Trends in Same-gender Sexual Partnering, 1988-1998.

In 1948, Kinsey and his colleagues noted that, given the extent of homosexual activity present in society despite considerable social and legal constraints, there would surely be an increase in such activity if those constraints were to disappear (Kinsey, Pomeroy, & Martin, 1948). Social and legal constraints against same-gender sexual activity have not disappeared, but they have declined in recent decades. Thus, it may be becoming easier for people to recognize and act on their sexual attraction to others of their own gender. If so, then it follows that there would have been an increase during recent years in same-gender sexual partnering. This study examines whether there was an increase in same-gender sexual partnering between 1988 and 1998. It also examines the relationship between same-gender sexual partnering and social-demographic characteristics.

A decline in cultural sanctions against same-gender sexual activity is evident on a number of fronts. Sodomy laws have been repealed by the legislatures in 25 states and the District of Columbia and have been struck down by the courts in 6 other states (American Civil Liberties Union, 1999). A number of states, counties, and municipalities now prohibit discrimination based on sexual orientation in private employment (Human Rights Campaign, 1999), and the early 1990s witnessed the fastest increase in the inclusion of sexual orientation in local civil rights codes (Button, Rienzo, & Wald, 1997). Whereas in 1990 fewer than six companies provided domestic partnership benefits to same-sex couples, by 1999 hundreds of businesses did (Human Rights Campaign, n.d.). There has also been a gradual trend toward granting parental rights to gay and lesbian parents (Stacey, 1998). In addition, the visibility of gay men and lesbians has increased in the media, and gay men and lesbians tend to be portrayed in a more favorable light than they had been in earlier decades (Kaiser, 1997; Parish, 1993).

Attitudes toward same-gender sexual behavior have become more tolerant among the general public since the late 1980s. After increasing during the 1970s and 1980s, the percentage of adults in the General Social Surveys who reported that sexual activity between two adults of the same sex is always wrong dropped from 74.9% in 1988 to 54.6% in 1998. (Samples sizes were 973 in 1988 and 1,874 in 1998.) Moreover, Americans are increasingly likely to agree that a man "who admits that he is a homosexual" should be allowed to make a speech in the community and to teach in a college or university, and they are increasingly less likely to favor removing from the local public library "a book he wrote favoring homosexuality." Figure 1 shows these trends in American public opinion from 1973 to 1998.

[Figure 1 ILLUSTRATION OMITTED]

There have also been changes in social and economic conditions that may have made it more likely for women to consider other women as sex partners. Women's wages have risen during the last couple of decades (Bianchi, 1995; Wetzel, 1995), thereby increasing women's financial ability to support themselves and their families. At the same time, the wages of men with a high school education or less have declined, making them less attractive as mates. Between 1979 and 1992, the earnings of full-time, year-around female workers rose from 60% to 71% of the earnings of full-time, year-around male workers (Bianchi, 1995). Equalizing the earning potential of men and women may enable women to consider family structures and sexual partnerships that do not include men. But perhaps just as important as current wage levels and female/male earnings ratios is that, as more recent generations replace earlier ones, there is an increasing proportion of women in the population who have grown up having internalized a broad set of future possibilities. These include a college education, interesting work, and economic independence. It is probable that girls born since the resurgence of the feminist movement in the 1970s and after the rise in divorce rates in the 1970s were less likely than earlier generations of girls to expect to be economically dependent on a man as an adult. Consequently, as adults they may be more likely than earlier generations of women to consider having a sexual relationship with another woman. Because men have generally not depended upon women for financial support, recent social and economic changes should have had a smaller impact on the gender of their sex partners.

The idea that the relaxing of constraints against same-gender sexual relationships behavior will result in an increase in same-gender sexual relationships is consistent with a social constructionist view of human sexuality. This theory posits that while the sexual drive is a biological phenomenon, many aspects of its expression, including who is considered an appropriate sex partner, are shaped by culture (see DeLamater & Hyde, 1998, for a discussion). Thus, it follows that as the culture becomes more accepting of a variety of forms of sexual expression, including same-gender sexual behavior, there will be an increase in these behaviors. However, increases in same-gender sexual behavior would not be inconsistent with an essentialist perspective. Even if people are either heterosexual or homosexual by nature, as essentialists argue, we might still see changes over time in the sexual behavior of homosexuals as the culture makes it easier or more difficult for them to express their true, or "essential," nature.

