Women’s Sexual Orientation and Sexual Behavior: How Well Do They Match?

If you know which sexual orientation a woman identifies with, how much does that tell you about whether she has sex with men, women, or both?  Here we answer this question for American women, using data from the 2006-2008 National Survey of Family Growth for women 15 to 44 years of age.

Detailed tables, along with how we generated our measures, are in the Technical Appendix at the end of this post. Here we zero in on a few specific questions of interest:

How Common Is It for Heterosexual Women to Have Sex with Women?

How common is it for women who say they are heterosexual to have sex with women? If we operationalize having had sex with a woman as having had oral sex with a woman, the answer is that 6% have ever done so. All of these have had sex with men as well (Table 1).

A second approach uses the question asking people how many male sexual partners they have had intercourse with, and how many female sexual partners they have had. One disadvantage of his approach is that we don’t know what individual respondents think “counts” as having had sex with another woman. Nonetheless, on the question of how many heterosexual women have had sex with women, this strategy gives us a similar answer: Almost none of the women identifying as heterosexual have had only female partners (0.3%), and 9% say they have had at least one female sexual partner (Table 2).

The fact that 6-9% (depending on the measure) of women who identify as heterosexual have had sex with a woman doesn’t necessarily mean that women’s identity and behavior are inconsistent at any single point of time.  It is possible these women identified as lesbian or bisexual at the time they had sex with women. Or it is possible that women engage in behavior inconsistent with their orientation. We would need panel data following the same people over time and repeatedly asking about orientation and recent behavior to distinguish (a) changing sexual orientation where behavior and identity are always consistent from (b) inconsistency between current identity and current behavior.

In the NSFG data, we can get a little closer to assessing how common inconsistency between identity and behavior is by comparing women’s current identity with whether they’ve had men and women as sexual partners in the last year.  Here we get a much smaller figure—only 2% of women who called themselves heterosexual on the survey have had sex with a woman in the last year (Table 3).  This suggests that, at least for women who identify as heterosexual, behavior usually aligns fairly well with identity in any short (one year) time range.

These results for heterosexual women—how many have ever or in the last year had sex with a woman—are  shown in the graph below:

 

How Common Is It for Lesbians to Have Sex with Men?

As the graph below (drawing from Tables 1 and 2) shows, depending on the measure used, between two-thirds and four-fifths of lesbians have had sex with a man sometime in their lives.  Eighty one percent report having had either oral sex, vaginal intercourse, or anal sex with a man, while 67% report having had a male intercourse partner sometime in their life. By either measure, the proportion of lesbians who have ever had sex with a man is drastically larger that the proportion of heterosexual women who have ever had sex with a woman.

However, if, we restrict our focus to the year before the survey, we get a very different picture.  Only 22% of women who identify as lesbian have had sex with a man last year. If these are all women whose behavior is inconsistent with their identity, then it seems a sizable share—over a fifth; it is very different than the under 2% of heterosexual women who had sex with a woman in the last year.  However, it is also possible that some sizable share of the 22% may be cases where women changed their identity and behavior in the last year, but identity was consistent with behavior at most all times.  The data don’t allow us to tell which it is.

How Does the Behavior of Bisexual Women Differ From That of Lesbians and Heterosexual Women?

Given that the common-sense meaning of the term “bisexual” is an interest in having sex and romantic relationships with either men or women, we would expect more bisexual than heterosexual or lesbian women to have had sex with both men and women.  Indeed, depending on whether we use the measure in Table 1 or 2, we find that 68%-75% of bisexual women have had sex with men and women, higher than the analogous figures for lesbians of 67% and 55%, and drastically higher than the proportion of heterosexual women of 6%-8%.  In this regard, bisexual women look more like lesbians than like heterosexual women.

Also of interest is how many women in each group have only had sex with men, or only with women. The percent who have had sex only with men is 78-84% for heterosexuals, 12-14% for lesbians and 14-24% for bisexuals (Tables 1 and 2); in this regard also, bisexual women are more similar to lesbians than heterosexuals.

What about who has had sex only with women? Less than 1% of the heterosexual women reported having had sex with only women, compared to 16-31% for lesbians and 2-5% for bisexuals; in this regard bisexual women are more similar to heterosexuals than to lesbians—almost none have had sex only with women.  (See Tables 1 and 2.)

Who have bisexual women had sex with in the last year?  As the graph below shows, drawing from Table 3, the percent that have had sex with both men and women in the last year is 23% for bisexual women, 10% for lesbians, and only 2% for heterosexuals.  Bisexuals are much more likely than lesbians to have had sex only with men in the last year—57% compared to 12% for lesbians.  But, not surprisingly, this figure is highest for heterosexuals—78%.  The flip side of this is that bisexuals are much less likely than lesbians to have had only women partners in the previous year—9% for bisexuals compared to 75% of lesbians (and less than 1% heterosexuals).

Conclusion:  Fluidity, Links Between Behavior and Identity, and the Need for Panel Data

We have shown that identity—the sexual orientation one identifies with—is strongly, but by no means perfectly, correlated with whether women have had sex with men, women, or both. Heterosexual women are the most likely to have had sex only with men, bisexual women are the most likely to have had sex with men and women, especially if we look only at the last year, and lesbians are the most likely to have had sex only with women, especially when we limit the scope to the last year.

