Attitudes toward Bisexual Men and Women among a Nationally Representative Probability Sample of Adults in the United States

Abstract

As bisexual individuals in the United States (U.S.) face significant health disparities, researchers have posited that these differences may be fueled, at least in part, by negative attitudes, prejudice, stigma, and discrimination toward bisexual individuals from heterosexual and gay/lesbian individuals. Previous studies of individual and social attitudes toward bisexual men and women have been conducted almost exclusively with convenience samples, with limited generalizability to the broader U.S. population. Our study provides an assessment of attitudes toward bisexual men and women among a nationally representative probability sample of heterosexual, gay, lesbian, and other-identified adults in the U.S. Data were collected from the 2015 National Survey of Sexual Health and Behavior (NSSHB), via an online questionnaire with a probability sample of adults (18 years and over) from throughout the U.S. We included two modified 5-item versions of the Bisexualities: Indiana Attitudes Scale (BIAS), validated sub-scales that were developed to measure attitudes toward bisexual men and women. Data were analyzed using descriptive statistics, gamma regression, and paired t-tests. Gender, sexual identity, age, race/ethnicity, income, and educational attainment were all significantly associated with participants' attitudes toward bisexual individuals. In terms of responses to individual scale items, participants were most likely to “neither agree nor disagree” with all attitudinal statements. Across sexual identities, self-identified other participants reported the most positive attitudes, while heterosexual male participants reported the least positive attitudes. As in previous research on convenience samples, we found a wide range of demographic characteristics were related with attitudes toward bisexual individuals in our nationally-representative study of heterosexual, gay/lesbian, and other-identified adults in the U.S. In particular, gender emerged as a significant characteristic; female participants’ attitudes were more positive than male participants’ attitudes, and all participants’ attitudes were generally more positive toward bisexual women than bisexual men. While recent population data suggest a marked shift in more positive attitudes toward gay men and lesbian women in the general population of the U.S., the largest proportions of participants in our study reported a relative lack of agreement or disagreement with all affective-evaluative statements in the BIAS scales. Findings document the relative lack of positive attitudes toward bisexual individuals among the general population of adults in the U.S. and highlight the need for developing intervention approaches to promote more positive attitudes toward bisexual individuals, targeted toward not only heterosexual but also gay/lesbian individuals and communities.

Introduction

Bisexual individuals commonly report experiencing stigma, prejudice, and discrimination regarding their sexual identity from both heterosexual and gay/lesbian individuals [1, 2]. This stigmatization of bisexuality among monosexual (i.e., exclusively heterosexual and exclusively homosexual) individuals has been documented in scientific and community literature as “biphobia” (or more recently “binegativity” or “anti-bisexual prejudice”) [3–5]. Biphobia derives in part from heterosexist reaction to sexual expression that is not heteronormative (same-gender sexual expression), as well as monosexist reaction to sexual expression that is not monosexual (e.g., sexual expression with more than one gender). Beyond default assumptions that privilege heterosexuality, negative attitudes toward bisexual groups are also grounded in monosexism, or the belief that people can only be either heterosexual or gay/lesbian, i.e., one or the other. Binary conceptions of sexuality and sexual identity can further fuel dismissive and denigrating attitudes toward bisexual persons, from both heterosexual and gay/lesbian groups [4].

Thus, while sexual minority persons do share some collective experiences of prejudice living in a heteronormative and heterosexist society (e.g., [6, 7]) bisexual individuals are additionally burdened by negative attitudes from gay and lesbian individuals. The deployment of oppressive structures often create multiple levels of oppressive dynamics, from the dominant group to the minority group, and also from minority group to other minority groups [8]. In other words, from a sexual rights perspective, negative attitudes from gay men and lesbian women toward bisexual individuals can be conceptualized as reinforcing of larger oppressive structures against non-heteronormative, non-monosexual forms of expression [9, 10]. Anderson & McCormack note ten forms of “bisexual burden” that may contribute to an increased burden of mental and other health challenges faced by bisexual individuals [11]. Prior research on social attitudes toward sexual minority individuals typically aggregate “lesbian, gay, and bisexual individuals” into a single “LGB” or “sexual minority” category, despite the distinct anti-bisexual prejudices faced by bisexual individuals. Among these, bisexuality is often trivialized as an illegitimate and transitional “phase” in which the person is assumed to be on their way to a “valid” (i.e., monosexual) identity—resulting in colloquial microaggressions such as “bi now, gay later” [5, 12–17]. Bisexual individuals are also commonly characterized as sexually promiscuous, even responsible for serving as a “bridge population” for HIV and other sexually transmitted infections (STI) from partners of one gender to others [3, 18–20]. This is also related to a broader discourse on any form of non-monogamy as “risky” in terms of HIV/STI. As such, another common stereotype involves equating bisexuality with non-monogamy, and bisexual individuals being inherently “incapable” of maintaining monogamous relationships [1, 5].