Large-scale national surveys of sexual behavior are a relatively recent phenomenon, and so we have been unable to examine empirically whether same-gender sexual behavior varies in its prevalence over time as cultural and other environmental conditions change. Laumann, Gagnon, Michael, and Michaels (1994) found no evidence of a trend in General Social Survey (GSS) data over the 1988-1993 period, but the cultural changes during that short time period may have been too slight to produce a change in sexual behavior that could be detected with survey data.

There is other evidence that same-gender sexual behavior may be sensitive to variation in environmental and cultural conditions. Carrier (1985), Magana and Carrier (1991), and Paz (1950) have pointed out that in Mexico, men who engage in same-gender sexual activity tend to restrict themselves to either an insertive or a receptive role, and men who play the insertive role are not stigmatized as "homosexual." It would follow, then, that sexual activity among Mexican men might be more common than it is in cultures where any sexual activity between men is stigmatized. Indeed, Billy, Tanfer, Grady, and Klepenger (1993), who studied a nationally representative U.S. sample of men ages 20 to 39, found that Hispanic men were more likely to have engaged in same-gender sexual activity than non-Hispanic men during the previous 10 years.

Laumann et al. (1994) combined the GSS data with data from the 1992 National Health and Social Life survey and found additional evidence that same-gender sexual behavior may be sensitive to variation in environmental and cultural conditions. First, having had a same-gender sex partner was positively associated with size of home town when growing up among men (but not among women), suggesting that the characteristics of large cities make it easier for men to explore nonnormative forms of sexuality than would be possible in smaller cities and towns. Second, educational attainment was positively associated with same-gender attraction, appeal, desire, and sex partners among both men and women. This suggests that as people become more open to new ideas through education, they become more in touch with socially stigmatized aspects of their sexuality. And because more educated people tend to work and socialize with similar others, the greater tolerance of same-gender sexuality among more educated people (Herek, 1984) is likely to create a more conducive setting for the development of same-gender sexual interest. Third, Laumann et al. found that younger people were more likely to report having had a same-gender sex partner since age 18 than were older people, despite having had fewer years to do so. Growing up during a more sexually permissive historical period should make it easier for people to explore same-gender sexuality. Laumann et al., however, did not report whether these relationships were statistically significant. Other studies have found limited Corroborating evidence. Billy et al. (1993) found the occurrence of same-gender sexual activity during the past 10 years to be positively associated with educational attainment. Rogers and Turner (1991), who combined data on same-gender sexual contact among males from the GSS, the 1970 Kinsey survey, and a local 1989 survey conducted in Dallas, Texas, found a positive relationship for same-gender sexual contact with father's education, but not with size of place growing up, age, or the respondent's own education. Thus, there is some indication that same-gender sexual behavior is influenced by cultural and environmental factors, but the evidence from large-scale social surveys is inconsistent.

One reason why it is difficult to determine the social-demographic correlates of same-gender sexual behavior is that small proportions of people report having had a same-gender sex partner. In the study by Laumann et al. (1994), for example, 2.7% the men and 1.3% of the women reported having had a same-gender sex partner during the previous 12 months. Consequently, very large probability samples are necessary to obtain statistical significance for effects of moderate magnitude. The current study has a larger sample size than earlier studies and consists of a series of cross-sectional surveys administered over an 11-year period. This makes it fruitful to examine whether the likelihood of same-gender sex partnering has changed over the 1988-1998 period, and to revisit the question of whether social-demographic characteristics of individuals are associated with same-gender sex partnering.

In this paper, I examine whether during the 1988-1998 period there was an increase in the likelihood that American men and women would report having had a same-gender sex partner during the previous year and during the previous 5 years. If an increase is found, it will be of interest whether the increase can be explained by a decline in exclusively opposite-gender sex partnering or by a decline in sexual abstinence. Moreover, we will want to rule out the possibility that the increase is simply an artifact explained by an increasing likelihood that people are willing to answer survey questions about sex. Therefore, I examine whether there were changes in the likelihood that Americans reported having had no partners or exclusively opposite-gender partners, and whether they failed to answer the questions about the gender of their partners. Next, I examine the relationships between a series of social and demographic characteristics of individuals and same-gender sexual partnering. If social-demographic factors are associated with same-gender partnering, then changes in the social-demographic composition of the population, rather than the relaxation of social mores and economic and legal restrictions, may be responsible for any increase in the likelihood of same-gender partnering. To rule out this competing explanation, I examine whether increases in same-gender partnering over the 1988-1998 period can be explained by the changing social and demographic characteristics of the population.

