The Rise of Bisexuality: U.S. Representative Data Show an Increase Over Time in Bisexual Identity and Persons Reporting Sex with Both Women and Men

ABSTRACT

Recognizing that social forces shape sexual behavior and sexual identities, we examined whether the recent liberalization in attitudes toward same-sex relationships has been accompanied by changes in sexual orientation as indicated by sexual behavior and sexual orientation identity. Using nationally representative data from the General Social Survey (n = 34,524), we compared reported patterns of sexual behavior of respondents from 1989 through 2021. Among the 2012–2018 waves (n = 7144, weighted), we compared patterns of sexual orientation, as indicated by behavior and identity, among different demographic groups. We also examined how sexual behavior varied among respondents with different sexual orientations. Results showed an increase in respondents reporting both male and female partners over time, with 3.1% of respondents reporting both male and female partners since age 18 among the 1989–1994 waves, 9.3% reporting this in the 2012–2018 waves, and 9.6% reporting this in the 2021 sample. Identifying as bisexual was more common among women (3.7%) than men (1.6%) and among younger persons than older persons. Over 6% of 18–29 year-olds but fewer than 2% of respondents over age 40 identified as bisexual. Findings demonstrate that sexual norms and behaviors have changed and that far more persons today than in earlier years identify as bisexual and/or have both male and female partners.

Opinion surveys in the U.S. indicate substantial increases in the acceptance of same-sex sexuality in general and same-sex marriage, particularly since 2010 (Adamczyk & Liao, 2019; Steffens & Preuß, 2020) and to a lesser extent over longer time periods (Monto & Supinski, 2014; Sullivan, 2004). Research on same-sex sexuality has also flourished, providing greater insight into the experiences and behaviors of sexual minority communities. There has been less attention directed toward bisexuality, and there are still significant gaps in our understanding of the trends and sexual behavior associated with this sexual orientation (Bowes-Catton & Hayfield, 2015; Kaestle & Ivory, 2012). Existing research on bisexuality relies primarily on convenience samples and often fails to compare changes over time. The present study addressed these gaps, using nationally representative samples of adults in the United States to describe trends in bisexuality, same-sex sexuality, and heterosexuality over time and identify patterns in bisexuality, in contrast to heterosexuality and same-sex sexuality, among recent cohorts.

Hertlein et al. (2016) argued that bisexuality is often misunderstood, and that bisexual persons remain less accepted than gay and lesbian individuals, both among heterosexual persons and among other sexual minorities. More fully considering bisexuality as a concept and as an identity can contribute to better understandings of the ways society and culture shape human sexuality (Callis, 2009). Defining bisexuality has been described as a “herculean task” (Swan, 2018), and definitions of the term have been inconsistent (Flanders et al., 2017). Using frameworks borrowed from other research (Jeffries, 2011; Kaestle, 2019; Mendelsohn et al., 2022), we define bisexuality as a sexual orientation that can be indicated by sexual attraction to persons of both sexes, sexual behavior involving both sexes, and/or a personal identity as bisexual. Existing research reveals consistently reveals close correlations between sexual attraction, sexual orientation identity, and sexual behavior (Chandra et al., 2011; Savin-Williams, 2014). However, people are more likely to report sexual behavior or attraction involving either the same sex or both sexes than they are to identify as lesbian, gay, or bisexual (Kaestle, 2019; Mishel, 2019). We also note that patterns and findings vary depending on which indicators are used (Jeffries, 2011). In this study, we employed two of these indicators of sexual orientation: identity and behavior. Using data gathered through the nationally representative General Social Survey (Smith et al., 1972–2021), we identified trends in sexual behavior since 1989, identified trends in sexual orientation identity since 2008, described demographic patterns in sexual orientation among recent cohorts, and compared bisexual persons to heterosexual persons and gay and lesbian persons across six sexual and relationship variables.

Bisexuality as a Sexual Orientation

Although sex researchers have acknowledged bisexuality as a sexual orientation since the time of Kinsey (Dodge et al., 2008), it is still often grouped with other minoritized sexualities in research and cultural dialogue (Kaestle & Ivory, 2012; Volpp, 2010; Worthen, 2013). Though people who identify as bisexual may share experiences of marginalization in common with lesbian and gay identified persons, recent findings suggest that they are less likely to be “out” to close family and friends (Pew, 2019) and tend to experience their sexual orientation as less important to their sense of self (Pew, 2019). Bisexual persons in relationships with partners of a different sex may escape some of the negative social judgments experienced by persons in gay or lesbian relationships. However, people who identify as bisexual also report significant “biphobia” or “binegativity” (Pollitt & Roberts, 2021). Bisexuality is sometimes treated as not a “real” sexual orientation, a phase or attention seeking strategy for women who are really heterosexual, or a transitory behavior for men who are gay and in denial (Alarie & Gaudet, 2013). Scholars have suggested that these beliefs contribute to making bisexuality invisible or the “forgotten sexuality” (Kaestle & Ivory, 2012), despite recent data on the rising proportion of persons identifying as bisexual (Jones, 2021) contributing to increased visibility. Bisexual people report experiencing stigma and erasure from both heterosexual and gay and lesbian people, an experience sometimes referred to as “monosexism,” i.e., the assumption that people naturally have sexual desires for just one sex (Chan & Leung, 2022; Roberts et al., 2015).

Polling on Same-Sex Sexuality and Bisexuality

Public opinion about same-sex marriage represents an indicator of the degree to which people accept sexual orientations besides heterosexuality. For example, the Pew Research Center reported that the proportion of Americans who oppose same-sex marriage declined from 60% in 2004 to 31% in 2019 (Pew, 2019). A recent Gallup poll captured a substantial increase in the proportion of persons identifying as lesbian, gay, bisexual, or transgender (LGBT), with 7.1% of respondents identifying as LGBT in 2022, twice the percentage reporting those identities in 2012 and an increase in comparison to the 5.6% reporting those identities in 2020 (Jones, 2022). While previous Gallup polls did not measure the proportion of persons identifying as bisexual, the 2022 report showed 56.8% of LGBT adults (about 4.0% of all U.S. adults) identified as bisexual. Another brief analysis, using the General Social Survey (Bridges & Moore, 2018), also showed that by 2016, identifying as bisexual, reported by 3% of respondents, had become more common than identifying as lesbian or gay, reported by 2.4% of respondents. However, when looking only at men, identifying as bisexual was less common than identifying as gay. Further, a 2016 study utilizing the National Survey of Family Growth (England et al., 2016) showed increasing rates of women identifying as bisexual and reporting sexual partners of both sexes in more recent cohorts. Interestingly, no corresponding change in these variables was found among men. Our study builds on these findings by using several behavioral indicators of sexual orientation in addition to sexual orientation identity, by capturing trends over a much longer period of time, by more thoroughly describing variations in bisexual, heterosexual, and same-sex sexual orientation by age, race, sex, marital status, and education level, and by examining how sexual orientation is associated with sexual behavior.