Impacts of Negative Attitudes on Health

Negative attitudes derived from such stereotypes may impact the health of bisexual individuals, as they develop into and are enacted in subsequent forms of stigma and prejudice which cumulatively contribute to a state of “minority stress” [21]. The deleterious effects of stigma and prejudice on the health of sexual minority individuals have been well-documented across both physiological and psychological domains. Theoretical models on relations between health disparities and poor health outcomes assume that sexual minority individuals experience distinct elevated levels of stress due to the social stigmatization of their sexual orientation/identity [21, 22]. Broadly, stress can be conceptualized as a biopsychosocial process occurring at three interconnected levels: 1) biological (e.g., elevated diastolic blood pressure); 2) psychological (e.g., stress affect); and 3) social (e.g., experiences of prejudice) [23]. Such a model demonstrates intersections between an individual’s social interactions and their physiological health may be a contributing factor in explaining the highest rates of many adverse health outcomes are found among bisexual individuals, who experience distinct and multiple levels of sexuality-related stigmatization.

Research on the etiologies of health disparities among sexual minority individuals, beneath the surface of “minority stress,” has also highlighted the importance of social support and material resources as moderators of stigma-induced stress [22]. Bisexual individuals’ experiences of biphobia and subsequent marginalization from gay/lesbian, and heterosexual communities (i.e., sources of social support and resources) place them at elevated risk for physical and mental health disparities [24]. Recent research has documented the existence of a wide range of distinct health disparities among bisexual individuals, relative to their exclusively heterosexual and homosexual counterparts [25, 26]. Compared to their gay and lesbian peers, bisexual individuals report higher rates of mood and anxiety disorders [27, 28], physical and emotional abuse [29], and substance abuse [30], which are strong predictors of early and excess mortality.

Previous Explorations of Attitudes toward Bisexuality in the United States

Data from large polling agencies reveal that attitudes toward “homosexuality” and “gays and lesbians” have become markedly more positive over the past decade [31], both in the U.S. and throughout much of the world [32]. Previous social and behavioral science researchers have made efforts to assess attitudes toward bisexuality, as distinct from heterosexuality and heterosexuality, in a range of convenience samples in the U.S. and other contexts [33]. Mohr and Rochlen developed the Attitudes Regarding Bisexuality Scale (ARBS) with convenience samples of lesbian, gay, and heterosexual college students and found two factors associated with attitudes towards bisexuality: tolerance and stability [34]. Tolerance determined whether bisexuality was viewed as morally acceptable, whereas stability determined whether bisexuality was perceived as a legitimate sexual orientation. Attitudes regarding bisexuality varied greatly in respect to the gender of the bisexual individual; specifically, bisexual men were rated less positively than bisexual women in studies with other convenience samples of heterosexual college students [18, 35–37]. Brewster and Moradi developed and psychometrically evaluated another measure, the Anti-Bisexual Experiences Scale, using a different scaling approach based on data from convenience samples of bisexual individuals [5]. Their scale development process involved separate measurements of anti-bisexual experiences directed from heterosexual and gay/lesbian individuals, and their findings align with negative attitudinal findings from other samples [1, 34].

Herek [38] conducted the only prior study using probability sampling to examine adults’ attitudes towards bisexual men and women in the U.S., although his sample was limited to heterosexual individuals. Attitudes towards bisexual individuals were less favorable than a wide range of religious, racial, political, and sexuality groups (e.g., Jewish individuals, Haitian individuals, pro-choice individuals, lesbian and gay individuals). Indeed, injection drug users were the only group to be perceived more negatively. Gender differences were observed among heterosexual respondents, with female participants rating bisexual men and women less favorably than gay/lesbian individuals, and male participants reporting more positive views towards lesbian and bisexual women than gay and bisexual men. Although strengthened by probability sampling, the study was limited in that it focused solely on heterosexual individuals’ attitudes (not those of gay/lesbian individuals’). Additionally, the findings are nearly two decades old and predate a number of structural and societal changes that have reflected changes in attitudes among heterosexual individuals toward sexual minority individuals in some ways, including legislation recognizing the legality of same-sex marriage [39].

Study Aims

Our study aimed to examine attitudes toward bisexual men and women in the general population of the U.S., using a probability sample of heterosexual, gay, lesbian, and other-identified participants from the 2015 National Survey of Sexual Health and Behavior (NSSHB). We used abridged versions of the Bisexualities: Indiana Attitudes Scale (BIAS) [40], incorporating two forms measuring distinct attitudes towards bisexual men and bisexual women—i.e., the BIAS-m and BIAS-f, respectively. (While numerous gender identities exist outside a male/female binary, the BIAS specifically measures attitudes towards bisexual individuals who identify as men and women without specific mention of whether these men/women are transgender or cisgender, as prior studies have documented differences in attitudes between these two groups.)

The BIAS assesses a number of domains that have emerged in previous studies, including bisexual men and women being confused or in transition regarding their sexual orientation, bisexual men and women as hypersexual, and bisexual men and women as vectors of STI. The current study is innovative not only in terms of sampling, by relying on a large probability sample of adults from throughout the U.S., but also because it is among the first to examine attitudes toward bisexual men and women in a nationally-representative sample of not only heterosexual but also gay, lesbian, and other-identified individuals.

