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].

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.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

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

https://doi.org/10.1371/journal.pone.0164430.t001

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.

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.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

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

https://doi.org/10.1371/journal.pone.0164430.t002

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.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

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

https://doi.org/10.1371/journal.pone.0164430.t003

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.

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.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

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

https://doi.org/10.1371/journal.pone.0164430.t004

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).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

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

https://doi.org/10.1371/journal.pone.0164430.t005

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.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

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

https://doi.org/10.1371/journal.pone.0164430.t006

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].

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.

Acknowledgments

The study team would like to express their appreciation to the members of the BIAS expert review panel, whose insight was essential in modifying and abridging the sub-scales. We would also like to thank the editor and reviewers for thorough feedback and suggestions that significantly improved the quality of this paper.