METHOD

Data

The data for this study came from the General Social Surveys (GSS) conducted by the National Opinion Research Center. The GSS is a series (once annual, now every other year) of cross-sectional, nationally representative samples of civilian, noninstitutionalized, English-speaking adults age 18 and older (see Davis & Smith, 1992, for a description of the General Social Surveys). Since 1988, the GSS has included questions on sex partners. Eight waves of the survey were used for the current study: 1988, 1989, 1990, 1991, 1993, 1994, 1996, and 1998. The sample consisted of 11,355 respondents age 18 to 59; 5,063 were men and 6,292 were women. Of these, 154 men and 117 women reported a same-gender sex partner during the previous year. Questions about sex partners during the previous 5 years were introduced in 1991. Of the 8,393 respondents interviewed since 1991, 3,726 were men and 4,667 were women. One hundred and fifty-four men and 136 women reported having had a same-gender sex partner during the previous 5 years.

I restricted the sample to adults age 59 and younger, because a relatively large percentage of older people reported having had no sex partner in recent years. Fifty-three percent of older respondents, age 60 and older, reported no sex partner in the previous year, compared to only 12% of younger respondents, ages 18 to 59. Rates for no sex partners in the previous 5 years were 40% and 6% for older and younger respondents respectively. The cutoff age of 59 was also chosen to be consistent with the research conducted by Laumann et al. (1994).

The GSS sampling design excludes people living in institutions and group quarters, such as college dormitories, prisons, and military barracks. The exclusion of these groups of primarily young adults from the sample may produce some downward bias in the estimates of same-gender partners, but it is unlikely to affect the trend over the 1988-1998 period.

Measures

Dependent variables. Same-gender sexual activity was measured in two ways: whether the respondent reported having engaged in any same-gender sexual activity during the previous year and during the previous 5 years. The method of data collection used in the General Social Survey is primarily a face-to-face interview, but the questions about sexual behavior were administered as a self-completion component. Respondents were given a form that included questions about their sexual behavior that they filled out in private and then returned to the interviewer in a sealed envelope (Smith, 1992). The questions were prefaced with the following statement:(1)

   There is a great deal of concern today about the AIDS epidemic and how to
   deal with it. Because of the grave nature of this problem, we are going to
   ask you some personal questions and we need your frank and honest
   responses. Your answers are confidential and will be used only for
   statistical reports.

The following four questions were used for the current study.

How many sex partners have you had in the last 12 months? PLEASE CIRCLE ONE ANSWER.

Respondents circling "no partners" skipped the next question.

Have your sex partners in the last 12 months been ... PLEASE CIRCLE ONE ANSWER.

Exclusively male

Both male and female

Exclusively female

Now, think about the past five years--the time since February/ March 19--, and including the past 12 months, how many sex partners have you had in that five-year period?

Respondents circling "no partners" skipped the next question.

Have your sex partners in the last five years been ... PLEASE CIRCLE ONE ANSWER.

Exclusively male

Both male and female

Exclusively female

The sexual behavior questions appeared as a supplement at the end of the interview. As reported by Smith (1992), the percentages of GSS respondents who completed the sexual behavior supplement were 93.9% in 1988, 91.2% in 1989, and 85.5% in 1990. About one half of the respondents who did not complete the sexual behavior supplement had ended the interview before they reached the supplement and so their failure to continue the interview cannot be attributed to the nature of the questions. A small proportion of respondents who completed the supplement failed to answer one or more of the sexual behavior questions. This appears to have been primarily due to either confusion about skip patterns or to the respondent's general uncooperativeness, as demonstrated by his or her refusal to answer a question about income and having answered don't know to a number of questions earlier in the interview. There is no indication that either supplement nonresponse or item nonresponse was caused by a reluctance to admit to having had same-gender sex partners (Smith, 1992). Nonresponse did not vary by gender of the respondent (Smith 1992).

I coded men as having had a same-gender sex partner in the past year if they reported partners in the past 12 months who were either exclusively male or both male and female. I coded women as having had a same-gender sex partner in the past year if they reported partners in the past 12 months who were either exclusively female or both male and female. Respondents who reported only opposite-gender partners or no partners were coded accordingly. Respondents who filled out the sexual behavior supplement who had reported having had at least one partner but left blank the item on gender of partner were coded as not having answered the question. Respondents who ended the interview before beginning the supplement were treated as missing data. A similar procedure was used to code sex partners in the past 5 years.(2)

Independent variables. Year of the interviews ranged from 1988 to 1998. It was recoded for the multivariate analysis so that 1988 = 0, 1989 = 1, 1990 = 2, and so forth.

