confound the associations Dependent Variables among variables and consider whether factors in- forming sexual decision making can account for Self-recorded cumulative incidence. On a self-ad- the effects of gender, race, age, and education. ministered questionnaire to be sealed in a ``pri- vacy'' envelope, respondents marked whether they had ever had sex with someone other than METHOD their husband or wife while they were married. The 1992 National Health and Social Life Survey This self-recorded item was less vulnerable to so- (NHSLS) is a national probability sample of 3,432 cial desirability bias than a person-to-person in- English-speaking Americans ages 18±59 who terview. To link sexual behavior and mate's char- 52 Journal of Marriage and the Family acteristics, we limited our analysis of this item to by an ordinal item on the wrongness of extramar- respondents who had been married only once, be- ital sex. cause those with multiple relationships did not in- dicate which one involved adultery. Extramarital Opportunities. Although it was not confounded sex was reported by 266 (15.5%) of the 1,717 with in®delity, early sexual experienceÐthat is, respondents in this category. number of partners between age 18 and the ®rst unionÐmeasured endowments and interpersonal Interview cumulative incidence. Two interview skills that attract partners. Work-place opportuni- parts determined whether ever-married and ever- ties were measured by a summated scale of 3 cohabited respondents had ever been unfaithful. items on whether the respondent's job required First, sexual histories showed the timing of sexual touching, talking, or being alone with others. A relationships in the 12 months before the inter- dummy variable measured residence in a large- or view. Sex was de®ned as ``mutually voluntary ac- medium-sized central city. Fidelity-supporting so- tivity with another person that involves genital cial networks were gauged by annual frequency contact and sexual excitement or arousal, that is, of religious attendance and by shared networks, a feeling really turned on, even if intercourse or or- summated scale based on 4 items regarding how gasm did not occur.'' Second, for earlier periods, much partners enjoyed spending time with one an- interviewers asked whether, while living in a giv- other's family and friends. en marriage or cohabitation, the respondent con- tinued a former sexual relation or began one with Primary relationship. Three dummy variables a new partner. The data excluded sexual relation- measured partners' social dissimilarity: (a) Relig- ships that occurred after a separation but before a ious difference indicated that partners did not divorce. Because recall is better and respondents' share the same major religion; (b) education dif- current characteristics more proximate in time, we ference referred to a major gap in educational at- focused on the current or most recent marriage or tainment; and (c) age difference re¯ected an age cohabitation. We treated as one union any cohab- gap of at least 5 years. Available only for unions itation that became formalized in a marriage. Sex- established in the previous 12 months, (d) dissat- ual in®delity was reported by 291 (11.2%) of isfaction was the sum of values for two items on 2,598 ever-married or ever-cohabited respondents. the union's emotional satisfaction and physical Interview 12-month prevalence. Interviewer-col- pleasure. With cumulative incidence, we consid- lected data on the timing of sexual relationships ered whether the respondent had ever cohabited showed 94 (4.7%) of the 2,010 respondents co- in the union. With prevalence, we asked whether habiting and/or married in the past year had been the respondent had been cohabiting during the unfaithful to their primary partner during this past 12 months. With cumulative incidence, union time. A narrower window on sexual activity re- duration controlled for exposure time for the risk sults in fewer instances of in®delity but a closer of in®delity. For 12-month prevalence, union du- temporal match between sexual events and the re- ration tested the investment versus habituation hy- spondents' characteristics. potheses.