Sexual Orientation and the Social Environment

Human sexuality is shaped by the rules and norms surrounding sexual practices, the attitudes about what and whom is attractive or desirable, and the terms used to identify sexual identities, all of which vary from culture to culture and within particular cultures over time. Changing norms shape whether particular sexual behaviors and identities are defined as acceptable or “deviant.” Becker (1963) argued that whether individuals or actions are identified as problematic depends not on the qualities of the actions or individuals but on the boundaries delineated by those with the power to define the rules. Foucault (1978) described the profound ways that the medical “invention” of “homosexuality” shaped cultural understandings and terminology associated with same-sex sexuality, as well as the identities through which people conceive themselves or are labeled by others (Grassi, 2016). Rubin (1984) argued that contemporary society places sexuality that is monogamous, heterosexual, and procreative within a “charmed circle” that serves to “rationalize the well-being of the sexually privileged and the adversity of the sexual rabble” (p. 152). However, norms surrounding sexuality are not static and are affected by changing power dynamics and definitional contests within societies. For example, the American Psychiatric Association’s 1973 decision to remove the diagnosis of “homosexuality” from the DSM was the result of effective mobilization by gay and lesbian practitioners and their advocates and included a vote by APA members (Drescher, 2015). Interestingly, it is not same-sex sexuality but rather fear, dislike, or prejudicial beliefs toward lesbian and gay persons that are currently problematized in the growing body of academic research on homonegativity (formerly misnamed homophobia; Rollè et al., 2022). More recent changes in sentiments toward same-sex marriage have also been shaped by the successful mobilization efforts of LGBT advocates and their allies, supported by the increasing visibility of high-profile celebrities and role models (Marzullo & Herdt, 2011). Concurrent with the changes in public sentiments, the U.S. Supreme Court decision legalizing same-sex marriage (Obergefell v. Hodges, 2015) provided another institutional process that has shaped and may continue to shape social rules and practices surrounding gay, lesbian, and bisexual relationships (Hart-Brinson, 2018).

With recent research providing additional evidence of the genetic correlates of sexual orientation (Ganna et al., 2019), it is important not to lose sight of the ways that social norms and stigma shape sexual behavior and the ways we choose to identify ourselves (Diamond, 2021). The vocabularies people use to identify their sexual orientation are socially negotiated and change over time (Baker & Harris, 2020; Grassi, 2016; Zosky & Alberts, 2016). Additionally, cultural beliefs surrounding the etiology of sexual orientation have been shown to influence public acceptance of same-sex sexuality, with people who view same-sex sexuality as innate or having biological origins more supportive than people who believe it is a choice (Frias-Navarro et al., 2015; Whitehead & Baker, 2012). In sum, sexual behavior and sexual orientation identity can be recognized as the product of a complex interplay between the genetics associated with desire, the social norms and definitions provided by society, and individuals negotiating their own actions and identities in much more particular and immediate situational contexts.

The social nature of human sexuality is reflected in the patterns of sexual orientation associated with demographic variables, including race and ethnicity, gender/sex, and age, as well as achieved characteristics, such as marital status and education level. Muñoz-Laboy (2019) pointed to substantial racial and ethnic differences in understandings of sexual orientation among different ethnic groups. Han (2015) argued that the cultural rhetoric surrounding Black men who have sex with men and the media hype suggesting that bisexual Black men on the “down low” were responsible for the transmission of sexually transmitted infections to women and the general population contributed to pathologizing Black male same-sex sexuality. Muñoz-Laboy (2019) argued that research that does not acknowledge racial and ethnic differences in sexual orientation risks reinforcing “color-blind racism” (Bonilla-Silva, 2017) and further marginalizing bisexual persons who are also members of minoritized ethnic or racial groups.

Gender/Sex and Sexual Orientation

Similarly, research that categorizes people by their sexual orientation but neglects variation between women and men misses a key factor that has been repeatedly shown to shape the sexual behavior of gay, lesbian, bisexual, and heterosexual persons. Gender/sex in conjunction with sexual orientation is associated with frequency of sexual behavior, number of partners (Glick et al., 2012; Jeffries, 2011; Oswalt & Wyatt, 2013; Peplau et al., 2004), paying or being paid for sex (Boyer, 2013), commitment to monogamy (Barrios & Lundquist, 2012; Potârcă et al., 2015; Schmitt, 2005), and sociosexuality (Schmitt, 2007), the “predisposition to engage in uncommitted sex” (Rodrigues & Lopes, 2017, p. 775). Sexual orientation, often in conjunction with gender/sex, is associated with education levels (Mittleman, 2022) and the likelihood of being married (Carpenter, 2020), with the latter variable clearly affected by changes in the legal status of same-sex marriage. The complex ways in which age, race, and gender/sex interact with sexual orientation and the cultural and subcultural beliefs within these groups that shape sexual and relationship behaviors are beyond the scope of this paper, better suited to qualitative research or more detailed sex surveys. However, the present study provides an unprecedented opportunity to describe the broad patterns associated with these variables among a representative sample of U.S. adults over time and to bring bisexuality more clearly into focus. We note that the GSS has used the term “sex” to indicate whether respondents were men or women since the survey began, making it a crude indicator of both biological sex and gender. We use the term gender/sex throughout this article when discussing the differences between women and men for several reasons. First, it avoids confusing this distinction with the act of sex. Second, we recognize that “gender,” the qualities assigned by culture to these categories, is critical in shaping our identities and our sexual experiences (van Anders, 2022). Third, we wish to avoid reinforcing the impression that biology is the sole determinant of sexual expression. Thus, though we can identify broad patterns associated with gender/sex, the survey data used here do not allow us to differentiate between the effects of sex and gender.

Increases over time in the proportion of persons in the U.S. identifying as gay and lesbian, as well as the increasing acceptability of same-sex relationships, suggest a growing acceptance of non-heterosexual identities and sexual behavior, trends that may also apply to bisexual orientation identity and may make having partners of both sexes more acceptable. Hence we hypothesized:

H1: Respondents in the 2012–2018 waves of the General Social Survey will be more likely to report having partners of both sexes than respondents from any of the preceding waves.

H2: Respondents from the 2012–2018 waves of the General Social Survey will be more likely to identify as bisexual than respondents from the 2008–2010 waves.

Additionally, recognizing the importance of better understanding bisexuality and its significance in the United States, the present study aimed to draw specific attention to this sexual orientation using nationally representative survey data to address the following questions:

1. How have patterns of sexual orientation changed over time?

2. What proportion of U.S. adults in recent cohorts report having sexual partners of both sexes or identify as bisexual, and how does that compare with the proportion who identify as heterosexual or gay and lesbian?

3. How, if at all, does sexual orientation, as indicated by sex of sexual partners or by sexual orientation identity, vary among different demographic groups, including gender/sex, age, race, marital status, and education level?

4. How, if at all, does sexual behavior differ by sexual orientation, as indicated by sex of sexual partners or by sexual orientation identity?