Materials & Methods

The National Survey of Sexual Health and Behavior (NSSHB) is an ongoing population-based survey of adults and adolescents in the U.S. [41]. The first wave of data was collected in 2009, and subsequently in 2012, 2013, 2014, and 2015. This paper presents the data from the 2015 NSSHB, which was conducted during November and December 2015, via the KnowledgePanel of GfK Research (GfK) (Menlo Park, California). Research panels accessed through GfK are based on a national probability sample established using both random digit dialing (RDD) and an address-based sampling (ABS) frame. ABS involves the probability sampling of a frame of residential addresses derived from the U.S. Postal Service’s Delivery Sequence File, a system that contains detailed information on every mail deliverable address in the U.S. Collectively, the sampling frame from which participants are recruited covers approximately 98% of all U.S. households. Randomly selected addresses are recruited to the research panel through a series of mailings and subsequently by telephone follow-ups to non-responders when possible. To further correct sources of sampling and non-sampling error, study samples are corrected with a post-stratification adjustment using demographic distributions from the March 2015 Current Population Survey (CPS), the monthly population survey conducted by the U.S. Bureau of the Census considered to be the standard for measuring demographic and other trends in the U.S. These adjustments result in a panel base weight that was employed in a probability proportional to size selection method for establishing the samples for this study. Oversampling of specific subgroups are corrected by adjusting the corresponding weights accordingly with the CPS benchmarks serving as reference points.

Once the sample frame was established, all individuals within that frame received a recruitment message from GfK that provided a brief description of the NSSHB and invited them to participate. A total of 2,999 adults (ages 18+) completed the survey, including 1,320 from the general population, 1,230 from an oversample of 18–34 year-olds, and 156 from an oversample of gay men. Given sufficient participation across sexual identity categories, the oversample of gay men was excluded in this analysis, resulting in a total unweighted sample of 2,843 adults.

Post-stratification weights, provided by GfK, were used to maximize generalizations. The post-stratification weights were produced using an iterative proportional fitting procedure that aligned the study sample to all study benchmark distributions from the March 2015 CPS. All results hereafter presents the weighted data. As we focused on the attitudes of non-bisexual individuals about bisexual persons, 33 self-identified bisexual male and 61 bisexual female participants were intentionally excluded from subsequent analyses. Given the very small number of transgender participants (N = 5), these individuals were also removed the analyses. The weighted total sample resulted in 3,221 adults, with 2,434 from the general population and 787 from the 18–34 year-old oversample. A sensitivity analysis excluding the oversample of 18–34 year-old adults was conducted for all analyses. No significant difference were observed, and therefore, the oversample of 18–34 year-old adults was pooled with adults from the general population.

Measures

In preparation for survey data collection, we engaged a panel of expert reviewers from a range of both scientific and community-based organizations who reviewed the original BIAS scales (26-item BIAS-m sub-scale for male participants, 27-item BIAS-f sub-scale for female participants) [1, 40]. Previous studies have demonstrated differences in attitudes toward bisexual women in comparison to bisexual men and, as such, we sought to determine whether this was the case in a national probability sample. A panel of six expert reviewers helped the researchers to identify common themes across scale items in order to identify those that tapped into the most important factors that emerged during prior data analysis [1, 12]. While our pilot study, as well as other studies, showed significant differences in attitudes toward bisexual men and bisexual women, the same 5 themes were selected by reviewers as most relevant for both men and women, specifically:

1. Perceptions of confusion

2. Perceptions of HIV/STI “riskiness”

3. Perceptions of non-monogamy

4. Perceptions of promiscuity

5. Perceptions of bisexuality as temporary

After examining feedback from reviewers, the study team identified the highest ranked items and conducted a second level of review with the expert panel. Reviewers agreed, despite similarity of the scale items, that separate items for attitudes toward bisexual men and bisexual women are necessary in order to capture potentially important gender differences in a probability sample.

Table 1 presents the final scale items for the abridged BIAS-m and BIAS-f measures. These abridged scale items were compiled for the 2015 wave of NSSHB in order to collect data from a nationally representative sample of self-identified heterosexual, gay, lesbian, and other participants. The Institutional Review Board of the Human Subjects Office at Indiana University—Bloomington reviewed and approved all study protocols.

Table 1. Abridged Bisexualities: Indiana Attitudes Scale (BIAS) items.

A. Abridged BIAS-f [7 point Likert response scale SA-SD]

Please respond to the next series of items in terms of the extent that you agree with the following statements about bisexual women (i.e., those with the capacity for physical, romantic, and/or sexual attraction to more than one sex or gender).

1. I think bisexual women are confused about their sexuality.

2. People should be afraid to have sex with bisexual women because of HIV/STD risks.

3. Bisexual women are incapable of being faithful in a relationship.

4. Bisexual women would have sex with just about anyone.

5. I think bisexuality is just a phase for women.

B. Abridged BIAS-m [7 point Likert response scale SA-SD]

Please respond to the next series of items in terms of the extent that you agree with the following statements about bisexual men (i.e., those with the capacity for physical, romantic, and/or sexual attraction to more than one sex or gender).

1. I think bisexual men are confused about their sexuality.

2. People should be afraid to have sex with bisexual men because of HIV/STD risks.

3. Bisexual men are incapable of being faithful in a relationship.

4. Bisexual men would have sex with just about anyone.

5. I think bisexuality is just a phase for men.

For measures of sexual identity, participants responded to the following item:

Which of the following commonly used terms best describes your sexual orientation?