Demographic risk factors. Dummy variables in- Independent Variables dicated whether the respondent was male or fe- As noted, sexual in®delity was associated with male, White or African-American. A categorical three sets of variables that might affect decision variable measured respondents' education. Age making: (a) sexual tastes and values that deter- (in years) at in®delity was determined only for the mined personal costs and gains to in®delity; (b) 12-month prevalence measure. Because duration opportunities for undetected sex; and (c) the pri- and age are highly correlated, they were not in- mary relationship jeopardized by in®delity. Inde- cluded in the same model. The trained interview- pendent variables and their operationalizations and er's rating of respondents' frankness on a 4-point weighted descriptive statistics appear in Table 1. ordinal scale proved a consistently signi®cant con- trol for underreporting of in®delities. Given the Tastes and values. Sexual interest was measured possible limitations of subjective impressions, we by an item on how often the respondent thought estimated all models with and without this vari- about sex. Nonpermissive values were measured able; in both cases, the models' ␹2 values were Sexual In®delity 53

TABLE 1. INDEPENDENT VARIABLES,MEANS, AND STANDARD DEVIATIONS:EVER-MARRIED AND EVER-COHABITED AMERICANS,AGES 18±59, 1992

Sexual interest ``On average, how often do you think about sex?'' (never ϭ 0; several times a day ϭ 5) Mean ϭ 3.02, SD ϭ 1.12 Nonpermissive ``What is your opinion about a married person having sexual relations with someone other than the marriage partner?`` (always wrong ϭ 4; not wrong at all ϭ 1) Mean ϭ 3.66, SD ϭ.68 Sexual experience Number of sexual partners between age 18 and start of ®rst marriage or cohabitation Mean ϭ 4.07, SD ϭ 14.46 Workplace opportunity Four-point summated scale based on three items: 1) frequently alone with clients, custom- ers, or co-workers; 2) job requires touching clients, customers, or coworkers; 3) requires discussing the personal concerns of clients, customers, or coworkers. No job or no oppor- tunities ϭ 0. Mean ϭ .97, SD ϭ .96, alpha ϭ .55 Central city Resident of a central city of 50,000 or more (resident ϭ 1, else ϭ 0) Mean ϭ .28, SD ϭ .45 Shared networks Five-point count based on highly positive responses to four items on how much respondent enjoys partner's family, respondent enjoys partner's friends, partner enjoys respondent's family, partner enjoys respondent's friends. Mean ϭ 1.62, SD ϭ 1.46, alpha ϭ.74 Religious attendance How often respondent attends religious services: nine categories recoded to number of times annually (never ϭ 0; several times a week ϭ 104) Mean ϭ 23.88, SD ϭ 31.90 Religious difference Dummy variable for respondent whose partner does not share the same major religion (dif- ferent ϭ 1, same ϭ 0) Mean ϭ .32, SD ϭ .47 Education difference Dummy variable (different ϭ 1, same ϭ 0) if partner has two or more category difference based on ®ve category education variable (less than high school ϭ 1; more than college graduate ϭ 5) Mean ϭ .16, SD ϭ .36 Age difference Dummy variable (different ϭ 1, same ϭ 0) if respondent's and partner's ages differ by 5 or more years Mean ϭ .16, SD ϭ .37 Couple cohabited Couple ever lived together without being married (yes ϭ 1, no ϭ 0) Mean ϭ .49, SD ϭ .50 Couple cohabiting Couple cohabiting during last 12 months (yes ϭ 1, no ϭ 0) Mean ϭ .18, SD ϭ .38 Duration Years married or cohabiting in relationship Mean ϭ 11.29, SD ϭ 10.69 Dissatisfaction Nine-point summated scale from two items on emotional satisfaction and physical pleasure with primary relationship (extremely satis®ed or pleasurable ϭ 1; not at all ϭ 5) Mean ϭ 3.35, SD ϭ 1.51, alpha ϭ .84 Sex Male ϭ 1, female ϭ 0 Mean ϭ .42, SD ϭ .49 African American African American ϭ 1, else ϭ 0 Mean ϭ .15, SD ϭ .36 Education Respondent's educational attainment (grade 8 or less ϭ 1; postsecondary trade school ϭ 4; advanced college degree ϭ 8) Mean ϭ 4.16, SD ϭ 1.64 Age Respondent's age in years recoded into four categories (18±30, 31±40, 41±50, 51ϩ) Mean ϭ 2.24, SD ϭ 1.03 Frankness Interviewer's assessment of respondent's frankness (probably not frank ϭ 1; entirely frank ϭ 4) Mean ϭ 3.66, SD ϭ .58 Note: n ϭ 2,598; n ϭ 2,010 for couple cohabiting, dissatisfaction, and age (12-month variables).