The effect of age on sexual behavior can be a function of at least two factors: a developmental effect and a cohort effect. Thus, older people may be less likely to have had a recent same-gender partner because they tend to be less sexually active than younger people are. Or, the effect may be due to their having been socialized during an historical period when same-gender sexual activity was considered unacceptable. The use of longitudinal data allows us to begin to sort out the two influences. The current data set provides us with same-aged respondents born over an 11-year period. Age was coded as 18 to 29, 30 to 39, 40 to 49, and 50 to 59 years old at the time of the interview. Birth cohort was categorized by decade of birth: 1929-1939, 1940-1949, 1950-1959, 1960-1969, and 1970-1980.

Race/ethnicity was determined from responses to two questions: "what race do you consider yourself?" and "from what countries or part of the world did your ancestors come?" Respondents were coded as either White, Black, or Other based on their answer to the first question, and were recoded as Hispanic if their ancestors came from Mexico, Puerto Rico, or "other Spanish."

Size of hometown at age 16 was coded as rural (lives in open country or on a farm), small or medium-sized city (population 250,00 or less), and large city or its suburb (city with a population over 250,000 or its suburb).

The respondent's education, mother g education, and father's education was measured using a series of questions to establish the highest grade of school they each completed. The variables were coded as less than high school, high school graduate, some college, and college graduate.

There were no missing data for the variables year, age, birth cohort, or race/ethnicity. Seventeen cases were missing for size of hometown at age 16 and 25 cases were missing for education. These cases were dropped from the multivariate analyses. In addition, 2,012 cases had data for mother's, but not father's, education and 474 cases had data for father's, but not mother's, education. In such cases, values were substituted using the mean-value imputation method (Kalton, 1983). Data on both mother's education and father's education were missing in 709 cases. I created a dummy variable to represent these missing cases (missing = 1). Table 1 shows the distribution of the social-demographic variables in the sample.

Table 1: Social-Demographic Characteristics of the GSS Sample, Weighted

Characteristic                 Percent

Gender
  Male                           45.8
  Female                         54.2
Age
  18-29                          28.7
  30-39                          29.1
  40-49                          25.9
  50-59                          16.2
Birth Cohort
  1930s                           8.1
  1940s                          22.1
  1950s                          28.5
  1960s                          28.5
  1970s                          13.0
Race/Ethnicity
  White, non-Hispanic            78.4
  Black, non-Hispanic            12.3
  Hispanic                        5.5
  Other                           3.8
Residence at age 16
  Rural                          23.5
  Small or medium city           47.9
  Large city/suburb              28.6
Education
  Less than high school          15.0
  High school graduate           31.0
  Some college                   28.4
  College graduate               25.6
Mother's education
  Less than high school          29.4
  High school graduate           40.2
  Some college                   13.6
  College graduate               10.7
  Missing                         6.0
Father's education
  Less than high school          41.4
  High school graduate           24.8
  Some college                   12.8
  College graduate               15.1
  Missing                         6.0

Note. N = 11,355.

Data Analysis

To illustrate the trends over time in responses to the sex-partners questions, the data were analyzed using cross-tabulations between sex partners (during the past year and during the past 5 years) and interview year. Cross-tabulations were conducted separately for men and women. In order to determine the statistical significance of the trends shown in the cross-tabulations, the data were analyzed using maximum likelihood logit analysis. Logit analysis allows the researcher to analyze data with a dichotomous dependent variable and both categorical and continuous predictor variables (Hanushek & Jackson, 1977). I recoded the two sex-partner variables (during the past year and during the past 5 years) to create two sets of four dichotomous dependent variables: (a) whether any of the respondent's partners in the past year/5 years were of the same gender as the respondent (yes = 1/no = 0), (b) whether the respondent's partners in the past year/5 years were exclusively of the opposite gender (yes = 1/no = 0), (c) whether the respondent reported no partners during the past year/5 years (yes = 1/no = 0), and (d) whether the respondent did not answer the question (yes = 1/no = 0). The predictor variable was year of interview.

For the sample to represent individuals, rather than households (which would cause underrepresentation of individuals in large households), I analyzed the data using the sample weights provided by the GSS (see Davis & Smith, 1992). The percentages of respondents reporting same-gender partners are somewhat lower when weights are used (reflecting the tendency for people with same-gender partners to live in smaller households than other people do), but the trend is essentially the same. In the following tables all statistical estimates are weighted, whereas the unweighted frequencies are reported.