1. Straight/heterosexual

2. Gay, lesbian, or homosexual

3. Bisexual

4. Asexual (I am not sexually attracted to others)

5. Other, please describe [textbox]

In addition to the widely used identity labels of heterosexual, gay/lesbian, and bisexual, the NSSHB offers participants the opportunity to indicate and briefly describe the other identities; in previous probability samples of U.S. adults, the other identity is endorsed by approximately 1% of male and 2% of female respondents [42]. As an increasing number of other identified NSSHB participants have described their orientation as asexual, we have offered this term as a separate option in recent waves of data collection. We were particularly interested in examining attitudes among other-identified individuals because, just as diverse individuals are problematically collapsed together under the auspices of “sexual minorities,” a number of non-monosexual identities (e.g., bisexual, pansexual, queer) are sometimes subsumed under the label of “bisexual” [43, 44] and experiences of anti-bisexual prejudice have been found to differ based on the label endorsed by the individual [45].

As anticipated, based on previous waves of NSSHB data collection, relatively small numbers of participants self-identified as asexual (N = 21) and other (N = 15). Among the participants who identified as other, participants’ self-reported identities included pansexual, demisexual, gray-asexual, and “I do not identify with any sexual orientation.” A very small number (N = 3) participants did not provide a textbox response for the other item. As the small numbers of asexual and other identified individuals did not provide sufficient statistical power for advanced analyses, we combined these individuals into an aggregate “other” category.

Statistical Analysis

Differences in sociodemographic characteristics between participants who responded to BIAS-m and/or BIAS-f were presented descriptively. For an overall measurement of attitudes towards bisexual men and women, we summed the responses for all participants who completed each BIAS scale, subtracting by the total number of items answered. BIAS-m and BIAS-f were also summed to generate a total BIAS score among participants who completed both scales. Due to the right-skewness of the scale response distributions, generalized linear modeling with gamma distributions (log link) were utilized to assess the association between sociodemographic characteristics and BIAS scores. For each of the three BIAS scores, pairwise comparisons by sexual orientation and race/ethnicity, respectively, using t-tests for weighted survey data were conducted. To test for differences between negative attitudes towards bisexual men versus bisexual women, weighted paired t-tests and Wilcoxon signed rank tests were conducted on BIAS-m and BIAS-f scores. All analyses were performed using Stata version 13 (StataCorp. 2013. Stata Statistical Software: Release 13. College Station, TX: StataCorp LP).

Results

Participants

Table 2 provides an overview of the demographic characteristics reported by participants who completed the BIAS-m and the BIAS-f measurements. Again, self-identified bisexual participants were intentionally excluded from these analyses since the study focused on the attitudes of non-bisexual individuals about bisexual persons. For the male version, most participants in our sample identified as heterosexual or straight (95%, N = 2,885), with a further 3.7% identifying as gay (N = 112) and a remaining 1.2% (N = 36) as other. Proportions were identical for the female version, with slightly more heterosexual participants completing the sub-scale. For the BIAS-m scale as a whole, Cronbach’s alpha was 0.911; for the BIAS-f scale as a whole, Cronbach’s alpha was 0.909.

Table 2. Sociodemographics of participants completing the Bisexualities: Indiana Attitude Scale (BIAS) targeting bisexual males (BIAS-m) and bisexual females (BIAS-f).

		BIAS-m (N = 3,033)	BIAS-f (N = 3,046)
Sociodemographics		n (%)	n (%)
Age
	18–24	348 (11.5)	348 (11.4)
	25–34	589 (19.4)	590 (19.4)
	35–44	443 (14.6)	443 (14.5)
	45–54	507 (16.7)	507 (16.6)
	55+	1,147 (37.8)	1,160 (38.1)
Gender
	Male	1,437 (47.5)	1,438 (47.3)
	Female	1,591 (52.5)	1,604 (52.7)
Race/Ethnicity
	White, non-Hispanic	1,983 (65.4)	1,985 (65.2)
	Black, non-Hispanic	358 (11.8)	363 (11.9)
	Other, non-Hispanic	199 (6.6)	199 (6.5)
	Hispanic	446 (14.7)	453 (14.9)
	Multiple races/ethnicities	47 (1.5)	47 (1.5)
Education
	Less than high school	342 (11.3)	349 (11.4)
	High school	917 (30.3)	917 (30.1)
	Some college	860 (28.4)	860 (28.2)
	Bachelor’s degree or higher	914 (30.1)	920 (30.2)
Household income
	<$25,000	522 (17.2)	522 (17.2)
	$25,000-$49,999	643 (21.2)	650 (21.3)
	$50,000-$74,999	529 (17.5)	529 (17.4)
	> = $75,000	1,339 (44.1)	1,346 (44.2)
Geographic region
	Northeast	558 (18.4)	560 (18.4)
	Midwest	649 (21.4)	649 (21.3)
	South	1,128 (37.2)	1,139 (37.4)
	West	698 (23.0)	698 (22.9)
Sexual orientation
	Heterosexual	2,885 (95.1)	2,898 (95.2)
	Gay/Lesbian	112 (3.7)	112 (3.7)
	Bisexual	-	-
	Other	36 (1.2)	36 (1.2)

Descriptive Findings

Table 3 provides an overview of the descriptive responses from all participants for each individual scale item on the BIAS-m and BIAS-f sub-scales. Regarding the response rate for the BIAS sub-scales, 97.0% responded to both scales, 2.5% did not respond to either scale, 0.02% responded to BIAS-m but not BIAS-f, and 0.5% responded to BIAS-f but not BIAS-m. In comparison to participants who responded consistently to both scales (either both responded or both non-response), the people who responded to one scale and not the other is extremely small (less than 1%). The location of the BIAS-f items before the BIAS-m items may account for the very slightly higher non-response rate of the BIAS-m items.