highly signi®cant (p Ͻ .0000), and the coef®- dependent variables, preliminary analyses found that cients were very similar. men were less likely to be sexually exclusive, not We posit no interactions between demographic because they weight factors differently than do and decision-making variables. Although one might women, but largely because men and women differ expect, say, gender differences in the effects of in- in their values (e.g., means) on these factors. 54 Journal of Marriage and the Family

TABLE 2. SEXUAL EXCLUSIVITY EXPECTATIONS AND who cohabited together (89%) (t ϭ 1.56, ns; p ϭ BEHAVIOR BY :MARRIED AND .12) or current cohabitors (88%) (t ϭ 1.52, ns; p COHABITING AMERICANS,AGES 18±59, 1992 ϭ .13). Nor were current cohabitors and previ- Married ously cohabiting marrieds statistically different Cohabiting Without Cohabited Hetero- from one another (t ϭ .20, ns; p ϭ .84). Of Variable Cohabiting First sexual course, cumulative incidence was affected by union duration, and cohabiting persons were apt Respondent expects ex- clusivity 99% 98% 94% to have had less time ``at risk.'' Partner expects exclu- sivity 99% 99% 95% Respondent exclusive 92% 89% 88% Multivariate Models of In®delity n 939 532 331 Cumulative incidence of extramarital sex. If con- Note: From National Health and Social Life Survey, cerns about social acceptability deter people from 1992. admitting sexual in®delity, the self-administered questionnaire offered better data than the person- to-person interview. The ®rst two columns of Ta- RESULTS ble 3 show logistic regression results for the self- recorded cumulative incidence of extramarital sex. Expectations for Sexually Exclusive Unions Tastes and values demonstrated the hypothesized NHSLS respondents, whether married or cohab- relationships: Greater interest in sex was positive- iting, held similarly high expectations for sexual ly associated with the likelihood of in®delity, exclusivity. This was an important fact to estab- while nonpermissive sexual values were negative- lish, because we compared married persons with ly associated. The exponentiated betas show the married and cohabiting heterosexuals to gauge the magnitude of effects. Controlling for other vari- robustness of results to different operationaliza- ables, thinking about sex daily instead of just a tions of sexual in®delity. few times a week meant a 22% increase in the Respondents who had had sex with a primary odds of ever having had extramarital sex. partner at least 10 times over the past year were The hypothesized link between in®delity and asked about expectations for sexual ®delity. As opportunities for undetected sex received mixed Table 2 reports, nearly 99% of married persons support. For once-married persons in central cit- expected their to have sex only in mar- ies, compared to other communities the odds of riage, and 99% assumed their partner expected extramarital sex were 39% higher. Partners' sexual exclusivity of them. Many important as- shared networks showed the predicted negative pects of life together (e.g., whether to have chil- association: All things being equal, enjoying time dren, how to meet expenses, who should wash the spent with a mate's family lowered the odds of dishes) might be rethought as circumstances extramarital sex by 24%. Prior sexual experience, change, but expectations for sexual exclusivity are attendance at religious services, and work-place not negotiable. In other tabulations not shown, we opportunities for extramarital sex were statistical- found less than 1% of heterosexuals, married or ly insigni®cant. cohabiting, reported that a partner had changed As for the couple's relationship, no statistically expectations for ®delity during the relationship. signi®cant positive association was found for any Although cohabitors held less conventional measures of social dissimilarity. However, living gender and family values (Clarkberg et al., 1995), together before marriage raised the net odds of cohabiting heterosexuals were only slightly less marital in®delity by 39%, even controlling for likely (94% versus 99%) to expect sexual exclu- sexual values and frequency of attendance at re- sivity than married persons who had never lived ligious servicesÐvariables that distinguish mar- together (t ϭ .333, p Ͻ .001). Once married, those ried couples who ®rst cohabited from the more who had once lived together (98%) held expec- conventional married couples who did not. tations that were not signi®cantly different from Gender and race were statistically signi®cant. the expectations of other married people (99%) (t All things considered, being male increased the ϭ .104, ns; p ϭ .30). odds of having engaged in extramarital sex by Were respondents sexually exclusive? People 79%. Being African-American raised them by who married without ®rst cohabiting were no 106%, even though education controlled for racial more faithful (92%) than either married persons differences in socioeconomic status. Education Sexual In®delity 55