Method

Respondents

The present study reports on 32,238 (unweighted) respondents who completed the General Social Survey (Smith et al., 1972–2021) between 1989 and 2018 and 2,313 (unweighted) additional respondents who completed the survey in 2021 and for whom data were available on their race, gender/sex, year, and sexual partners in the previous year (including those with none). Results are presented using weighted numbers (to be described shortly), which differed little from unweighted numbers. For example, the weighted n of the 2021 wave was 2297 (compared to 2,313 unweighted). The COVID-19 pandemic necessitated methodological changes in administration of the 2020–21 GSS that may affect its comparability with other waves (Davern et al., 2021). Though we report results from 2020–21, we do not statistically compare the 2021 data to other waves.

The GSS is a leading source of data on social trends over the last five decades in the United States. Conducted by the National Opinion Research Center, the GSS is a cross-sectional study that employs a sampling frame designed to yield a nationally representative sample of U.S. households. Many of the analyses described here included only respondents who completed questions on the number of male and female sexual partners they had since their 18th birthday, questions first asked in 1989. Beginning in 1994, the GSS was conducted in even years, until the pandemic affected administration in 2020–2021. We employed the GSS weighting variable COMPWT for the 1989–2018 waves and the weighting variable WTSSNRPS for the 2021 data. Both were designed to weight responses so that they yield parameter estimates that represent individuals rather than households. The COMPWT variable also addressed problems and deliberate oversampling in earlier waves of the survey. About 44.1% of respondents were identified as male, and 55.9% of respondents were identified as female. Most (79.6%) were identified as White, with 13.3% identified as Black and 7.0% identified as “other.” Newer waves of the survey also included “Asian/Pacific Islander” and “Hispanic,” which we use in describing later waves. These and subsequent percentages and other statistics reported in this article are weighted.

Measures

This study employed four measures of sexual orientation as indicated by sexual behavior and one measure of sexual orientation as indicated by identity. Beginning in 1989, the GSS began asking respondents about the number of male partners they had had sex with since age 18, as well as the number of female partners since age 18. We recoded these variables and combined them with respondents’ gender/sex to yield four categories: “no partners,” “only partners of a different sex,” “only partners of the same sex,” and “both female and male partners.” Among only those who reported having had sexual partners, two additional variables asked “have your sex partners in the last [five years/12 months] been” “exclusively male,” “both male and female,” or “exclusively female.” We used the “12 months” variable with all respondents and then separately with only respondents who reported having had 2 or more partners, because including respondents with only one partner eliminated the possibility of both male and female partners. We used these four indicators to examine trends in sexual orientation as indicated by sex of sexual partners from 1989 to the present. The GSS began including a variable asking respondents to identify their sexual orientation in 2008 that asked “which of the following best describes you?” and included three options, “gay, lesbian, or homosexual,” “bisexual,” and “heterosexual or straight.” We used this indicator to examine sexual orientation as indicated by identity from 2008 to the present. Among recent cohorts (2012–2018), we also compared men and women across the four behavioral and one identity indicator of sexual orientation.

We used our broadest behavioral indicator of sexual orientation, “partners since age 18,” and our sexual orientation identity variable to compare reported sexual orientation patterns among recent cohorts (2012–2018) across the following background variables: sex, race/ethnicity, age, marital status, and highest degree. We compared six variables addressing respondents’ sexual and relationship behaviors. These included: 1) marital status; 2) in a relationship with last sexual partner; 3) had extramarital sex while married; 4) ever paid or been paid for sex; 5) had 2 or more sexual partners over the previous year; and 6) had sex more than once a month over the last year.

Data Analysis

Statistical calculations were performed with SPSS version 27 (IBM Corp, 2020). Missing values were excluded pairwise from all analyses. Since the goals of this study were descriptive, we reported odds ratios and confidence intervals where possible. First, we provided frequencies of each sexual orientation indicator, across four time periods from 1989 to 2018 to examine changes over time. We calculated confidence intervals for the proportion of respondents in each category using the standard error formula sqrt [p (1 – p)/n]. This allowed us to provide parameter estimates of the percentage of adults in the United States falling into these categories. We ran chi-square tests to evaluate whether the differences between waves were statistically significant and pairwise comparisons to find out which sexual orientations differed significantly between waves.

Focusing the remaining analyses on the 2012–2018 respondents, we examined patterns in sexual orientation as indicated by sex of sexual partners since age 18 and as indicated by sexual orientation identity across the five previously mentioned background variables, using chi-square tests to evaluate whether associations were statistically significant. We also reported pairwise comparisons to show which demographic categories differed significantly.

Additional analyses examined whether sexual orientation and gender/sex were associated with six dichotomized sexuality and relationship variables. We ran multiple logistic regression analyses that included gender/sex, sexual orientation (analyzed as two dummy variables: gay/lesbian vs heterosexual, and bisexual vs heterosexual), and/or their interaction terms together in each regression model to test the relationship between these variables and each dichotomous sexuality variable. Because our interest was only in the statistical significance of these individual parameter coefficients, we did not report model fit characteristics. We ran separate logistic regression analyses excluding heterosexual persons to examine whether lesbian and gay persons differed from bisexual persons across these variables.

Results

Table 1 depicts the proportion of respondents falling into different sexual orientation categories, as indicated by their sexual orientation identity (first asked during the 2008 wave), and the sex of their sexual partners since age 18, over the past five years, and over the past year. For sexual orientation identity, among the 2008 and 2010 waves (grouped together), people were more likely to identify as gay and lesbian (1.4%) than bisexual (1.2%), while among recent waves (2012–2018) people were more likely to identify as bisexual (2.7%) than gay and lesbian (1.8%). We note that the 95% confidence intervals for the proportion of respondents identifying themselves as bisexual in these two time periods do not overlap, meaning this sexual orientation identity was significantly more common in the 2012–2018 waves than in the 2008–2010 waves. Pairwise comparisons show that the proportions of respondents identifying as heterosexual, but not the proportions identifying as lesbian or gay, also differed significantly between these two waves. Among the 2021 wave, 4.5% of respondents identified as bisexual, compared to 3.5% who identified as gay or lesbian. Although the confidence intervals for the proportion of respondents in the 2021 wave identifying as bisexual do not overlap with the confidence intervals for the 2012–2018 waves, we caution against assuming a significant increase in bisexual identity between the 2012–2018 waves and 2021, due to pandemic-related sampling differences in 2021.

Table 1. Responses to indicators of sexual orientation by time period (weighted).