Table 3. Bisexualities: Indiana Attitudes Scale response distributions for BIAS-m and BIAS f sub-scales.

BIAS-m	Strongly disagree	Disagree	Somewhat disagree	Neither agree nor disagree	Somewhat agree	Agree	Strongly Agree
	%	%	%	%	%	%	%
I think bisexual men are confused about their sexuality	12.4	13.5	6.0	34.0	10.2	11.2	12.6
People should be afraid to have sex with bisexual men because of HIV/STD risks	11.5	9.2	6.1	29.4	15.3	13.2	15.0
Bisexual men are incapable of being faithful in a relationship	17.8	15.0	8.5	37.1	7.0	7.2	7.3
Bisexual men would have sex with just about anyone	17.5	14.9	8.1	34.5	8.5	7.2	9.2
I think bisexuality is just a phase for men	18.7	17.1	9.9	37.7	7.4	3.6	5.0
Bisexualities: Indiana Attitudes Scale-Female (5 Items) response distributions (N = 3,046)
BIAS-f	Strongly disagree	Disagree	Somewhat disagree	Neither agree nor disagree	Somewhat agree	Agree	Strongly Agree
	%	%	%	%	%	%	%
I think bisexual women are confused about their sexuality	12.5	14.4	5.4	35.7	10.4	11.0	10.6
People should be afraid to have sex with bisexual women because of HIV/STD risks	13.8	12.8	7.1	35.0	11.3	9.5	10.1
Bisexual women are incapable of being faithful in a relationship	18.5	17.3	7.5	38.0	7.2	5.1	6.0
Bisexual women would have sex with just about anyone	19.3	17.3	8.1	36.0	7.9	4.3	6.3
I think bisexuality is just a phase for women	17.3	17.2	10.0	38.1	8.7	3.7	4.4

Bisexualities: Indiana Attitudes Scale-Male (5 Items) response distributions (N = 3,033)

In terms of responses, for both male and female sub-scale items, participants were most likely to “neither agree nor disagree” with attitudinal statements (Table 3). For all items, the proportions of “neither agree nor disagree” responses were over one-third; indeed, nearly 40% neither agreed nor disagreed regarding the capability of bisexual men and/or bisexual women to be “faithful” in a relationship.

Difference between Overall Attitudes towards Bisexual Men versus Bisexual Women

Average BIAS scores for participants completing both scales were 13.64 (95% CI: 13.08–14.20) towards bisexual men and 12.80 (95% CI: 12.25–13.35) towards bisexual women. Negative attitudes towards bisexual men was significantly greater than negative attitudes towards bisexual women (p<0.001 by both weighted paired t-test and Wilcoxon signed rank test).

Factors Influencing Attitudes toward Bisexual Men and Women

Table 4 presents the relationships between specific demographic characteristics and mean BIAS scores. Age emerged as a related factor with attitudes toward bisexual men and women. Participants under the age of 25 years old reported significantly lower BIAS scores; thus, younger participants reported more positive attitudes (p<0.001). Gender was also a significant factor, such that, relative to men in the sample, women reported significantly lower scores on the BIAS-f. There were not; however, any significant gender differences between participants on the BIAS-m.

Table 4. The relationship between sociodemographic characteristics and BIAS scores towards bisexual males (BIAS-m) and females (BIAS-f).

		BIAS-m			BIAS-f			Total BIAS
Variables		β	SE	p	β	SE	p	β	SE	p
Intercept		2.61	0.03	<0.001	2.47	0.03	<0.001	3.23	0.03	<0.001
Age < 25 years		-0.22	0.04	<0.001	-0.16	0.04	<0.001	-0.19	0.04	<0.001
Gender
	Male (ref)	0.00	--	--	0.00	--	--	0.00	--	--
	Female	-0.11	0.03	0.101	-0.05	0.03	0.100	-0.08	0.03	0.005
Race/Ethnicity
	White	0.00	--	--	0.00	--	--	0.00	--	--
	Black	0.15	0.05	0.002	0.13	0.05	0.006	0.14	0.05	0.004
	Other	0.02	0.06	0.700	0.09	0.05	0.079	0.05	0.05	0.314
	Hispanic	0.07	0.04	0.089	0.11	0.04	0.011	0.08	0.04	0.049
	2+ races	0.09	0.08	0.240	0.11	0.07	0.102	0.10	0.07	0.162
High school or less		0.10	0.02	<0.001	0.15	0.02	<0.001	0.12	0.02	<0.001
Income < $25,000		0.18	0.04	<0.001	0.20	0.04	<0.001	0.19	0.04	<0.001
Sexual orientation
	Heterosexual (ref)	0.00	--	--	0.00	--	--	0.00	--	--
	Gay/Lesbian	-0.59	0.13	<0.001	-0.57	0.18	0.002	-0.57	0.15	<0.001
	Other	-0.66	0.13	<0.001	-0.62	0.12	<0.001	-0.65	0.12	<0.001