TABLE 3. LOGISTIC REGRESSION RESULTS FOR EXTRAMARITAL SEX:AMERICANS AGES 18±59, MARRIED ONLY ONCE, 1992 Self-Recorded Interview Married Once Married Once Without Cohabiting Variable beb beb beb Sexual tastes and values Sexual interest .197** 1.22 .212* 1.24 .324** 1.38 Nonpermissive Ϫ.881*** .42 Ϫ.921*** .40 Ϫ1.084*** .34 Opportunities Sexual experience .006 1.01 .013 1.01 .010 1.01 Workplace opportunity .085 1.09 .128 1.14 Ϫ.069 .93 Central city .331* 1.39 .234 1.26 .420 1.52 Shared networks Ϫ.268*** .76 Ϫ.300*** .74 Ϫ.466*** .63 Religious attendance Ϫ.002 1.00 Ϫ.003 .99 Ϫ.004 1.00 Relationship Religious difference .250 1.28 .076 1.08 Ϫ.044 .96 Education difference .295 1.34 .249 1.28 .315 1.37 Age difference Ϫ.178 .84 Ϫ.279 .76 Ϫ.579 .56 Couple cohabited .328* 1.39 Demographic and control Male .582*** 1.79 .951*** 2.59 .660** 1.94 African American .721*** 2.06 1.053*** 2.87 .976** 2.65 Education Ϫ.095* .91 Ϫ.082 .92 Ϫ.104 .90 Frankness .436** 1.55 .418* 1.52 .557** 1.75 Marital duration .021** 1.02 .032*** 1.03 .021* 1.02 Constant Ϫ1.050 Ϫ1.333 Ϫ1.046 Chi-square 230.16 151.70 164.69 Degrees of freedom 16 15 15 n 1717 1102 1102 % Ever unfaithful 15.5 13.5 10.5 Note: eb ϭ exponentiated beta. *p Ͻ .05. **p Ͻ .01 ***p Ͻ .001. One-tailed signi®cance tests. showed a weak negative association. Both frank- showed somewhat greater likelihood of in®delity ness and marital duration (i.e., exposure time) only at the extremes of the education distribution showed the expected positive relationships. (eighth grade or less and master's degree or high- Before analyzing person-to-person interview er), categories that together encompassed only 5% reports of in®delity for both married and cohab- of the sample. iting persons, we compared interview and self-re- Regardless of how extramarital sex was mea- corded measures of the cumulative incidence of sured, sexual interest showed the predicted posi- extramarital sex. Because the interview item did tive relationship, and nonpermissive values not distinguish extramarital sex from cohabiting showed the hypothesized negative relationship. in®delities, Table 3 focuses on those who married Shared networks were negatively related to extra- only once and who did not live with the mate marital sex in both interview and self-recorded before marriage. The self-recorded item yielded a data. The fact that demographic factors were also higher estimate of extramarital sex (13.1%) than consistent suggests little bias results from mode of the person-to-person interview item (10.5%), data collection. Given the general comparability which did not count in®delities between separa- of the self-recorded and interview data, we turn tion and divorce. The patterns were quite consis- to the analysis of in®delity reported to interview- tent: Sexual interest, nonpermissive values, shared ers. networks, gender, race, marital duration, and frankness were statistically signi®cant and in the Cumulative incidence for married and cohabiting anticipated direction for both interview and self- persons. Both cohabitors and married people ex- recorded items. Neither central city residence nor pect sexual exclusivity. Interview data measured education was signi®cant. Closer inspection whether respondents ever had a secondary sex 56 Journal of Marriage and the Family

TABLE 4. LOGISTIC REGRESSION RESULTS FOR SEXUAL :EVER-MARRIED AND EVER-COHABITED AMERICANS, AGES 18±59, 1992 Model A Model B Model C Model D Variable beb beb beb beb Sexual tastes and values Sexual interest .287*** 1.33 .283*** 1.33 .259*** 1.30 Nonpermissive Ϫ.688*** .50 Ϫ.671*** .51 Ϫ.627*** .53 Opportunities Sexual experience .006* 1.01 Workplace opportunity .059 1.06 Central city .386** 1.47 Shared networks Ϫ.300*** .74 Religious attendance Ϫ.004 1.00 Relationship Religious difference .002 1.00 Ϫ.108 .90 Education difference .059 1.06 .084 1.09 Age difference Ϫ.596 .55 Ϫ.595 .55 Couple cohabited .338* 1.40 .258 1.29 Demographic and control Male .786*** 2.20 .442** 1.56 .409** 1.51 .385** 1.47 African American .600*** 1.82 .600*** 1.82 .588*** 1.80 .504** 1.65 Education .015 1.01 Ϫ.021 .98 Ϫ.027 .97 Ϫ.029 .97 Frankness .298** 1.35 .290** 1.34 .281* 1.32 .305** 1.36 Duration .012* 1.01 .022*** 1.02 .031*** 1.03 .034*** 1.03 Constant Ϫ3.855 Ϫ2.137 Ϫ2.310 Ϫ2.131 Chi-square 52.84 156.50 169.94 222.84 Degrees of freedom 5 7 11 16 n 2,598 2,598 2,598 2,598 Note: eb ϭ exponentiated beta. *p Ͻ .05. **p Ͻ .01 ***p Ͻ .001. One-tailed signi®cance tests. partner while married or cohabiting. Table 4 much diminish the effect of being African-Amer- shows four sets of independent variables. These ican. are considered step-wise, beginning with the de- The complete Model D again shows that those mographic and control variables and proceeding with strong interest in sexÐthose apt to gain most from individual tastes to the microsocial couple from sexual encountersÐwere signi®cantly more relationship and then to macrosocial sexual