		I:1989–1994		II:1996–2002		III:2004–2010		IV:2012–2018		V:2021		Columns
Item		n	%	n	%	n	%	n	%	n	%	Compared^a	X^2	(df)	Significant Pair Comparisons^b
Sexual Orientation Identity						(n = 3551)		(n = 7119)		(n = 2221)
Don’t know (N)						14	0.4	20	0.3	0.0	0
Heterosexual or straight (H)						3443	97.0	6774	95.2	2043	92.0
Gay, lesbian, homosexual (GL)						50	1.4	131	1.8	77	3.5
Bisexual (B)						44	1.2	194	2.7	101	4.5
95% CI for Bisexual							(0.8–1.5)		(2.3–3.1)		(3.6–5.4)	III,IV	27.82	(3)***	B, H
Sex of Partners Since Age 18		(n = 6354)		(n = 8388)		(n = 7498)		(n = 6609)		(n = 2032)
No partners (N)		289	4.5	381	4.5	367	4.9	375	5.7	126	6.2
Partners of Other Sex Only (H)		5804	91.3	7562	90.2	6622	88.3	5522	83.6	1668	82.1
Same Sex Partners Only (GL)		61	1.0	136	1.6	82	1.1	98	1.5	42	2.1
Partners of Both Sexes (B)		200	3.1	309	3.7	427	5.7	614	9.3	196	9.6
95% CI for Both sexes			(2.7–3.5)		(3.3–4.1)		(5.2–6.2)		(8.6–10.0)		(8.3–10.9)	I,II:	15.41	(3)**	H, GL
												I,III:	54.28	(3)***	H, B
												I,IV:	232.4	(3)***	N, H, B, GL,
												II,III:	45.13	(3)***	H, GL, B
												II,IV:	217.1	(3)***	N, H, B
												III,IV:	79.03	(3)***	N, H, GL, B
Sex of Partners in Past 5 Years		(n = 4697)		(n = 9314)		(n = 7967)		(n = 7105)		(n = 2230)
No partners (N)		555	11.8	1051	11.3	901	11.3	846	11.9	302	13.6
Partners of Other Sex Only (H)		4024	85.7	7909	84.9	6819	85.6	5957	83.8	1786	80.1
Same Sex Partners Only (GL)		71	1.5	216	2.3	158	2.0	170	2.4	79	3.5
Partners of Both Sexes (B)		47	1.0	138	1.5	89	1.1	132	1.9	63	2.8
95% CI for Both sexes			(0.7–1.3)		(1.3–1.7)		(0.9–1.3)		(1.6–2.2)		(2.1–3.5)	I,II:	16.40	(3)**	GL, B
												I,III:	4.67	(3)
												I,IV:	25.59	(3)***	H, GL, B
												II,III:	6.82	(3)	B
												II,IV:	5.47	(3)
												III,IV:	19.46	(3)***	H, B
Sex of Partners in Past Year		(n = 7186)		(n = 9425)		(n = 8107)		(n = 7174)		(n = 2266)
No partners (N)		1334	18.6	1676	17.8	1485	18.3	1438	20.0	521	23.0
Partners of Other Sex Only (H)		5742	79.9	7468	79.2	6431	79.3	5494	76.6	1651	72.9
Same Sex Partners Only (GL)		88	1.2	228	2.4	156	1.9	167	2.3	78	3.5
Partners of Both Sexes (B)		22	0.3	53	0.6	35	0.4	75	1.0	16	0.7
95% CI for Both sexes			(0.2–0.4)		(0.4–0.8)		(0.3–0.5)		(0.8–1.2)		(0.4–1.0)	I,II:	38.11	(3)***	GL, B
												I,III:	13.59	(3)**	GL
												I,IV:	62.80	(3)***	N, H, GL, B
												II,III:	7.05	(3)	GL
												II,IV:	27.26	(3)***	N, H, B
												III,IV:	32.46	(3)***	N, H, B
Sex of Partners in Past Year
(excluding persons with 0 or 1 partner)		(n = 825)		(n = 1197)		(n = 1000)		(n = 860)		(n = 217)
Partners of Other Sex Only (H)		775	93.9	1074	89.7	930	93.0	752	87.4	182	83.9
Same Sex Partners Only (GL)		28	3.4	70	5.8	41	4.1	38	4.4	23	10.4
Partners of Both Sexes (B)		22	2.7	53	4.4	29	2.9	70	8.1	12	5.7
95% CI for Both sexes			(1.6–3.8)		(3.2–5.6)		(1.9–3.9)		(6.3–9.9)		(2.6–8.8)	I,II:	11.10	(2)**	H, GL, B
												I,III:	0.73	(2)
												I,IV:	26.19	(2)***	H, B
												II,III:	7.34	(2)*	H
												II,IV:	13.77	(2)**	B
												III,IV:	25.54	(2)***	H, B

^aFirst four chi-square tests compared two time periods and four sexual orientation categories and have 3 degrees of freedom.

^bEach letter represents a category that differs significantly (p < .05) between the two time periods based on z-tests for independent proportions (2-tailed, Bonferroni corrected). H indicates heterosexual; GL indicates lesbian or gay; B indicates bisexual; N indicates no partners or don’t know their sexual orientation.

*p < .05. **p < .01. ***p < .001.

In terms of sex of sexual partners, the proportion of persons with only same-sex partners since age 18 increased from the 1989–1994 waves (1.0%) to the 1996–2002 (1.6%) waves but did not consistently increase through the 2012–2018 waves. The high proportion in 2021 may indicate real change but may reflect pandemic-related sampling changes in the 2021 sample. In contrast, the proportion of persons reporting both male and female partners since age 18 increased substantially over this time period, particularly in the most recent waves, from 3.1% (95% CI [2.7–3.5]) in the 1989–1994 waves to 9.3% (95% CI [8.6–10.0]) in the 2012–2018 waves and 9.6% in the 2021 wave. The differences between 95% confidence intervals show that these changes were statistically significant, and chi-square test statistics showed significant differences between each time period (p < .001, except for the contrast between the 1989–1994 period and the 1996–2002 period, which was significant at the p < .01 level).

In terms of partners over the past year and past five years, the proportion of respondents reporting only same-sex partners was greater than the proportion reporting both male and female partners across both variables and all waves, though both increased over time. When we examined only responses of those who reported two or more partners over the past year, we found that the proportion of respondents reporting only same-sex partners was greater than the proportion reporting both male and female partners across all waves except 2012–2018. Though these indicators show a general trend of increases in bisexuality over time, the differences in the reported sex of sexual partners were not consistent or statistically significant across every time period. The clearest and most remarkable change across these variables and time periods is that respondents in the 2012–2018 waves were much more likely to report partners of both sexes than respondents in the 2004–2010 waves. Bisexuality, as indicated by sex of sexual partners, was more commonly reported in the 2012–2018 waves than it was in the 2004–2010 waves no matter which behavior indicator was used. Across all variables and waves there was a general trend toward fewer persons reporting only partners of a different sex. And, across the “partners since age 18” and “partners in past year” indicators, which included the response “no partners,” there was a gradual trend toward more persons without any sexual partners.

Table 2 examines whether there were variations between different demographic categories in the proportion of respondents from the 2012–2018 waves in different sexual orientation categories, as indicated by sexual orientation identity and sex of sexual partners since age 18. Women were significantly more likely to report partners of both sexes (10.3%) than were men (8.0%) (p < .01) and to identify as bisexual (3.7%) than were men (1.6%) (p < .001). Men were more likely than women to report only same-sex partners and to identify as gay or lesbian (combined) than were women. There were no statistically significant differences in the proportion of respondents reporting partners of both sexes or identifying as bisexual by the race/ethnicity variable; however, persons in the White non-Hispanic category were more likely than persons in the Hispanic and Asian/Pacific Islander categories to report having only partners of a different sex, with those in the Asians and Pacific Islander category also much more likely than White non-Hispanic respondents to report having no sexual partners since the age of 18. Respondents across all racial and ethnic categories were not statistically significantly more likely to identify as bisexual than lesbian/gay (combined).