Race/ethnicity was significantly related to the participants’ attitudes toward bisexual individuals. In terms of the total BIAS measure, White/non-Hispanic participants reported the lowest mean scores (i.e., most positive scores), while Black/non-Hispanic participants reported the highest score (i.e., most negative scores). Similar patterns were found with the male and female sub-scales. As seen in Table 5, significant differences were found between White/non-Hispanic and Black/non-Hispanic participants (p = 0.003), as well as Other/non-Hispanic and Black/non-Hispanic participants. (p = 0.006).

Table 5. Estimated mean differences in BIAS scores by race/ethnicity.

	White, non-Hispanic	Black, non-Hispanic	Other, non-Hispanic	Hispanic	2+ Races, non-Hispanic	Mean difference	SE	p
BIAS-m score	13.28	15.56	--	--	--	-2.28	0.74	0.003*
	13.28	--	13.03	--	--	0.25	0.73	0.738
	13.28	--	--	13.94	--	-0.67	0.49	0.175
	13.28	--	--	--	14.45	-1.18	1.06	0.272
	--	15.56	13.03	--	--	2.53	0.88	0.006*
	--	15.56	--	13.94	--	1.62	0.71	0.028*
	--	15.56	--	--	14.45	1.11	1.23	0.371
	--	--	13.03	13.94	--	-0.91	0.62	0.145
	--	--	13.03	--	14.45	-1.42	1.34	0.295
	--	--	--	13.94	14.45	-0.51	1.19	0.671
BIAS-f score	12.28	14.45	--	--	--	-2.18	0.74	0.005*
	12.28	--	12.62	--	--	-0.35	0.56	0.535
	12.28	--	--	13.50	--	-1.22	0.50	0.019*
	12.28	--	--	--	14.04	-1.77	0.94	0.065
	--	14.45	12.62	--	--	1.83	0.83	0.032*
	--	14.45	--	13.50	--	0.95	0.72	0.193
	--	14.45	--	--	14.04	0.41	1.21	0.736
	--	--	12.62	13.50	--	-0.88	0.59	0.141
	--	--	12.62	--	14.04	-1.42	1.11	0.208
	--	--	--	13.50	14.04	-0.54	1.10	0.623
Total BIAS score	25.53	29.82	--	--	--	-4.30	1.46	0.005*
	25.53	--	25.65	--	--	-0.12	1.26	0.922
	25.53	--	--	27.23	--	-1.70	0.99	0.094
	25.53	--	--	--	28.50	-2.97	1.97	0.138
	--	29.82	25.65	--	--	4.17	1.67	0.016*
	--	29.82	--	27.23	--	2.60	1.46	0.081
	--	29.82	--	--	28.50	1.33	2.42	0.586
	--	--	25.65	27.23	--	-1.57	1.18	0.189
	--	--	25.65	--	28.50	-2.84	2.42	0.246
	--	--	--	27.23	28.50	-1.27	2.26	0.577

* p<0.05

Income was significantly related to participants’ attitudes. In particular, participants who earned less than $25,000 annually reported significantly higher BIAS mean scores. Thus, higher income participants were more likely to report more positive attitudes toward bisexual men and women. Similarly, educational attainment was significantly associated with participants' attitudes. Participants whose educational attainment was a high school diploma or less reported significantly higher BIAS scores. Thus, participants with higher levels of educational attainment were more likely to report more positive attitudes toward bisexual men and women.

Last, sexual identity was also significantly related to participants’ attitudes. Specifically, other-identified participants reported the lowest BIAS scores, followed by gay/lesbian participants and lastly heterosexual participants. Table 6 demonstrates relationships among various sexual identity categories for both BIAS-m and BIAS-f measures. Overall, sexual minority individuals (other and gay/lesbian) reported significantly more positive attitudes toward bisexual men and women than heterosexual participants. This was finding was consistent for both subscales on attitudes toward bisexual men and women.

Table 6. Estimated mean differences in BIAS scores by sexual orientation.