Next, bivariate relationships between same-gender sex partnering and the social-demographic variables were examined with cross-tabulations and, for the variables measured at the ordinal level, Spearman's coefficient of rank-order correlation ([r.sub.s]). Finally, multivariate logit analysis was used to determine whether the trend in same-gender partnering remained when the social-demographic variables were controlled in the model. The control variables (except for race/ethnicity, which is measured on the nominal level) were entered alternately as sets of dummy variables (e.g., age 18-29, 30-39, and 40-49, with age 50-59 as the excluded category) and as continuous variables with exponential terms (e.g., age, age(2)) to allow the relationship to be curvilinear. The best-fitting models are shown in Tables 4 and 5. Relationships are considered statistically significant if their probability of occurring by chance was estimated to be less than .05.

RESULTS

The results indicate a positive and statistically significant (p [is less than] .001) relationship between year and likelihood of same-gender sex partners in the previous year for both men and women (see Table 2). There was no trend in the tendency for either men or women to report exclusively opposite-gender partners during the previous year. However, when the data for men and women were pooled (not shown), there was a decline in the likelihood that respondents reported exclusively opposite-gender partners in the previous year (p = .028). Finally, there was no change in the likelihood of reporting no partners in the previous year or in not answering the question.

[TABULAR DATA 2 NOT REPRODUCIBLE IN ASCII]

There was a positive relationship between year and likelihood of same-gender sex partners during the previous 5 years for women, but not for men. In addition, the likelihood that women reported exclusively opposite-gender partners during the previous 5 years declined over the 1988-1998 period.

Table 3 presents the results of the cross-tabulations and correlations for same-gender partnering (during the previous year and during the previous 5 years) and each of the social-demographic variables for men and women. There is no significant association between age and same-gender partnering among men. However, men born in more recent decades are more likely than men born in earlier decades to report having had a same-gender partner during the past year. The relationship between age and same-gender sex partnering is negative for women--the older the woman, the less likely she was to have had a same-gender partner during either the previous year or the previous 5 years. The relationship between birth cohort and same-gender partnering among women is positive. More recent birth cohorts, especially respondents born in the 1970s, are more likely than earlier birth cohorts to have had a same-gender partner during either the previous year or the previous 5 years.

[TABULAR DATA 3 NOT REPRODUCIBLE IN ASCII]

The larger the population of their hometown at age 16, the more likely were men to report having had at least one same-gender partner during the previous year and the previous 5 years. The association between size of hometown at age 16 and same-gender partnering was also positive for women, but less pronounced and did not reach statistical significance for the past year. There is no association between sex partnering and race/ethnicity for either men or women.

There is no correlation between same-gender partnering and either the respondent's education or the respondent's father's education. Among women, however, there is a positive correlation between mother's education and same-gender partnering during the previous year and the previous 5 years.

Women's educational attainment, the proportion of Americans living in metropolitan areas, and the proportion of the population between ages 45 and 64 increased during the 1990s (U.S. Bureau of the Census, 1998). Consequently, the relationships found between same-gender partnering and size of hometown, mother's education, age, and birth cohort may explain at least some of the increase in the likelihood of same-gender partnering over the 1988-1998 period. To explore this possibility, the effect of year was examined when the social-demographic variables were controlled in a multivariate logit analysis. Education and father's education were excluded as control variables because they were not associated with same-gender partnering in the bivariate analyses and they are correlated with one another and with mother's education, creating a potential problem of multicollinearity. Birth cohort was also excluded; analyses indicated that because age and birth cohort are so highly correlated, a larger sample is necessary to tease apart the effects of the two variables.

The results of the multivariate analysis indicate that for both men and women, the effect of year on same-gender partnering during the previous year loses little or none of its magnitude when the control variables are introduced (Table 4). The effect of year continues to be statistically significant at the .001 level. Similarly, the effect of year on same-gender partnering during the previous 5 years declines very slightly and remains statistically significant at the .05 level for women (Table 5).

[TABULAR DATA 4-5 NOT REPRODUCIBLE IN ASCII]

The effects of the social-demographic variables in the multivariate analyses indicate that (other factors being equal) the likelihood that men reported a same-gender partner in either the previous year or previous 5 years increases with age until the mid-30s and then declines. For women, the likelihood of reporting a same gender partner is similar among women ages 18 to 49, and lower among women age 50 to 59. Black men were more likely than White men (the excluded category in the analysis) to report having had a same-gender partner during the previous year, but the relationship is not statistically significant for same-gender partnering during the previous 5 years. In contrast to the findings of Billy et al. (1993), there is no indication that Hispanic men were more likely than non-Hispanic men to report having had a same-gender partner. Women whose mothers had less than a high school education were the least likely to report a same-gender partner. Women who did not report the education of either parent were more likely than women who reported this information to have had a same-gender partner during the previous year and the previous 5 years. It is not clear why this would be the case. Also contrary to expectations, men whose mothers completed high school were less likely to report a same-gender partner during the previous year than were men whose mothers had not completed high school.