Table 2. Percentage of respondents in different demographic categories falling into each sexual orientation category, 2012–2018 waves (weighted).

Item	Sex of Sexual Partners Since 18				X^2	df	Sexual Orientation Identity^1
	Both Sexes %	Same Sex Only %	Diff Sex Only %	No Partners %			Bi %	L/G %	Het %	X^2	df
Sex/Gender	(n = 6609)				24.53	(3)***	(n = 7100)			33.26	(2)***
Male	8.0^a	2.1^a	84.5	5.4			1.6^a	2.2	96.2^a
Female	10.3^b	1.0^b	82.8	5.9			3.7^b	1.6	94.7^b
Race/Ethnicity	(n = 6526)				30.53	(12)**	(n = 7017)			6.83	(8)
White non-Hispanic	8.7	1.6	84.9^a	4.9^a			2.6	1.7	95.7
Black	10.4	1.3	81.6	6.6			2.9	2.1	95.0
Hispanic	10.3	1.4	81.3^b	7.1			3.0	2.2	94.8
Asian/Pacific Islander	13.2	0.0	76.8^b	9.9^b			4.6	1.5	93.8
Other (Pairwise Diff)	13.1	2.0	78.6	6.1			0.9	2.8	96.3
Age	(n = 6593)				309.79	(12)***	(n = 7081)			115.44	(8)***
18–29	10.3^a,b	2.3	73.3^a	14.1^a			6.1^a	2.5	91.4^a
30–39	12.1^b	1.1	84.0^b	2.7^b			3.7^b	2.3	94.0^a,b
40–49	9.6^a,b	1.4	87.2^b	1.8^b			1.9^b,c	1.9	96.2^b,c
50–59	8.4^a,c	1.3	87.1^b	3.2^b,c			1.2^c	1.6	97.3^c,d
60 and over	6.3^c	1.2	87.5^b	5.0^c			0.8^c	1.2	98.0^d
Marital Status	(n = 6606)				466.94	(12)***	(n = 7097)			191.94	(8)***
Married	6.8^a	0.9^a	90.2^a	2.0^a			1.3^a	0.7^a	98.0^a
Widowed	6.8^a,b	1.3^a,b	86.8^a,b	5.1^b			1.2^a,b	0.3^a	98.5^a,b
Divorced	12.6^c	0.6^a	84.5^b	2.3^a,b			3.2^b,c	1.3^a	95.5^b
Separated	14.7^c	0.0^a,b	81.4^b	4.0^a,b			4.7^b,c	1.0^a,b	94.3^b,c
Never Married	12.1^b,c	3.0^b	71.2^c	13.8^c			5.2^c	4.5^b	90.3^c
Highest Degree	(n = 6605)				42.05	(12)***	(n = 7097)			9.24	(8)
Some High School	11.2	0.8	80.7^a	7.2^a			3.5	0.8	95.8
High School	9.8	1.6	82.0^a	6.6^a			2.6	1.9	95.5
Junior College	10.0	1.2	84.2^a,b	4.7^a,b			2.6	1.5	96.0
Bachelor	7.6	1.4	86.8^b	4.2^b			2.7	2.2	95.1
Graduate	7.4	2.1	87.4^b	3.1^b			2.9	2.2	94.9

Within-column means with different superscripts and no shared superscripts differ significantly from one another at the p < .05 level based on z-tests for independent proportions (2-tailed, Bonferroni corrected).

¹Bi refers to bisexual; L/G refers to gay or lesbian; Het refers to heterosexual. Respondents indicating “don’t know” were excluded due to insufficient n’s.

*p < .05. **p < .01. ***p < .001.

Younger respondents were much more likely to identify as bisexual than older persons and somewhat more likely to report partners of both sexes, though the proportions were lower for 18–29 year-olds (10.3%) than for 30–39 year-olds (12.1%). In contrast, respondents 60 and over were much less likely than other respondents to identify as bisexual (0.8%) and report partners of both sexes since age 18 (6.3%) despite having more years of life experience. Divorced (12.6%; 3.2%), separated (14.6%; 4.7%), and never married persons (12.1%; 5.2%) were significantly (p < .001) more likely to report partners of both sexes and to identify as bisexual than were married (6.8%; 1.3%) or widowed persons (6.8%; 1.2%). People with bachelor’s or higher degrees were less likely to report having sexual partners of both sexes than people with less education.

Tables 3 and 4 examine the interaction between gender/sex and sexual orientation among the 2012–2018 waves across 6 sexual and relationship variables. Table 3 uses the sexual orientation identity variable, while Table 4 uses the sex of sex partners since age 18 variable. For simplicity, we report only statistically significant (p < .05) main or interaction effects here. Women in general were more likely than men to report being in a relationship with their last sexual partner and being married or widowed. Men in general were more likely than women to report having had sex more than once a month over the previous year, having two or more partners over the previous year, having paid for sex, and having had sex with someone other than a spouse while married.

Table 3. Sexuality variables according to gender/sex and sexual orientation identity, 2012–18 waves (weighted).

	Heterosexual		Lesbian or Gay		Bisexual		Logistic Regression Coefficients^a
Item	Men %	Women %	Men %	Women %	Men %	Women %	Odds Ratio		95% C.I. for OR
Married or Widowed (n = 7097)	56.3 (1767/3140)	58.7 (2132/3632)	18.3 (13/71)	23.7 (14/59)	32.7 (17/52)	24.5 (35/143)
							I	0.905	(0.822–0.997)*
							II	0.221	(0.121–0.402)***
							III	0.228	(0.155–0.336)***
							IV	0.773	(0.329–1.816)
							V	1.591	(0.786–3.217)
In a Relationship with Last Sexual Partner (n = 6894)	86.1 2639/3064	92.7 3254/3509	60.6 43/71	93.2 55/59	66.7 34/51	80.0 112/140
							I	0.486	(0.413–0.573)***
							II	1.027	(0.379–2.786)
							III	0.314	(0.204–0.485)***
							IV	0.240	(0.079–0.730)*
							V	1.044	(0.499–2.186)
Have Sex Other Than Spouse While Married (n = 5020)	19.2 415/2161	11.8 321/2729	52.6 10/19	36.4 8/22	58.3 14/24	44.6 29/65
							I	1.782	(1.522–2.088)***
							II	3.888	(1.598–9.459)**
							III	6.057	(3.673–9.986)***
							IV	1.205	(0.336–4.326)
							V	0.975	(0.372–2.554)
Paid or Been Paid For Sex Since 18 (n = 7055)	9.9 309/3115	2.0 73/3614	15.5 11/71	5.0 3/60	11.5 6/52	6.3 9/143
							I	5.302	(4.092–6.872)***
							II	2.352	(0.689–8.026)
							III	3.132	(1.519–6.459)**
							IV	0.700	(0.174–2.812)
							V	0.345	(0.110–1.087)
Had Two or More Partners in Last Year (n = 7098)	14.9 469/3141	7.2 261/3633	29.6 21/71	16.9 10/59	57.7 30/52	44.4 63/142
							I	2.271	(1.935–2.664)***
							II	2.749	(1.393–5.424)**
							III	10.340	(7.260–14.726)***
							IV	0.894	(0.381–2.097)
							V	0.740	(0.382–1.434)
Had Sex More than Once a Month Last Year (n = 6474)	60.2 1738/2889	56.4 1847/3275	56.7 38/67	69.6 39/56	72.9 35/48	62.6 87/139
							I	1.167	(1.054–1.292)**
							II	1.718	(0.971–3.037)
							III	1.284	(0.905–1.820)
							IV	0.502	(0.237–1.065)
							V	1.405	(0.676–2.918)