	Heterosexual	Gay/Lesbian	Other	Mean difference	SE	P
BIAS-m score	13.94	8.04	--	5.91	0.96	<0.001*
	13.94	--	7.05	6.90	0.95	<0.001*
	--	8.04	7.05	0.99	1.48	0.507
BIAS-f score	13.05	7.40	--	5.65	1.18	<0.001*
	13.05	--	6.93	6.11	0.93	<0.001*
	--	7.40	6.93	0.47	1.63	0.775
Total BIAS score	26.92	15.43	--	11.49	2.13	<0.001*
	26.92	--	13.90	13.03	1.84	<0.001*
	--	15.43	13.90	1.54	3.08	0.619

* p<0.05

Discussion

Our findings are among the first to document attitudes toward bisexual men and women among a probability sample of adults in the general population of the U.S. Our data from the 2015 NSSHB provide a snapshot of attitudes toward bisexual men and women in the U.S. that is both nationally representative and inclusive not only of self-identified heterosexual individuals but also, importantly, among individuals of diverse sexual identities. While a small number of studies have examined attitudes toward bisexual individuals (or bisexuality) in a range of convenience samples, our paper is only the second that explores such attitudes in a nationally representative sample, and is the first to do so in a sample of gay, lesbian, and other-identified individuals. In the sole prior probability study of heterosexual individuals’ attitudes toward bisexual men and women, self-reported attitudes (toward bisexual men, in particular) were extremely negative [38]. As such, our findings represent a “shift” in attitudes toward bisexual men and women from negative to more neutral in the general population. Further nationally representative data on attitudes toward bisexual men and women are needed in order to provide a clearer sense of change over time in such attitudes.

Positive attitudes toward gay and lesbian individuals among heterosexual individuals have dramatically increased over the past decade [31], both in the U.S. and around the world [32]. In the U.S., every demographic group has seen an increase in acceptance of gay and lesbian individuals including religious groups, all political parties, and young people [46]. These rates have remained constant almost a year after the Supreme Court ruling Obergefell v. Hodges [46]. However, in terms of attitudes toward bisexual men and women, the majority of participants in our study remained “middle of the road” and were most likely to report neither agreeing nor disagreeing with the range of sentiments and stereotypes embedded within the BIAS items. These responses can be interpreted in a variety of ways, including the participant not knowing, the participant being unsure, or the participant not caring [47]. Additionally, mid-point responses may also reflect social desirability bias [48]. While it is encouraging that most participants did not report more explicitly negative attitudes, that over one-third of participants neither agreed nor disagreed with every scale item may also be indicative of a general ambivalence toward bisexual men and women within the general population [49]. This may be reflective of larger cultural shifts away from expressing explicitly negative or discriminatory attitudes toward minority groups, while implicit or unconscious biases may still operate [50]. This may also be partially due to the relative invisibility of bisexual individuals and positive bisexual role models, relative even to increasing portrayals of gay men and lesbians in media and social consciousness [3, 28]. It may also reflect respondents’ unfamiliarity with bisexual people as bisexual men and women are significantly less likely than gay and lesbian individuals to disclose their identities and behaviors to friends and family members [51, 52].

In terms of specific scale items, the most commonly endorsed belief was that bisexual men and women are at a heightened risk for HIV/STI. This finding is in line with prior assessments of attitudes toward bisexual men and women in convenience samples in terms of “sexual riskiness” [1], as well as descriptions of “sexual irresponsibility” as a commonly reported anti-bisexual experience among bisexual individuals [5]. This sentiment is likely an artifact from the early days of the HIV epidemic, in which bisexual men, in particular, were stereotyped as a “bridge population” between their male partners and their presumably unaware female partners [53, 54]. [1, 5] However, recent research has demonstrated that, in the U.S., the number of bisexual men who are HIV-positive is likely no greater than the number of heterosexual men who are HIV-positive [55]; that bisexual men who are HIV-positive report lower proportions of female sexual partners, and lower rates of sexual intercourse with women, than bisexual men who are HIV-negative [56]; and that Black men who are gay-identified but behave bisexually are more likely to use HIV pre-exposure prophylaxis (PrEP) than Black gay men who do not report sex with women [52]. Taken together, these findings suggest that social discourse about bisexual men’s viral bridging behavior does not take into account current strategies that bisexual men have developed to minimize risk to themselves and their partners, whether male, female, or other. Additionally, while it is important to document the existence of health disparities, researchers and public health professionals are encouraged to be mindful of their role in perpetuating these stereotypes. Researchers are encouraged to engage with bisexual communities in both research design and dissemination in order to ensure that their research does not cause further unintentional damage to these communities [24].

Several findings on demographic factors associated with self-reported attitudes are parallel to results from prior studies focused on convenience samples and warrant further discussion [1, 33]. Gender was significantly associated with self-reported attitudes. Specifically, women were more likely to report more positive attitudes for bisexual men and women. These findings are consistent with the larger body of research on attitudes toward “LGB” populations, wherein women consistently demonstrate more positive attitudes toward sexual and gender minority groups than men [38, 57]. This may be because, overall, women tend to be less conservative and look more favorably upon social issues related to equality [58]. Similarly, attitudes toward sexual minority women are less negative than those towards sexual minority men. Worthen offers a number of explanations for this “gender gap”, including the sexualization of bisexual and lesbian women [59]. As it pertains to our findings, we would also posit that society constantly monitors women’s sexual behavior and uses that behavior to define and control women (e.g., being labelled as a “slut” versus a “prude”) [60]. Being made aware of the damage that this constant regulation may have on the way that they are able to experience their life may make women less prone to posing similar regulations on the lives of others. Additionally, more women identify as bisexual than men and women have also been thought to be more sexually fluid than men [14]. It is possible that women are less judgmental about bisexuality because many have considered their own bisexuality and/or sexual fluidity. Additionally, as men are traditionally expected to more rigidly conform to gender explicitly heteronormative norms and stereotypes, their attitudes may be markedly less “fluid” than women’s [61], though attitudes about stringent masculinity expectations are changing rapidly in youth populations in the U.S. and elsewhere [62].