DISCUSSION

The likelihood that men and women reported a same-gender sex partner in the previous year increased over the 1988-1998 period. There was a slight decline in exclusively opposite-gender partnering, detectable only when data for men and women were pooled. There was no apparent change in sexual abstinence or in the willingness of respondents to answer questions about sexual behavior.

When looking at sexual partnering during the previous 5 years, trends were found for women over the 1988-1998 period, but not for men. Women were increasingly likely over time to report having had a same-gender partner during the previous 5 years, and they were increasingly less likely over time to report having had exclusively opposite-gender sex partners. The finding of a changing likelihood that women, but not men, reported same-gender and exclusively opposite-gender partners in the previous 5 years is consistent with the argument that greater gender equity in earnings capacity increases the tendency for women, but not men, to choose same-gender sex partners. But the absence of a pattern among men may also be a consequence of the greater difficulty for people to accurately recall behaviors over a 5-year period than over a one-year period, thus reducing the reliability of the 5-year measure. The longer the reference period, the more likely people are to include events that occurred outside the period or to omit events that occurred within the period (Sudman, Bradburn, & Schwarz, 1996). This may be more problematic for men than for women. A study by Brown and Sinclair (1999) found that methods of estimating number of sex partners in the previous year are similar for men and women. Men, however, were more likely than women to make rough approximations of the number of sex partners in their lifetime, whereas women were more likely than men to obtain the total by counting names. It follows that men may be more likely than women to guess when recalling their sex partners from the previous 5 years, leading to more error-prone counts.

Some respondents may have failed to report a same-gender sex partner because same-gender sexual behavior is stigmatized in contemporary American society. This could have occurred despite the efforts made by the GSS to encourage honest answers--specifically, the self-administered instrument, the assurances of confidentiality, and the emphasis on the importance of gathering accurate information on sexual behavior. Thus, the actual incidence of same-gender sexual behavior may be higher than that reported here. On the other hand, errors in recording respondents' sex, which are estimated to occur in 1.97% of the cases, would lead to an overestimate of the proportion of the population that has had a same-gender partner (Black, Gates, Sanders, & Taylor, 2000). A crucial question for this study is whether reporting and recording errors can account for some or all of the increase in reports of same-gender partners over the 1988-1998 period. There is no reason to think that errors in recording respondents' sex would have increased over time. However, attitudes toward homosexuality became more tolerant during the 1988-1998 period and, consequently, respondents in the more recent waves of the GSS may have felt more comfortable admitting to having had a sex partner of their own gender. Note, however, that the tendency not to answer the questions did not decline over time and, therefore, would not appear to account for the rising reports of same-gender sex partnering.

These estimates of same-gender sex partnering should not be taken as estimates of the proportion of the population that is gay or lesbian. Some people may engage in same-gender sexual activity and yet identify as heterosexual, whereas other people may identify as gay or lesbian but may not have been sexually active in recent years. These estimates are comparable to those of other recent large-scale studies on same-gender sexual behavior conducted in the United States (Billy et al., 1993; Laumann et al., 1994; Rogers & Turner, 1991), Great Britain (Wellings, Field, Johnson, & Wadsworth 1994), and France (Spira, 1993, cited in Laumann et al., 1994).

The associations between same-gender sex partnering and social-demographic factors are consistent with previous research findings in some cases but not others. In contrast to the findings of research conducted by Billy et al. (1993), this study found no effect of Hispanic ancestry on the likelihood of same-gender sex partnering but found a greater likelihood of same-gender partnering among African-American men than among White men. This racial difference might be due to the greater incarceration rates of African-American men than of White men (U.S. Bureau of the Census, 1998). The absence of female sex partners in jails and prisons creates conditions conducive to same-gender sexual activity among men who might otherwise choose women as sex partners. But without information on whether sample members had been recently incarcerated, the validity of this explanation cannot be determined. Consistent with the findings by Laumann et al. (1994), the current study found that men who lived in a metropolitan area at age 16 were more likely to report having had a same-gender partner. In contrast to Rogers and Turner (1991), the current study did not find a positive association between same-gender sex partnering and father's education among men. Women, however, were least likely to report a same-gender partner if their mother had not graduated from high school.

The patterns found in the data suggest that age and birth cohort may have independent effects on the likelihood of same-gender sex partnering. The likelihood that men had had a recent male sex partner increased with age until their thirties and then decreased, whereas the likelihood that women had had a recent female sex partner was similar for women ages 18 to 49 and lower for women in their fifties. In contrast, more recent generations of men and women were more likely to report a same-gender partner than were earlier generations. Adding additional waves of the General Social Survey in future years should show whether the generation effects found here continue as these birth cohorts age. We may find that age effects reflect the tendency for sexual activity to vary at various stages of life, while throughout their lives some generations may be more likely than other generations to have same-gender sex partners because of their socialization as children and young adults.