^aCoefficients: I-Men in contrast to women, with women as reference group; II-Gay or lesbian in contrast to heterosexual, with heterosexual as reference group; III-Bisexual in contrast to heterosexual; IV–Interaction effects between sex and sexual orientation (heterosexual vs gay/lesbian); V–Interaction effects between sex and sexual orientation (heterosexual vs bisexual).

Table 4. Sexuality variables according to gender/sex and sexual orientation as indicated by sex of sexual partners since age 18, 2012–2018 waves (weighted).

	Only Partners of Other Sex		Only Partners of Same Sex		Partners of Both Sexes		Logistic Regression Coefficients^a
Item	Men Men (%)	Women Women (%)	Men Men (%)	Women Women (%)	Men Men (%)	Women Women (%)	Odds Ratio		95% C.I. for OR
Married or Widowed (n = 6230)	58.0 (1453/2505)	61.6 (1858/3014)	34.4 (21/61)	38.9 (14/36)	47.7 (113/237)	36.9 (139/377)
							I	0.860	(0.772–0.958)**
							II	0.409	(0.209–0.799)**
							III	0.364	(0.292–0.454)***
							IV	0.941	(0.400–2.211)
							V	1.800	(1.272-.547)**
In a Relationship with Last Sexual Partner (n = 6187)	88.2 2202/2497	94.3 2812/2983	62.7 37/59	91.7 33/36	78.4 185/236	86.4 325/376
							I	0.454	(0.373–0.553)***
							II	0.584	(0.189–1.799)
							III	0.389	(0.278–0.543)***
							IV	0.390	(0.112–1.361)
							V	1.265	(0.789–2.028)
Have Sex Other Than Spouse While Married (n = 4554)	19.7 353/1791	10.2 239/2346	29.2 7/24	17.6 3/17	33.1 50/151	36.0 81/225
							I	2.162	(1.811–2.581)***
							II	1.641	(0.443–6.079)
							III	4.926	(3.635–6.674)***
							IV	1.014	(0.207–4.952)
							V	0.408	(0.255–0.653)***
Paid or Been Paid For Sex Since 18 (n = 6228)	10.2 256/2505	1.4 43/3013	8.1 5/62	2.8 1/36	17.7 42/237	7.7 29/375
							I	7.863	(5.666–10.913)***
							II	2.741	(0.487–15.427)
							III	5.834	(3.599–9.455)***
							IV	0.258	(0.036–1.864)
							V	0.328	(0.180–0.598)***
Had Two or More Partners in Last Year (n = 6223)	16.3 409/2507	6.9 207/3015	29.5 18/61	22.2 8/36	24.1 57/237	23.6 89/377
							I	2.646	(2.218–3.157)***
							II	3.887	(1.754–8.615)**
							III	4.199	(3.185–5.536)***
							IV	0.562	(0.213–1.483)
							V	0.391	(0.257–0.594)***
Had Sex More than Once a Month Last Year (n = 5829)	63.5 1505/2369	58.2 1623/2787	59.3 35/59	62.9 22/35	58.4 132/226	70.0 247/353
							I	1.249	(1.116–1.398)***
							II	1.239	(0.623–2.127)
							III	1.674	(1.317–2.127)***
							IV	0.684	(0.287–1.629)
							V	0.481	(0.334–0.694)***

^aCoefficients: I-Men in contrast to women; II-Only same sex partners in contrast to only partners of the other sex; III-Partners of both sexes in contrast to only partners of the other sex; IV–Interaction effects between sex and sexual orientation (heterosexual vs gay/lesbian); V–Interaction effects between sex and sexual orientation (heterosexual vs bisexual).

Respondents identifying as bisexual or reporting both male and female partners were less likely than persons identifying as heterosexual or reporting only partners of a different sex to be in a relationship with their last sexual partner or to be married or widowed. They were more likely to report having had sex with someone other than a spouse while married, having paid or been paid for sex, having had two or more partners over the previous year, and having had sex more than once a month over the previous year, with the latter pattern statistically significant only for the sex of sex partners indicator. Respondents identifying as gay or lesbian or reporting only partners of the same sex were less likely than heterosexual persons to be married or widowed and more likely to report having had two or more partners over the previous year. Further, respondents identifying as gay or lesbian were more likely than those identifying as heterosexual to report having had sex with someone other than a spouse while married. The difference was not statistically significant when the “sex of sex partners” indicator of sexual orientation was used. Statistically significant interactions between sex and having partners of both sexes (in comparison to having only partners of a different sex) appeared across five variables, with women who had both female and male partners disproportionately likely to report having paid or been paid for sex, having had two or more partners over the past year, and having sex more than once a month over the past year, and being disproportionately less likely to report being married or widowed. In fact, women reporting both male and female partners since age 18 were more likely than all other combinations to report having had sex more than once a month over the last year. Separate logistic regression analyses comparing lesbian and gay persons with bisexual persons (and excluding heterosexual persons) revealed no statistically significant differences, and these results are not reported.

Discussion

This investigation used nationally representative survey data collected between 1989 and 2021 to describe the proportion of U.S. adults identifying as bisexual or reporting having sexual partners of both sexes in contrast to the proportion of adults reporting other sexual orientations. Further, we sought to investigate whether these proportions had changed over time, whether there were demographic patterns in the proportion of persons reporting different sexual orientations, and whether women and men in different sexual orientation categories reported different sexual and relationship patterns. Notable findings include the following:

• Hypothesis 1 was supported, as there was a threefold rise in the proportion of respondents reporting partners of both sexes over time, with 3.1% of respondents reporting partners of both sexes since age 18 among the 89–94 waves and 9.3% reporting this in the 2012–18 waves.

• Hypothesis 2 was supported, as respondents from the 2012–2018 waves were more than twice as likely to identify themselves as bisexual than respondents from the 2008–2012 waves.

• Among the 2012–18 waves and the 2021 wave, respondents were more likely to identify as bisexual or report having had partners of both sexes since age 18 than they were to identify as gay/lesbian (combined) or to report having only same-sex partners.

• When respondents reported their sexual behavior over shorter time periods (one year or five years), more respondents reported having only same-sex partners than reported having partners of both sexes.