Age also emerged as significant, specifically that lower age (under 25 years) was associated with more positive attitudes. These findings are consistent with recent qualitative cohort research on bisexual men, in which those in the 18–24 year old age cohort reported more acceptance from their heterosexual and gay and lesbian peers, than those in the older cohorts [63]. Contemporary youth are more likely to consider and define their sexuality and gender in ways that are less rigid than older generations. This is reflected both in national surveys that have found an increasing number of youth adopting non-monosexual identity labels (e.g., bisexual, pansexual, queer) and further highlighted by the number of young celebrities who have publicly claimed similar labels [64]. Youth may be more sensitive and resistant to stereotypes about those who adopt a bisexual label based on reflections of their own sexualities.

In relation to sexual identity, this may also explain why other-identified participants were more likely than others to report more positive attitudes toward bisexual men and women. Along the same lines as above, those who have questioned their identity and do not define within ‘traditional’ categories may relate to and be more sensitive about similar stereotypes. It is positive that gay/lesbian-identified individuals responded with less negative attitudes toward bisexual men and women than heterosexual participants. However, their attitudes were also less positive than other-identified individuals. This may be further evidence that nominally inclusive “LGB” support mechanisms may not be as relevant to bisexual individuals as they are to gay men and lesbian women [1]. Additionally, regarding sexual prejudice more broadly, oppressive structures operate through multiple levels, and to a large extent monosexism might be a fundamental source of oppressive attitudes faced by bisexual individuals. It is important to also consider that monosexism is dominated by heteronormative discourse thus, placing non-heterosexual forms of expression in antithesis to heteronormativity but also in conflict with each other.

In relation to race/ethnicity, Black/non-Hispanic participants were most likely to report more negative attitudes toward bisexual men and women. This is interesting given that, in many convenience samples of gay and bisexual men, Black men are significantly more likely to identify as bisexual than as gay [53, 65]. More negative attitudes among Black participants may also be the result of recent sensationalized negative media portrayals of Black behaviorally bisexual men “on the down low,” which has been alluded to in prior qualitative research [53]. In our pilot research, biracial/multiracial participants reported more positive attitudes toward bisexual men and women than their counterparts [1]. Null significance may be due to statistical power, due to the relatively smaller number of participants who report 2+ races/ethnicities in our nationally representative sample, and this issue may be explored in future research.

Both educational attainment and income level were associated with participants’ attitudes. Specifically, lower levels of income were reported with less positive attitudes toward bisexual men and women. This was also true of education, in that lower educational attainment was associated with less positive attitudes toward bisexual men and women. When away from home in an academic setting, students are often encouraged to question their beliefs, tolerate and accept diversity, and sometimes explore their sexuality. They may also be exposed to others whose sexual lives differ from their own. This exposure (whether in their lives or to the lives of others) may reduce judgement that they may have had about an otherwise unknown group [2].

Limitations

As with all social science research, this study is not without limitations. Future research would benefit from exploring attitudes toward bisexual transgender, non-binary, gender-queer, and other groups of individuals who may not identify as male or female. While some perhaps have not included these non-binary identities in previous studies on bisexuality, due to the assumption that the numbers of people who would report these identities is too low, it may be the case that we simply have not afforded individuals the option to report such identities consistently in research.

Additionally, as in prior waves of NSSHB, small numbers of individuals self-identified their sexual orientation as other. Unfortunately, we were not able to capture a large enough sub-sample size to explore the diversity among individuals who are captured under the other umbrella (e.g., similarities or differences among asexual, pansexual, and others who do not self-identify their sexual orientation with “traditional” labels, or who do not label themselves at all). An interesting finding that emerged, however, is that attitudes toward bisexual men and women among the aggregated other individuals were the most positive of all sexual identity sub-groups. Future research on sexuality-related attitudes and concerns among other-identified individuals is warranted.

Conclusions

Although it was not the aim of our study to directly examine previously established health disparities, the pervasive absence of positive attitudes toward bisexual men and women may be one of a number of other complex factors that continue to drive higher rates of adverse health outcomes among bisexual men and women. Researchers should continue to explore these unique health disparities in the context of social attitudes toward bisexual men and women in the U.S. After documenting the existence of an absence of positive attitudes toward bisexual men and women in the general U.S. population, we encourage future research to explore intervention opportunities focused on assessing, understanding, and eliminating biphobia—for example, among clinicians and other service providers—and determining how health disparities among bisexual men and women can be alleviated.

Data Availability

The minimal data set underlying the findings is available in the openICPSR (Inter-University Consortium of Political and Social Science Research) data repository (National Survey of Sexual Health & Behavior: Bisexualities Indiana Attitudes Scale) [http://doi.org/10.3886/E100265V1].

Funding Statement

Funding for the National Survey of Sexual Health & Behavior (NSSHB) is provided by Church & Dwight, Co., Inc. (Debby Herbenick, Principal Investigator). Additional funding for data collection using the BIAS was awarded by Indiana University School of Public Health-Bloomington (Brian Dodge, Principal Investigator). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.