op- likely to have been unfaithful. Those facing stiffer portunity in¯uences. Each set of factors added sig- personal costsÐfor example, those with nonper- ni®cantly to the explanation of sexual in®delity, missive valuesÐwere signi®cantly less likely to as indicated by increments to the ␹2 statistic, but engage in in®delity. relationship measures added the least. Early sexual experience and central city resi- Note that gender's effect was markedly re- dence were positively associated with the likeli- duced when we added other variables hypothe- hood of having ever been unfaithful. Each addi- sized to affect sexual decision-making: Being tional sex partner between age 18 and the ®rst male raised the odds of in®delity by 120% in union increased the net odds of in®delity by 1%, Model A, which contains only demographic and compared to a 47% increase associated with living control variables, but by only 56% when sexual in a central city. Sharing a mate's social network interest and nonpermissive values were consid- was negatively associated with in®delity. All ered. Other analyses (not shown) demonstrated things considered, befriending a partner's family that controlling for permissiveness eliminated was associated with a 26% decrease in the odds most gender differences in in®delity: Because of sexual in®delity. Workplace sexual opportuni- men's sexual values are more permissive, men ties and religious service attendance were not sta- faced fewer impediments to in®delity. By contrast, tistically signi®cant. Relationship measures of het- variables in¯uencing decision-making did not erogamy were statistically insigni®cant, too. Sexual In®delity 57

TABLE 5. LOGISTIC REGRESSION RESULTS FOR SEXUAL INFIDELITY DURING LAST 12 MONTHS:MARRIED AND COHABITING AMERICANS,AGES 18±59, 1992 Comparable Model Causal Variables Only Full Model Variable beb beb beb Sexual tastes and values Sexual interest .267* 1.31 .304** 1.36 Nonpermissive Ϫ.718*** .49 Ϫ.675*** .51 Opportunities Sexual experience .010* 1.01 .014*** 1.01 .011* 1.01 Workplace opportunity .233* 1.26 .206* 1.23 .235* 1.26 Central city .389 1.48 .602** 1.83 .358 1.43 Shared networks Ϫ.273*** .76 Ϫ.203* .82 Religious attendance Ϫ.013* .99 Ϫ.013* .99 Relationship Religious difference .089 1.09 .436* 1.55 .058 1.06 Education difference Ϫ.026 .97 Ϫ.003 1.00 Ϫ.116 .89 Age difference Ϫ1.086 .34 Ϫ.915 .40 Ϫ1.142 .32 Couple cohabited .505* 1.66 Couple cohabiting .759** 2.14 Duration Ϫ.008 .99 Dissatis®ed .247*** 1.28 Demographic and control Male .210 1.23 .673** 1.96 .300 1.35 African American .904** 2.47 .767** 2.15 .893** 2.44 Education Ϫ.046 .96 Ϫ.044 .96 .004 1.00 Age 18±30 .698* 2.01 .202 1.22 Age 31±40 .075 1.08 Ϫ.269 .76 Age 41±50 Ϫ.098 .91 Ϫ.360 .70 Frankness .557** 1.75 .513* 1.67 .579** 1.78 Constant Ϫ3.535 Ϫ6.007 Ϫ5.093 Chi-square 136.11 66.50 157.81 Degrees of freedom 16 13 19 n 2,010 2,010 2,010 Note: eb ϭ exponentiated beta. *p Ͻ .05. **p Ͻ .01 ***p Ͻ .001. One-tailed signi®cance tests.

Cohabitation fell short of statistical signi®cance at ble 5 to Model D in Table 4 shows the effect of the .05 level. the time referent: Results for short-term in®delity largely paralleled those from earlier analyses. Per- Prevalence of in®delity for married and cohabit- sonal tastes and values were signi®cantly associ- ing persons. Sexual interest might prompt in®del- ated with the likelihood of in®delity. With the ex- ity, but in®delity might stimulate interest in sex, ception of central city residence, so were leading to frequent erotic thoughts. Although per- measures of sexual opportunities. Even workplace missive values no doubt encourage adultery, adul- opportunity, which was not statistically signi®cant terers might rationalize their behavior by adopting in earlier analyses, showed the hypothesized pos- permissive views. Data did not permit us to sort itive relationship for the previous 12 months. This out all causal relationships, if only because the suggests that characteristics of the job mattered, timing of in®delity was not known for cumulative but the current job's social interactions did not incidence measures. Independent variables mea- adequately capture previous work conditions in- sured at the interview date did not necessarily cor- ¯uencing past sexual behavior. Although previous respond to prior circumstances that gave rise to cohabitation increased the odds of in®delity, other in®delity. Focusing on the year before the inter- features of the relationship did not prove statisti- view clari®ed causation by narrowing the time cally signi®cant. Union duration, introduced here frame for in®delities. to test habituation versus investment hypotheses, Table 5 displays the results of in®delity in the was not signi®cant either, implying that the pas- prior 12 months. Comparing the ®rst model in Ta- sage of time had no effect on the marital gains 58 Journal of Marriage and the Family that would be jeopardized by sexual in®delity. than twice as high for cohabitors than