• Younger persons and women were more likely than older persons or men to report partners of both sexes and to identify as bisexual.

• Bisexual and gay/lesbian (combined) respondents differed in similar ways from heterosexual respondents across sexual and relationship behaviors.

Conceptualizing Sexual Orientation

Our findings present an opportunity to look more analytically at the term “sexual orientation” and how it is defined both in research and in U.S. culture more broadly. Sexual orientation can be indicated by the sex of one’s sexual partners, how one identifies their sexual orientation, and by the sex to which one is attracted. We used the first two of these indicators in our analyses, because the third was not included on the General Social Survey. This way of conceptualizing sexual orientation and others inevitably fail to fully capture the range and diversity of human sexual desire, behavior, and identity (Swan, 2018). The fact that the two measures we used yielded different proportions and sometimes different patterns points to the importance of recognizing that sexual behavior and sexual orientation identity do not line up perfectly for many persons. Women in our sample were over twice as likely to identify as bisexual than were men but only slightly more likely to report sexual partners of both sexes, perhaps indicating that bisexual identity remains more stigmatized for men than for women (Sarno et al., 2020). Using sexual behavior as an indicator of bisexuality means that scholars are left to draw a line as to what degree of frequency and recency of sexual experience with partners of both sexes can be defined as bisexuality (Bauer & Brennan, 2013). Our results show that reporting partners of both sexes during the past year and past five years was much less common than it was when we asked about the sex of partners since age 18.

Our decision to use “sex of sexual partners since age 18” was deliberate, designed to contrast a very broad behavioral measure of bisexuality with a person’s sexual orientation identity. The two are clearly different, with the overwhelming majority of persons who have had partners of both sexes since age 18 not identifying themselves as “bisexual.” There are several likely explanations for this. First, sexual behavior can be exploratory, taking place before a person has clarified their own sexual orientation identity, a process that continues throughout young adulthood (Kaestle, 2019). Our measure likely captures gay and lesbian persons who had partners of a different sex before they came out or clarified their sexual orientations, which can occur later in life (Rosario, 2019). Second, sexual orientation can be fluid (Campbell et al., 2021; Diamond, 2021), with some people changing their sexual behavior and/or their sexual orientation identities over time. In both of these cases, using the sex of sexual partners since age 18 measure means that we may be capturing sexual experiences that have taken place many years ago, especially for older respondents. Finally, even persons who have more recently had partners of both sexes may not consider themselves bisexual, and the term may not fit their understandings of themselves and their sexual behavior. When we used the “sex of sex partners in the past 5 years” indicator, more people had only partners of the same sex than partners of both sexes. This is likely because many respondents were involved in ongoing monogamous relationships. In other words, even people who identify as bisexual may have partners of only one sex for significant periods of time.

Demographic Patterns in Bisexuality

The increase in the proportion of persons identifying as bisexual and reporting partners of both sexes accompanies a decline in the proportion of persons identifying exclusively as heterosexual and reporting partners of only the other sex. This suggests a loosening of the social norms and institutional enforcement that have privileged heterosexuality over other sexual orientations (Seidman, 2009). As we have noted, women and younger persons were more likely than men or older persons to report having had partners of both sexes or to identify as bisexual. The association with age likely reflects the greater acceptance that young persons have for same-sex and bisexual relationships and identities. That acceptance may mean that some people who would not have had sexual experiences with partners of both sexes in the past due to social stigma are having those experiences today. Additionally, the growing recognition and acceptance of bisexuality as a “real” sexual identity likely means that people who are attracted to both sexes but would have previously identified themselves as heterosexual, gay, or lesbian, may now be more willing to identify themselves as bisexual. This may be especially true for individuals currently in monogamous relationships with one partner who could identify themselves as either heterosexual or gay or lesbian.

The proportion of respondents who identified as bisexual or reported sexual partners of both sexes was not statistically significantly associated with the racial or ethnic category with which they identified. However, there were some statistically significant race/ethnicity differences in the proportion with partners of a different sex and having no sexual partners. These differences underscore the possibility that the norms and social conditions shaping sexual behavior and sexual orientation identity may differ among different racial and ethnic groups. Studies with the capacity to more qualitatively examine the relationship between sexuality and ethnicity and to disaggregate different racial categories can shed additional light on this topic.

Bisexuality and Sexual Behavior

Our findings also revealed that bisexual and gay/lesbian (combined) respondents differed in similar ways from heterosexual persons across sexual and relationship behaviors. Although logistic regression analyses revealed no statistically significant differences between lesbian and gay persons in comparison to bisexual persons, smaller effect sizes and a smaller sample of respondents with only partners of the same sex or who identified as lesbian or gay may have obscured differences that could be revealed by research using larger samples and more nuanced measures. We note that what appears to be an interesting interaction between gender/sex and sexual orientation, with women who have had both male and female sexual partners since age 18 more likely than other combinations to report having sex one or more times per month over the previous year, may not be what it seems. First, the same pattern was not found when using the sexual orientation identity measure. Second, the percentage of women reporting partners of both sexes since age 18 is more than twice as high as the proportion identifying as bisexual, 10.3% compared to 3.7% (see Table 2), meaning most women who have had partners of both sexes since age 18 do not consider themselves bisexual. Instead of the pattern indicating that bisexual women have more frequent sex, the finding may simply reflect that women with high “sociosexuality” (Schmitt, 2005; Wongsomboon et al., 2020) have more frequent sex as well as greater openness to same-sex sexuality, with the latter trait potentially shaped by genetic patterns (Diamond, 2021).

With recent research indicating that genetic traits may explain as much as 25% of the variance in same-sex sexual behavior among both persons with only same-sex partners and those with partners of both sexes (Ganna et al., 2019), it is important to recognize that social and cultural factors affect how genetic tendencies are expressed. The increase we found in the proportion of persons in the U.S. reporting partners of both sexes since age 18 over the past 30 years and the increase in the proportion identifying as bisexual cannot be explained by genetics. Society draws lines around sexual behavior, and these lines, though often in dispute, may move over time. These changing norms appear to have affected the proportion of people who identify as bisexual and likely also affected actual sexual behavior. Hart-Brinson (2018) made a similar claim, exploring much more thoroughly the social factors, from activism to media, that have led younger persons in the U.S. to embrace same-sex marriage and accept same-sex relationships.

Society and Sexual Orientation

To propose that individual sexual orientation may be shaped by society should not be misconstrued to suggest that it is a choice or that it is experienced as a choice. The lived experiences of lesbian, gay, and bisexual persons, which are often described as a growing recognition of one’s sexuality or a knowing from a young age, belies any sense that we choose our sexual orientation. But neither should this fact be interpreted to mean that society and the social environment are irrelevant in the development of sexual desire and sexual orientation. Cross-cultural and historical variations in sexual norms and practices (Penniston & Chivers, 2022; Tskhay & Rule, 2015), sociological analyses (Foucault, 1978; Gagnon & Simon, 2017; Giddens, 2013), and the present findings demonstrate that the norms associated with sexuality are socially negotiated and changeable. Many aspects of who we are and what we desire are not experienced as choices but are rather a complicated interplay between biologically mediated desire and the social environment.

