Differences in Sexual Identity Dimensions Between Bisexual and Other Sexual Minority Individuals: Implications for Minority Stress and Mental Health

Abstract

Bisexual individuals experience poorer mental health than other sexual minority individuals. One explanation for this is that biphobia predisposes bisexual individuals to have a more ambiguous sexual identity and fewer opportunities for stress-ameliorating forms of coping and support. This study explores sexual identity and sexual identity dimensions—prominence, valence, integration, and complexity—in bisexual and other sexual minority individuals. We describe differences in sexual identity dimensions between bisexual and other sexual minority individuals and test two explanations for mental health disparities between them: whether sexual identity dimensions directly impact mental health and whether they moderate the impact of stress on mental health.
Data came from a longitudinal study of a diverse sample of sexual minority individuals (N = 396, 71 bisexual respondents) sampled from community venues in New York City. Sexual identity was prominent for both bisexual and other sexual minority individuals, but bisexual individuals reported lower valence and integration of sexual identity in their identity structures. The hypothesis that sexual identity dimensions moderate the impact of minority stress on mental health was not supported. Following several longitudinal assessments, however, we concluded that identity valence (but not integration or complexity) and depressive symptoms were bidirectionally associated so that differences in valence between bisexual and other sexual minority individuals explained, in part, disparities in depressive symptoms.

Full text

Differences in Sexual Identity Dimensions between Bisexual and
Other Sexual Minority Individuals: Implications for Minority
Stress and Mental Health
Chaïm la Roi1, Ilan H. Meyer2, and David M. Frost3
1Department of Sociology and Interuniversity Center for Social Science Theory and Methodology
(ICS), University of Groningen, Grote Rozenstraat 31, 9712 TG, Groningen, The Netherlands.
Electronic c.la.roi@rug.nl. Telephone: +31503638938
2The Williams Institute, UCLA School of Law, University of California in Los Angeles, 385 Charles
E. Young Dr. East, Los Angeles, CA, 90095, United States of America. Electronic
meyer@law.ucla.edu. Telephone (department): (310) 825-4841
3Thomas Coram Research Unit, UCL Institute of Education, University College London, 27-28
Woburn Square, Bloomsbury, London WC1H 0AA, United Kingdom. Electronic d.frost@ucl.ac.uk.
Telephone (department): +44 (0)20 7612 6957.
Abstract
Bisexual individuals experience poorer mental health than other sexual minority individuals. One
explanation for this is that biphobia predisposes bisexual individuals to have a more ambiguous
sexual identity and fewer opportunities for stress-ameliorating forms of coping and support. This
study explores sexual identity and sexual identity dimensions—prominence, valence, integration,
and complexity—in bisexual and other sexual minority individuals. We describe differences in
sexual identity dimensions between bisexual and other sexual minority individuals and test two
explanations for mental health disparities between them: whether sexual identity dimensions
directly impact mental health and whether they moderate the impact of stress on mental health.
Data came from a longitudinal study of a diverse sample of sexual minority individuals (N = 396,
71 bisexual respondents) sampled from community venues in New York City. Sexual identity was
prominent for both bisexual and other sexual minority individuals, but bisexual individuals
reported lower valence and integration of sexual identity in their identity structures. The
hypothesis that sexual identity dimensions moderate the impact of minority stress on mental health
was not supported. Following several longitudinal assessments, however, we concluded that
identity valence (but not integration or complexity) and depressive symptoms were bi-
directionally associated so that differences in valence between bisexual and other sexual minority
individuals explained, in part, disparities in depressive symptoms.
Keywords
Sexual identity; Minority stress; Bisexuality; Mental health; Depressive symptoms
Correspondence to: Chaïm la Roi.
HHS Public Access
Author manuscript
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Research has consistently shown that sexual minority individuals experience worse mental
health outcomes than heterosexual individuals (Institute of Medicine, 2011). This research
has typically combined sexual minority respondents into one group, obscuring differences
within the group. Yet, studies that assessed bisexual individuals separately from other sexual
minority individuals have shown that bisexual individuals have a comparatively higher risk
for mental health problems such as suicidality (Marshal et al., 2011), depressive symptoms
(Ross et al., 2017), and other mood and anxiety disorders (Bostwick, Boyd, Hughes, &
McCabe, 2010). Minority stress theory explains health disparities as related to prejudice and
stigma, which predispose sexual minority individuals to excess stress that may lead to
adverse health (Meyer, 2003). Social identity and identity structures potentially impact
mental health, but studies on the relationship between sexual identity structures and mental
health among sexual minority individuals remain scarce (Kertzner, Meyer, Frost, & Stirratt,
2009; Major & O’Brien, 2005; Meyer, 2003).
We seek to add to the literature on identity and minority stress by assessing whether
differences in sexual identity explain disparities in mental health outcomes between bisexual
and other sexual minority individuals. This study is novel in assessing how bisexual
individuals differ from other sexual minority individuals on a range of sexual identity
dimensions and in analyzing how sexual identity dimensions are related to mental health. A
cross-sectional analysis of data from the same dataset used in this study found that identity
valence was related to mental health disparities between bisexual and other sexual minority
individuals (Kertzner et al., 2009). Here we examine two hypotheses for explaining mental
health disparities between bisexual and other sexual minority individuals. First, we assess
whether bisexual individuals differ from other sexual minority individuals in identity
dimensions, and if so, whether differences in identity dimensions explain observed mental
health disparities. Second, we test whether dimensions of sexual identity moderate the
impact of stress on mental health and well-being (Meyer, 2003, p. 678). For example, that a
minority stressor would have a greater impact on health outcomes when sexual identity is
prominent than when it is secondary to a person’s self-definition.
Characteristics of a Bisexual Identity
Research on the particularities of a bisexual identity can be classified in four broad themes
(Sarno & Wright, 2013). First, bisexual individuals may be subject to biphobia both from
outside and within the LGBT community (Roberts, Horne, & Hoyt, 2015; Rust, 2002).
Examples of biphobia include its devaluation by the attitude that bisexuality is not a valid
sexual identity and that it is a transitory stage in the development of a lesbian or gay identity
(Brewster & Moradi, 2010; Hequembourg & Brallier, 2009). Second, bisexual individuals
conceal their sexual identity more than lesbian women and gay men (Balsam & Mohr, 2007;
Durso & Meyer, 2013; Mohr, Jackson, & Sheets, 2017). Concealment is a minority stressor
that may increase risk for mental health problems (Meyer, 2003; Riggle, Rostosky, Black, &
Rosenkrantz, 2017). Third, in part because of biphobia within the LGBT community,
bisexual individuals are less connected with the community than other sexual minority
individuals (Balsam & Mohr, 2007), which may deprive them of resources and opportunities
for coping and social support (Kwon, 2013; Meyer, 2015). And fourth, bisexual individuals
may have a somewhat more ambivalent stance towards their sexual identity than other
la Roi et al. Page 2
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