for married Gender's effectÐreduced when other factors were persons. Although cohabitation increased the like- controlledÐwas not statistically signi®cant at lihood signi®cantly, we did not ®nd the predicted even the .05 level for the previous 12 months. association between partners' social dissimilarity Race continued to be strongly signi®cant, how- and sexual in®delity. Subjective dissatisfaction, ever. however, was positively and signi®cantly associ- To address issues of causation, the second ated with the likelihood of in®delity in the pre- model in Table 5 includes only variables reason- ceding 12 months. Most people reported high sat- ably assumed to precede any in®delity in the pre- isfaction with both the emotional and physical ceding 12 months. These include sexual experi- aspects of their union, but the exponentiated beta ence between age 18 and the ®rst union, stable implied that the net odds of in®delity increased demographic characteristics (e.g., gender, race, ed- 28% when one was, say, merely ``very'' pleased ucation, age), and relatively ®xed partner dissim- as opposed to ``extremely'' so. ilarities (e.g., education differences). ``Causal'' Subjective perception of the relationship was variables also included city residence and work- more closely associated with in®delity than objec- place opportunities. Although one might posit that tive heterogamy measures, whichÐalthough sta- adultery prompts people to move to a big city, it ble and causally prior to in®delityÐdid not dem- is more plausible to assume that urban residence onstrate uniform effects on sexual behavior encourages adultery, particularly because current (Forste & Tanfer, 1996). Whether subjective dis- residence is a much weaker predictor of permis- satisfaction prompted in®delity or vice versa, any sive sexual values than is the size of the com- effect might be relatively short-term, especially if munity in which one was raised (Stephan & unhappy partners either reconciled or separated McMullin, 1982). Nor is it likely that sexual tastes after an in®delity. If subjective evaluations of the dominate the job search process. match were not very stable, this might explain Sexual opportunity measuresÐearly sexual ex- why prior studies did not always ®nd current mar- perience, workplace opportunity, central city res- ital evaluations to be signi®cantly associated with idenceÐwere all statistically signi®cant. Heter- cumulative incidence of in®delity over the course ogamy in education and age showed no of a union. hypothesized effects, but dissimilar religions raised the likelihood of sexual in®delity, suggest- DISCUSSION ing that excluded variables like sexual values and religious service attendance accounted for the re- Although previous research has reported personal ligious heterogamy effect. Gender and race were values, sexual opportunities, and the marital rela- statistically signi®cant. Youngest ages were asso- tionship as determinants of extramarital sex, these ciated with greater in®delity; all things being studies have been largely piecemeal and based on equal, the odds of in®delity were twice as high for small samples of limited generalizability. To the those ages 18±30 as for those over 50. best of our knowledge, our research is the ®rst to The third model in Table 5 adds attitudes and include measures of all three sets of determinants relationships possibly affected by the experience in multivariate analyses based on a large, repre- of sexual in®delity itself: sexual values, interest in sentative sample of the U.S. population. The anal- sex, attendance at religious services, social net- yses show that values, opportunities, and the mar- works, and whether a couple is cohabiting or mar- ital relationship are associated with sexual ried. Also included is dissatisfaction with the pri- in®delity, even when other factors and demo- mary relationship's emotional and physical graphic risk variables are controlled. rewards, a subjective measure available only with As we predicted, people who were more inter- the prevalence measure of in®delity. ested in sex were more likely to have multiple This model, too, shows that tastes and values partners. As we hypothesized, people with non- were statistically signi®cant. When ``cosmopolitan'' permissive values were less likely to engage in values were incorporated, the effect of central city sexual in®delity. Considering sexual opportuni- residence ceased to be statistically signi®cant. ties, we found evidence that prior sexual experi- Other measures of sexual opportunityÐwhether ences were positively associated with in®delity. clearly prior or less remote in timeÐshowed the The behavioral constraints posed by overlap of predicted associations. Controlling for other vari- mates' social networks reduced the likelihood of ables, the odds of a recent in®delity were more in®delity. In the short run, so did involvement in Sexual In®delity 59 a religious community: Those who often attended racial