Interpretation of associations between same-gender sexual behavior and social-demographic factors must be made with caution because the stigma of homosexuality may affect truthfulness in reporting same-gender sexual behavior. Homophobia has been found to be more prevalent among people living in small cities and towns, among less educated people, and among older people (Herek, 1984). Living in a hostile environment could lead those who have same-gender sexual interests to be more circumspect in disclosing information about their sexuality to interviewers. Homophobia appears to be higher among African-Americans than among Whites, however (Battle & Bennett, 2000; Smith & Seltzer, 1992), and so cannot account for the higher rates of reported same-gender sex partnering among African-American men.

The increase in the likelihood that Americans had had a recent same-gender sexual relationship during the 1988 to 1998 period cannot be accounted for by the changing age structure, increasing urbanization, increasing educational attainment, or the changing racial and ethnic (i.e., Hispanic) mix in the population. An alternative explanation for the trend was suggested at the beginning of the paper. Declining social, legal, and economic sanctions against same-gender sexual behavior in recent years and more positive images of gay men and lesbians in the media may have made it easier for people to recognize their same-gender sexual interest and to act on it. Testing this hypothesis directly, however, is a challenge. Consider the possibilities if the researcher had detailed location data for respondents and could code each case for whether the respondent's employer offered domestic partnership benefits, the attitudes in the community toward same-gender sexual activity, and the presence of a human rights ordinance that prohibited discrimination based on sexual orientation. Despite this wealth of data, the meaning of a negative association between the degree of sanctions and same-gender sex partnering would be ambiguous. On the one hand, the relaxation of social mores and legal restrictions might have enabled people to explore their sexual attraction to others of their gender, as hypothesized here. Yet people who are sexually attracted to others of their gender are likely to seek out jobs in firms with domestic partnership benefits and to move to areas of the country that have liberal attitudes and legal protections for gay men and lesbians. In other words, the relationship between sanctions and sex partnering is probably reciprocal and therefore difficult to sort out empirically.

Although doubling or tripling between 1988 and 1998, the absolute increase in the likelihood of having had a same-gender sex partner is only several percentage points. The proportion of men and women who reported having had a sex partner of their own gender during the previous year remained low in 1998--4.1% for men and 2.8% for women. But while social and legal constraints regarding same-gender sexual behavior are loosening, they are still strong. Children are ridiculed by their peers for displaying behavior that does not conform to gender stereotypes (Thorne, 1993), more than half of the American public continues to think that sexual relations between two adults of the same sex is "always wrong," and gay men and lesbians continue to face discrimination in many parts of the country and to be victims of hate crimes (People for tile American Way Foundation, 1993-99). Women continue to earn less than their male counterparts, making female coupling financially disadvantageous for many women. However, if the recent trend continues toward equal treatment of people regardless of their gender or sexual orientation, we may see additional increases in the likelihood of same-gender sex partnering.

(1) A randomly selected group of respondents in 1988 did not get the introductory sentences that referred to AIDS. Instead, they received the following statement: "Now we are going to ask some additional questions. Your answers are confidential and will be used only for statistical reports." Respondents who received the AIDS introduction were less likely than respondents who received the shorter introduction to report a same-gender sex partner, but the difference was not statistically significant (Smith, 1992).

(2) Eleven respondents who reported same-gender partners during the previous 12 months and who did not answer the question about sex partners in the past 5 years were recoded as having had a same-gender partner in the past 5 years. Six respondents who reported same-gender partners during the previous 12 months, but exclusively opposite-gender partners during the previous 5 years, were recoded as having had a same-gender partner in the past 5 years. The reassignment of these 17 respondents did not affect the conclusions drawn from the results.

REFERENCES

American Civil Liberties Union. (1999, January). Status of U.S. sodomy laws. New York: Author. Retrieved December 4, 1999, from the World Wide Web: http://www.aclu.org/issues/gay/sodomy.html.

Battle, J., & Bennett, M. (2000). Research on lesbian and gay populations within the African American community: What have we learned? African American Research Perspectives, 6 (2): 35-47.

Bianchi, S. (1995). Changing economic roles of women and men. In R. Farley (Ed.), State of the union: America in the 1990s, (Vol. 1, pp. 107-154). New York: Russell Sage.