sexual minority individuals, like lesbian women and gay men. They report higher levels of
internalized homonegativity (Cox, Berghe, Dewaele, & Vincke, 2010; Kuyper & Fokkema,
2011), identity confusion (defined as uncertainty about one’s sexual identity), lower levels
of identity centrality (defined as the importance of sexual identity for one’s overall sense of
self) (Balsam & Mohr, 2007; Dyar, Feinstein, & London, 2015), and greater variability in
their sexual identity over time (Dyar et al., 2015; Galupo, Ramirez, & Pulice-Farrow, 2017;
Weinberg, Williams, & Pryor, 1994).
Sexual Identity Dimensions
Sexual identity dimensions—for example, whether identity is viewed positively or
negatively by a sexual minority person—may be important for understanding the impact of
stress on health outcomes (Meyer, 2003). Sexual identities, like other social identities, have
multiple dimensions that are interconnected in identity hierarchies (Ashmore, Deaux, &
McLaughlin-Volpe, 2004; Roccas & Brewer, 2002; Rosenberg & Gara, 1985; Stirratt,
Meyer, Ouellette, & Gara, 2008). Four dimensions have been discussed, sometimes using
different terms: Identity prominence - also referred to as identity centrality or identity
importance - addresses the importance of a sexual identity within one’s identity hierarchy
(Ashmore et al., 2004). Identity valence—or private regard—refers to the extent to which an
identity is positively or negatively evaluated by an individual (Ashmore et al., 2004).
Identity integration represents the degree to which a person’s sexual identity shows close
relationships to other identities in the person’s identity hierarchy, or the extent to which
multiple identities are perceived to share the same characteristics. Identity complexity, lastly,
refers to the extent of cohesiveness versus differentiation within one’s self-concept, or the
extent to which multiple identities are conceived as a single convergent social identity
(Roccas & Brewer, 2002; Stirratt et al., 2008). Identity integration and complexity are
related yet distinct concepts: identity integration is concerned with the overlap in
characteristics between several social identities; identity complexity taps the extent to which
several identities converge to represent one higher-level social identity.
As discussed above, research has suggested that bisexual individuals are more ambivalent
and less positive about their sexual identity than other sexual minority individuals (in
particular lesbian women and gay men). Based on these observations, we test whether,
compared with other sexual minority individuals, the sexual identities of bisexual
individuals would be characterized by lower identity prominence, valence, and integration,
and higher identity complexity (H1).
Sexual Identity as a Predictor of Mental Health
We also examine the direct impact of sexual identity dimensions on mental health and assess
whether differences between bisexual and other sexual minority individuals in identity
dimensions might explain mental health disparities between the groups. Research provides
mixed results on the association between identity prominence and mental health. On the one
hand, it has been argued that strong identification with a minority group might have
detrimental consequences for the well-being of minority individuals, making minority
individuals prone to view themselves as victims of discrimination (Major & O’Brien, 2005).
la Roi et al. Page 3
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