differences in the sex ratio might in¯uence religious services were less likely to have had the likelihood of having multiple partners. multiple sex partners in the previous year, even We argue for thinking about sexual in®delity when sexual values associated with religiosity as the product of rational decision-making. As- were controlled. Sexual opportunities of the work- suming sexual behavior is subject to rational cal- place also increased the likelihood of in®delity culation, we derived a series of testable hypothe- during the last 12 months. At least in the short ses. NHSLS measures did not permit us to run, however, any effect of city residence was sub- examine intrapsychic, cognitive processes or to stantially reduced when ``cosmopolitan'' sexual compare directly preferences for alternative cours- values and tastes were controlled. es of behavior. To the extent that preferences are The nature of the primary relationship proved revealed in behavior, however, we can evaluate important. We found cohabitors more likely than our approach by asking whether empirical results married people to engage in in®delity, even when are consistent with predictions. Indeed, they are we controlled for permissiveness of personal val- largely consistent even given different operation- ues regarding extramarital sex. This ®nding sug- alizations of the dependent variable, sexual in®- gests that cohabitors' lower investments in their delity. unions, not their less conventional values, ac- Previous research reported that sexual in®del- counted for their greater risk of in®delity. Cohab- ity is associated with values, opportunities for se- itors who went on to marry were no less likely to cret sex, the quality of the primary relationship, demand sexual exclusivity than people who mar- and sociodemographic risk factors. Integrating ried without having lived together. Neither the ha- these piecemeal ®ndings into a uni®ed model re- bituation nor the investment hypothesis about the vealed some well-documented relationships to be effects of union duration was empirically sup- spurious. Our multivariate model also clari®ed the ported. mechanisms by which variables might in¯uence As for measures of marital quality, partners' in®delity. For example, differences in tastes and social dissimilarity was statistically insigni®cant, values largely accounted for the effects of city res- but subjective dissatisfaction with a union was as- idence and male gender. Controlling for sexual sociated with greater likelihood of recent in®del- values, however, did not eliminate the signi®cant ity. Prior studies yielded inconsistent results on association between in®delity and cohabitation, a whether poor relationships led to extramarital sex. result that pointed to commitment mechanisms as Our ®nding underscores the need to attend to op- likely in¯uences on sexual behavior. Nor could erationalizing relationship ``quality'' and to sort- sexual values account for the negative association ing out causal order. Current relationship quality of church-going and recent in®delity. The multi- might not demonstrate an association with cu- variate analysis suggested that religiosity con- mulative incidenceÐthat is, having ever been un- strained sexual behavior not only through inter- faithful: Relationship problems were apt to be nalized moral beliefs, but also via supportive short-term if couples either reconciled or divorced social networks. The integrated model pointed to soon after an in®delity. one clear result: Being subject to preferences, con- Although epidemiological research consistent- straints, and opportunities, sexual behavior is so- ly reports men to be at higher risk of in®delity cial behavior. than women, studies have not usually included in- dicators of sexual values and tastes. When we NOTE controlled for interest in sex and permissiveness of sexual values, we found that the main effects This research was supported by a grant from the Na- tional Science Foundation (SBR-9730171). An earlier of gender were markedly reduced or even elimi- version of this paper was presented at the annual meet- nated. Consistent with prior research, we found ing of the American Sociological Association, New that being African-American was positively as- , 1996. sociated with multiple sex partners, even when ed- ucational attainment (an indicator of socioeco- REFERENCES nomic status) and other variables were controlled. The persistence of this effect points to the need Atwater, L. (1982). The extramarital connection: Sex, intimacy, and identity. New York: Irvington. for further research to clarify the role of race. Be- Bell, R. R., Turner, S., & Rosen, L. (1975). A multi- cause we found the sexual opportunity structure variate analysis of female sexual coitus. Journal of to be important in understanding sexual behavior, Marriage and the Family, 37, 375±384. 60 Journal of Marriage and the Family