If we make the assumption—undoubtedly unwarranted for some—that the women surveyed have had the sexual orientation they report all their lives, then there is substantial evidence of behavior inconsistent with identity. For example, a strong majority of lesbians have had sex with a man, 24% of bisexuals have had sex with a man but never had oral sex with a woman, and 9% of heterosexual women say they have had a female sexual partner.  If sexual orientation doesn’t change over the life course, all of these are evidence of inconsistency, with heterosexuals being the least likely to show inconsistency.

However, both behavior and sexual orientation can change over time, and if they do, then the patterns just discussed don’t necessarily indicate inconsistency between orientation and behavior. Consider the hypothetical example of a woman who grew up assuming she was heterosexual—since virtually all the songs, stories, and movies suggested this was normal.  Following custom, she dated and had sex with men, but later noticed she was attracted to women, which led to sex with women, and eventually to a stable lesbian identity.  The data we’ve shown here, wherein most lesbians have had sex with men sometime, but the vast majority (about 80%) did not have sex with a man in the last year, would be consistent with this hypothetical woman’s sequence, which would have entailed inconsistency only around the time of change. But the data are also consistent with some amount of recurrent mismatch between identity and behavior.

To distinguish mismatch between behavior and identity from changes in either sexual orientation or behavior, we need panel data that survey the same group of people repeatedly over time asking about both behavior and orientation near the time of the survey.  Unfortunately, none of our national probability-sample panel data sets ask extensive questions about sexuality, and none of the studies, like the NSFG, that ask extensive questions on non-heterosexual behavior and identities also survey the same respondents repeatedly over time. We know of only one panel study, by Lisa Diamond (2008, 2009), following women for 10 years and examining their change in identities and sexual behavior. Unfortunately, it did not include women who self-defined as heterosexual at the outset, and it is a relatively small sample (no more than 100).

---

Technical Appendix

Our analysis uses data from the 2006-2008 National Survey of Family Growth on women 15 to 44 years of age.

To explore the relationship between sexual orientation and behavior, we began with the measure of sexual orientation in the survey.  Women were asked whether they “think of themselves as” “heterosexual or straight” (which we’ll call “heterosexual” here), “bisexual,” or “homosexual or lesbian” (which we’ll call “lesbian” here).  In some of the years of the survey that asked the sexual behavior questions below, respondents were also given the option to call their sexual orientation “something else.”  This option was only given in 2006-2008, then dropped because so few respondents chose it. Since the small proportion of women choosing “something else” might be different with respect to the link between their behavior and identity, we chose to use only the years where the option was given, and to put these women in a separate sexual orientation category, so that the sexual orientation categories available would be comparable for all these years.

We generated three different measures of sexual behavior from various questions in the survey.

1. Have you ever done specific sexual behaviors with a man/woman? For this measure, used in Table 1 below, having ever had sex with a woman is defined as reporting that you have ever had oral sex with one or more women, and having ever had sex with a man is defined as reporting that you have had vaginal, anal or oral sex with one or more males. Then we combined this information to create a variable indicating whether women had had neither oral sex with a woman nor oral, vaginal, or anal sex with a man; oral sex with a woman but none of the three types of sex with a man; one of the three types of sex with a man but never oral sex with a woman; or oral sex with a woman as well as one of the three types of sex with a man.
2. Have you ever had a male/female sexual partner? For this measure, we relied on two questions. One question asked women the number of female sexual partners they have had during their lifetime, but did not specify what sexual behavior was meant for “sexual partners.”  If women said they had had one or more, we considered them to have had a female partner for this measure. Regarding sex with men, women were asked how many men they had ever had vaginal intercourse with; if this number was 1 or more we considered them to have had a male partner during their lifetime. These items were used to create a variable indicating whether each woman had had no sexual partners of either sex, sex only with one or more women, sex only with one or more men, or sex with one or more men and one or more women.
3. Did you have a male/female sexual partner last year? This is constructed just like 2) above, but using questions about the number of male and female sexual partners one has had in the last 12 months.

One might worry that respondents would not be honest about same-sex sex or non-heterosexual identities, given the social bias against them. To try to avoid respondents saying what they thought the interviewer wanted to hear rather than the truth, the questions on sexual identity and sex with same-sex partners were asked through an Audio Computer-Assisted Self-Interview (ACASI) system. The interviewer handed the respondent a computer and earphones and stepped away to provide privacy while the respondent keyed answers into the computer. This ACASI approach was used for questions on sexual orientation, as well as the questions on number of same-sex partners, and what specific sexual behaviors respondents had done with a man and with a woman. However, the question about number of male vaginal intercourse partners women have had was taken from the main questionnaire, not the ACASI part.

Note:  Women are considered to have had sex with a woman if they report ever having had oral sex with a woman. They are considered to have had sex with a man if they report ever having had oral sex or vaginal or anal intercourse with a man.  N= 5851.

Note:  Women are considered to have had sex with a woman if they report ever having had a female sexual partner (regardless of what behavior they report they have done with a woman).  They are considered to have had sex with a man if they report ever having had vaginal intercourse with a man. N=5851.

Note:  Women are considered to have had sex with a woman if they report having had a female sexual partner in the last year.  They are considered to have had sex with a man if they report having had vaginal intercourse with a man in the last year. N=5851.

---

References:

Diamond, L. M. (2009). Sexual fluidity. John Wiley & Sons, Ltd.

Diamond, L. M. (2008). “Female bisexuality from adolescence to adulthood: results from a 10-year longitudinal study.” Developmental psychology, 44(1), 5.