On the other hand, models of sexual identity development describe identity centrality as a
positive component of the development of a sexual minority identity, for instance leading to
lower levels of internalized homonegativity (Mohr & Kendra, 2011).
In terms of identity valence, negatively valued identity has been associated with depression
and other negative mental health outcomes (Allen, Woolfolk, Gara, & Apter, 1996; Hughes,
Kiecolt, Keith, & Demo, 2015). Traditional models of gay identity development have
highlighted this component, showing that positive self-evaluations are signs of positive
identity development, conducive for mental health (Cass, 1984; Troiden, 1989). For
example, an earlier paper, using the same data as the data used in our study, reported a
positive cross-sectional association between identity valence and both psychological and
social well-being (Kertzner et al., 2009).
Traditional gay identity development models describe the association between identity
integration and mental health (Cass, 1984; Troiden, 1989). These models view the
integration of a sexual minority identity with a person’s other identities as the final and
optimal stage of sexual identity development, which, in turn, is related to high levels of self-
acceptance and emotional well-being (Meyer, 2003). Empirical studies supported this claim,
finding that high levels of identity integration were related with positive mental health for
sexual minority individuals (Levitt et al., 2016).
Using a similar rationale, identity complexity may reflect the presence of incongruent self-
concepts and may be associatied with negative mental health outcomes. Furthermore, the
maintenance of complex, diverse, and perhaps conflicting identities could create
interpersonal strains and increased demands for an individual’s time and attention (Simmel
& Wolff, 1964; Stirratt et al., 2008).
Modern critics of gay identity development models have pointed out that the models err by
describing identity development as a linear progression as well as for focusing on sexual
identity as a primary identity rather than considering sexual identity at the intersection of
other identities (Eliason & Schope, 2007). These critiques complicate what we can learn
from gay identity development models but do not contradict the ideas examined here: that a
positive and integrated identity, and a less complex identity structure, are associated with
positive mental health. Based on these writings, we expect identity valence and identity
integration to positively impact mental health and identity complexity to negatively impact
mental health.
We thus assess whether identitiy dimensions are associated with a bisexual identity so that
bisexual individuals will show a more negative sexual identity valence, less identity
integration, and greater complexity than other sexual minority individuals, which, in turn,
would explain mental health disparities between the groups. We hypothesize that lower
mental health in bisexual compared with other sexual minority individuals is mediated by
differences in identity valence, integration, and complexity (H2).
la Roi et al. Page 4
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Sexual Identity Dimensions as Moderators of the Impact of Stress on
Mental Health
In addition, it has been argued within the minority stress framework that qualities of sexual
identity dimensions might moderate the impact of minority stressors on mental health
(Meyer, 2003). Here, identity dimensions can be thought of as the context for the experience
of minority stress. For example, when a stressful experience occurs (e.g., anti-LGBT
violence), a person who views her or his sexual identity negatively may experience a sense
of dejection, identifying with the aggressor in rejecting her or his own sexual minority
identity, magnifying the impact of the assault. In contrast, a person who viewed her or his
identity positively may garner resilience and be better ready to initiate support, which, in
turn, would ameliorate the negative impact of the stressful experience (Meyer, 2003). In line
with this, research on ethnic minorities has shown that favorable evaluations of one’s
minority racial identity in the face of discrimination functions as a coping mechanism
against negative societal opinions of people’s social group (Ellemers & Haslam, 2012;
Hughes et al., 2015).
Another identity dimension, identity prominence, may work to increase vulnerability to
minority stressors. That is, minority stressors that threaten the person’s prominent sexual
minority identity may have a greater negative impact on his or her mental health as
compared with a person who suffers similar stress but for whom the minority identity is less
prominent (Thoits, 1991). For example, a lesser identity prominence might better allow the
person to cope with the minority stressor through disengagement (Major & O’Brien, 2005).
We hypothesize that sexual identity dimensions moderate the association between minority
stressors and mental health (H3).
Method
Sample and Procedure
Data came from project STRIDE, a study on the relationships between identity, stress, and
mental health in a diverse LGB sample from New York City. Participants were sampled
from venues selected to ensure a wide diversity of cultural, political, ethnic, and sexual
representations. Recruitment venues included bars, nonbar commercial establishments,
outdoor venues (e.g., parks), groups (e.g., Latin dance clubs), and events (e.g., gay pride).
During recruitment, quotas were maintained to ensure that no one venue type/time was
overrepresented by more than 3 participants in the final sample. At the venues, the research
workers engaged potential respondents in a brief conversation to explain the purpose of the
study and asked them to fill out a brief screening form that would determine eligibility for
participation in the study. Respondents were eligible if they were 18–59 years old, resided in
New York City for 2 years or more, self-identified as lesbian, gay, bisexual, or similar terms
(e.g., queer); were male or female (and their gender identity matched sex at birth); and
White, African American, or Latinx (or defined themselves using similar terms, e.g., Black).
Eligible respondents were contacted by an interviewer and invited to complete the in-person
research interview at the research office or at another convenient private space. Baseline
data were collected between February 2004 and January 2005 (N = 396) and wave 2 data
la Roi et al. Page 5
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