Blumstein, P., & Schwartz, P. (1983). American couples. coitus. Journal of Marriage and the Family, 32, 449± New York: Morrow. 456. Bozon, M. (1996). Reaching adult sexuality: First in- Lancaster, J. (1994). , life histories, and tercourse and its implications. In M. Bozon & H. Ler- evolutionary ecology. In A. S. Rossi (Ed.), Sexuality idon (Eds.), Sexuality and the social sciences (pp. across the life course (pp. 39±62). Chicago: Univer- 143±175). Aldershot, England: Dartmouth. sity of Chicago Press. Brown, E. M. (1991). Patterns of in®delity and their Laumann, E. O., Gagnon, J. H., Michael, R. T., & Mi- treatment. New York: Brunner-Mazel. chaels, S. (1994). The social organization of sexual- Bumpass, L. L., Sweet, J. A., & Cherlin, A. (1991). The ity. Chicago: University of Chicago Press. role of cohabitation in declining rates of marriage. Lawson, A. (1988). Adultery: An analysis of love and Journal of Marriage and the Family, 53, 913±927. . New York: Basic Books. Choi, K.-H., Catania, J. A., & Dolcini, M. M. (1994). Lehrer, E. L., & Chiswick, C. U. (1993). Religion as a Extramarital sex and HIV risk behavior among Amer- determinant of marital instability. Demography, 30, ican adults: Results from the National AIDS Behav- 385±404. ioral Survey. American Journal of Public Health, 84, Leigh, B. C., Temple, M. T., & Trocki, K. F. (1993). The 2003±2007. sexual behavior of US adults: Results from a national Clarkberg, M., Stolzenberg, R. M., & Waite, L. J. survey. American Journal of Public Health, 83, (1995). Attitudes, values, and entrance into cohabi- 1400±1408. tational versus marital unions. Social Forces, 51, Maykovich, M. K. (1976). Attitudes versus behavior in 609±633. extramarital sexual relations. Journal of Marriage Constitutional barriers to civil and criminal restrictions and the Family, 38, 693±699. on pre- and extramarital sex. (1993). Harvard Law Pittman, F. (1989). Private lives: In®delity and the be- Review, 104, 1660±1680. trayal of intimacy. New York: W. W. Norton. Dolcini, M. M., Catania, J. A., Coates, T. J., Stall, R., Prins, K. S., Buunk, B. P., & VanYperen, N. W. (1983). Hudes, E. S., Gagnon, J. H., & Pollack, L. M. (1993). Equity, normative disapproval, and extramarital re- Demographic characteristics of heterosexuals with lations. Journal of Social and Personal Relationships, multiple partners: The National AIDS Behavior Sur- 10, 39±53. veys. Family Planning Perspectives, 25, 208±214. Reiss, I. L., Anderson, R. E., & Sponaugle, G. C. Edwards, J. N., & Booth, A. (1976). Sexual behavior in (1980). A multivariate model of the determinants of and out of marriage: An assessment of correlates. extramarital sexual permissiveness. Journal of Mar- Journal of Marriage and the Family, 38, 73±83. riage and the Family, 42, 395±411. Edwards, J. N., & Booth, A. (1994). Sexuality, mar- Reiss, I. L., & Miller, B. C. (1979). Heterosexual per- missiveness: A theoretical analysis. In W. R. Burr, R. riage, and well-being: The middle years. In A. S. Ros- Hill, F. I. Nye, & I. L. Reiss (Eds.), Contemporary si (Ed.), Sexuality across the life course (pp. 233± theories about the family (Vol. 1, pp. 57±100). New 299). Chicago: University of Chicago Press. York: Free Press. Forste, R., & Tanfer, K. (1996). Sexual exclusivity Siegel, M. J. (1992). For better or worse: Adultery, among , cohabiting, and married women. Jour- crime and the Constitution. Journal of , nal of Marriage and the Family, 58, 33±47. 30, 45±95. Gagnon, J., & Simon, W. (1973). Sexual conduct: The Smith, T. W. (1991). Adult sexual behavior in 1989: social sources of human sexuality. Chicago: Aldine. Numbers of partners, frequency of intercourse and Giddens, A. (1992). The transformation of intimacy: risk of AIDS. Family Planning Perspectives, 23, Sexuality, love and eroticism in modern societies. 102±107. Cambridge, England: Polity Press. Smith, T. W. (1994). Attitudes toward sexual permis- Gilmartin, B. G. (1974). Sexual deviance and social net- siveness: Trends, correlates, and behavioral connec- works: A study of social, family, and marital inter- tions. In A. S. Rossi (Ed.), Sexuality across the life action patterns among co-marital sex participants. In course (pp. 63±97). Chicago: University of Chicago J. R. Smith & L. G. Smith (Eds.), Beyond Press. (pp. 291±323). Baltimore, MD: Johns Hopkins Uni- Stephan, E. G., & McMullin, D. R. (1982). Tolerance versity Press. of sexual nonconformity: City size as a situational Glass, S. P., & Wright, T. L. (1992). Justi®cations for and early learning determinant. American Sociolog- extramarital relationships: The association between ical Review, 47, 411±415. attitudes, behaviors, and gender. The Journal of Sex Vaughn, D. (1986). Uncoupling. Oxford, England: Ox- Research, 29, 361±387. ford University Press. Greeley, A. M. (1991). Faithful attraction. New York: Wellings, K., Field, J., Johnson, A., & Wadsworth, J. A Tom Doherty Associates Book. (1994). Sexual behavior in Britain. : Penguin Johnson, R. E. (1970). Some correlates of extramarital Books.

Top View