Billy, J. O. G., Tanfer, K., Grady, W. R., & Klepenger, D. H. (1993). The sexual behavior of men in the United States. Family Planning Perspectives, 25, 52-60.

Black, D., Gates, G., Sanders, S., & Taylor, L. (2000). Demographics of the gay and lesbian population in the United States: Evidence from available systematic data sources. Demography, 37, 129-142.

Brown, N. R., & Sinclair, R. C. (1999). Estimating number of lifetime sex partners: Men and women do it differently. The Journal of Sex Research, 36, 292-297.

Button, J. W., Rienzo, B. A., & Wald, K. D. (1997). Private lives, public conflicts: Battles over gay rights in American communities. Washington, DC: Congressional Quarterly.

Carrier, J. M. (1985). Mexican male bisexuality. In F. Klein & T. Wolf (Eds.), Bisexualities: Theory and research (pp. 75-86). New York: Haworth Press.

Davis, J. A., & Smith, T W. (1992). The NORC General Social Survey: A user's guide. Newbury Park, CA: Sage.

DeLamater, J. D., & Hyde, J. S. (1998). Essentialism vs. social constructionism in the study of human sexuality. The Journal of Sex Research, 35, 10-18.

Hanushek, E. A., & Jackson, J. E. (1977). Statistical methods for social scientists. New York: Academic Press.

Herek, G. M. (1984). Beyond "homophobia": A social psychological perspective on attitudes toward lesbians and gay men. Journal of Homosexuality, 10, 1-21.

Human Rights Campaign. (1999, March 5). States, cities and counties that prohibit discrimination based on sexual orientation in private employment. Washington DC: Author. Retrieved December 4, 1999, from the World Wide Web: http//www.hrc.com/issues/workplac/nd/ndyuris.html.

Human Rights Campaign. (n.d.). Domestic partnership benefits for same-sex couples. Washington DC: Author. Retrieved December 4, 1999, from the World Wide Web: http//www.hrc.com/issues/workplac/dp/index.html.

Kaiser, C. (1997). The gay metropolis: 1940-1996. Boston: Houghton Mifflin.

Kalton, G. (1983). Compensating for missing survey data. Ann Arbor, MI: Institute for Social Research, University of Michigan.

Kinsey, A. C., Pomeroy, W. B., & Martin, C. E. (1948). Sexual behavior in the human male. Philadelphia: W.B. Saunders.

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

Magana, J. R., & Carrier, J. M. (1991). Mexican and Mexican American male sexual behavior and spread of AIDS in California. The Journal of Sex Research, 28, 425-441.

Parish, J. R. (1993). Gays and lesbians in mainstream cinema. Jefferson, NC: McFarland.

Paz, O. (1950). The labyrinth of solitude: Life and thought in Mexico. New York: Grove Press.

People for the American Way Foundation (annual, 1993-1999). Hostile climate: A state by state report on anti-gay activity. Washington DC: Author.

Rogers, S. M., & Turner, C. F. (1991). Male-male sexual contact in the U.S.A.: Findings from five sample surveys, 1970-1990. The Journal of Sex Research, 28, 491-519.

Smith, R. C., & Seltzer, R. (1992). Race, class, and culture: A study in Afro-American mass opinion. Albany, NY: State University of New York Press.

Smith, T. W. (1992). A methodological analysis of the sexual behavior questions on the General Social Surveys. Journal of Official Statistics, 8, 309-325.

Stacey, J. (1998). Gay and lesbian families: Queer like us. In M. A. Mason, A. Skolnick, & S. D. Sugarman (Eds.), All our families: New policies for a new centre? (pp. 117-143). New York: Oxford University Press.

Sudman, S., Bradburn, N. M., & Schwarz, N. (1996). Thinking about answers: The application of cognitive processes to survey methodology. San Francisco: Jossey-Bass.

Thorne, B. (1993). Gender play: Girls and boys in school. New Brunswick, NJ: Rutgers University Press.

U.S. Bureau of the Census (1998). Statistical abstract of the United States, 1998. Washington DC: Government Printing Office.

Wellings, K., Field, J., Johnson, A., & Wadsworth, J. (1994). Sexual behavior in Britain: Ire National Survey of Sexual Attitudes and Lifestyles. New York: Penguin.

Wetzel, J. R. (1995). Labor force, unemployment, and earnings. In R. Farley (Ed.), State of the union: America in the 1990s, (Vol. 1, pp. 59-105). New York: Russell Sage.

Manuscript accepted May 1, 2000

Address correspondence to Amy Butler, 308 NH, School of Social Work, University of Iowa, Iowa City, IA 52242; e-mail: [e-mail].