Apart from the question of whether one’s sexual desires are shaped by society, the categories through which we identify ourselves are necessarily social (Grassi, 2016). The fact that there is stigma associated with being anything other than heterosexual (Jeffries et al., 2021; Rubin, 1984) means that the category with which one chooses to identify reflects not only one’s patterns of sexual attraction but also the levels of social approval associated with that identity. Bisexual persons in committed relationships with just one partner could choose to identify as gay even if they have had partners of a different sex, or straight even if they have had partners of the same sex. Other research consistently reveals that a proportion of self-identified lesbian persons have had male partners (Bauer & Jairam, 2008), and a proportion of men who have sex with men identify as straight (Silva, 2018). In fact, Hoy and London (2018) reported that as much as half of the same-sex behavior reported in the National Survey of Family Growth occurred between self-identified heterosexual persons. Contemporary culture provides a range of new terms that allow flexibility for persons who experience same-sex desire and/or have same-sex sexual experiences but do not identify as gay or lesbian. Among terms and phrases described in recent research and popular articles are “bi-curious,” “heteroflexible,” and “straight with a pinch of bi,” all of which may contribute to a greater elasticity in what it means to be “heterosexual” (Carrillo & Hoffman, 2018; Zosky & Alberts, 2016).

Our findings reveal changes in sexual orientation, as indicated by behavior and identity, over time within the United States. However, because social norms affect sexual behavior, and social definitions and social stigma affect sexual orientation identity, it would be reasonable to expect substantial cross-cultural variation in these indicators of sexual orientation. Unfortunately, representative data from countries where bisexuality and same-sex sexuality are highly stigmatized or criminalized are not available. More sexually tolerant European countries have only recently begun to collect data on sexual orientation identity, and data on bisexual and same-sex sexual behavior is limited as well. The German Socio-Economic Panel’s (SOEP) household panel survey indicated that only about 1.9% of Germans identified as gay, lesbian, or bisexual in 2016 (Organization for Economic Cooperation and Development [OECD], 2023). The German Health and Sexuality Survey (GeSiD), conducted from 2018–2019 (Briken et al., 2021) should give researchers an opportunity to examine these issues in greater detail with more recent data. Findings from the United Kingdom’s Annual Population Survey (APS) indicated that about 3.1% of persons aged 16 or older identified as lesbian, gay, or bisexual in 2020, nearly double the 1.6% reported in 2014 (Office of National Statistics, 2022). These proportions are lower than the 4.5% we found in the 2012–2018 waves and the 8.0% we found in the 2021 wave of the General Social Survey. The U.K.’s National Survey of Sexual Attitudes and Lifestyles (Natsal-3, 2010–2012), which provides greater detail on sexual behavior, found that 4.7% of men and 5.9% of women reported having sexual experiences with persons of both sexes during their lifetimes (Mercer et al., 2013), percentages similar to the 5.7% (combining both men and women) we found for the 2008–2010 waves but lower than the 9.3% we found for the 2012–2018 waves of the GSS. Natsal-4 is currently being conducted, which will allow researchers to find out whether having partners of both sexes has increased in the U.K. as it has in the U.S. Future research using these and other data sources could reveal whether countries besides the U.S. are experiencing increases in the percentages of persons identifying as bisexual or reporting partners of both sexes.

Strengths and Limitations

The present study addressed a critically important issue in the experience of human sexuality using high quality, nationally representative data. By combining waves, we were able to analyze samples large enough to provide meaningful population parameters for the proportion of adults in the U.S. who identify as bisexual or lesbian and gay, or who have had partners of both sexes. Further, because the General Social Survey employs many of the same measures from wave to wave, the data provide an excellent opportunity to examine and document changes in responses over time. Finally, the GSS provides highly accessible, public domain data, meaning other researchers can freely access the same measures and items we used in order to study additional topics and explore other covariates.

The limitations of this study include crude measures, imprecision, and the noise associated with survey-based research. As described throughout this article, the terminology and language associated with sexuality and gender is a product of society and can change over time. Because of this, using consistent measures over time introduces challenges in the interpretation of survey research. When the General Social Survey first began asking people to identify their sexual orientation in 2008, the choices listed were “Gay, lesbian, or homosexual,” “Bisexual,” and “Heterosexual or straight” (Smith et al., 1972–2021), a limited array of choices that may not capture the identity of some respondents. Additionally, the sex to which one is attracted, a key indicator of sexual orientation, is not assessed in the General Social Survey, leaving open the question of whether the proportion of people who are sexually or romantically attracted to persons of both sexes has changed over time and how this indicator is related to other indicators of sexual orientation. The present study relied on a set of questions on sexual behavior from the General Social Survey that were first asked in 1989 and continue to be asked today. It is important to acknowledge that the imprecision of the questions means that respondents are left on their own to determine what is meant when they are asked about their “sex” partners. The heteronormative definition of sex does not always match the definitions held by sexual minorities. While heterosexual penetrative intercourse is nearly universally recognized as counting as “sex,” other intimate or sexual behaviors vary as to whether they are defined as sex among heterosexual persons and persons with minoritized sexualities. There is no single agreed upon act that qualifies as “having sex” for many lesbian, gay, and bisexual identifying persons (Sewell et al., 2017). Additionally, while this study provides a much-needed focus on bisexuality, we acknowledge that research examining sexualities through the categories of heterosexuality, gay and lesbian sexuality, and bisexuality has tended to assume cisgender identities, missing the richness and complexity that would come from thoughtfully incorporating trans and queer gender identities methodologically and conceptually. Though this is not possible in a trend study like this one, future research should take advantage of a question on transgender identity recently added to the General Social Survey (Smith & Son, 2019). We also note that questions asked at one point in time inevitably fail to capture the experiences of people whose gender identity and sexual orientation are fluid and vary over time.

Finally, in survey research, issues of social desirability arise, especially for sensitive behaviors and stigmatized identities. We can assume that some of the respondents in earlier waves of the General Social Survey who experienced sexual attraction for both men and women may not have acted on those desires. While it is likely that some of the increase in reporting sexual partners of both sexes reflects the increasing acceptance of bisexuality rather than actual changes in behavior, this also reflects the power of society to shape human sexuality.

Conclusion

This study used GSS data that employ the same measures over time and a nationally representative probability sample. Therefore, we can be reasonably confident that the population parameters reported here are indicative of trends in the larger culture. The proportion of persons identifying as bisexual and the proportion of persons who report engaging in sex with partners of both sexes in the U.S. have increased markedly since 1989. Future research on sexual orientation can better recognize that the terms with which we identify ourselves are social and that there is some degree of choice about how to identify our sexual orientation, particularly among individuals with histories of both male and female partners. Additionally, the recognition that sexual behavior is normative should be more thoroughly considered in research on sexual orientation and sexual behavior.

Acknowledgments

We wish to thank the editor and the anonymous reviewers as the Journal of Sex Research for their editorial suggestions and feedback.

Disclosure Statement

No potential conflict of interest was reported by the authors.