were collected 12 months later (N = 371), resulting in a 94% retention rate. Respondents
were compensated $80 for participation in the baseline interview and $60 for participation at
wave 2. Sample characteristics and differences between bisexual and other respondents are
displayed in Table 1. Bisexual respondents included more women, more race/ethnic
minorities, and they were less educated, more often in debt, yet less often unemployed.
Approval was granted by Western Institutional Review Board, 2003–2006, Protocol
number: 20030059 and Columbia University Medical Center, since May 31, 2006, Protocol
Number: IRB-AAAB9794.
Measures
Sexual orientation.
Bisexual identity status: was measured at wave 1 by asking respondents what best
described their sexual orientation. Individuals who indicated bisexual as their sexual identity
were categorized as bisexual here (n = 71). Participants who chose gay, lesbian, or other
labels indicating a sexual minority status (e.g., queer, homosexual) were included in a
general other sexual minority category (n = 325), which was used as reference to the
bisexual participants. The large majority of respondents in this reference category identified
as gay (n = 178), lesbian (n = 111), or homosexual (n = 16); 15 respondents identified as
queer, and 5 respondents used other terms to describe their sexual identity. Sexual identity
was measured again at wave 2. As a robustness check, we reran our analyses, this time
counting all respondents that identified as bisexual at either wave 1 or wave 2, as bisexual.
The results (available upon request) did not change from the results reported here.
Furthermore, we compared bisexual respondents to a broad reference category of other
sexual minorities, rather than separating the group by gender (e.g., lesbian women and gay
men). This was done because our goal was to assess how bisexual men and women compare
with others within the LGB population in terms of sexual identity. Whereas lesbian women
and gay men might differ from each other in various ways, both groups differ in similar
ways from bisexual individuals, in that both lesbian women and gay men are romantically
and sexually attracted to others of one gender, which is the contrast of interest in our study.
To assess this, we compared lesbian women and gay men in our sample on mental health
and sexual identity (results available upon request). Lesbian women and gay men did not
differ significantly in any of the sexual identity dimensions. They also did not differ much in
mental health, except that lesbian women had higher level of depressive symptoms. This
finding is in line with evidence that women experience more depression than men (e.g.,
Nolen-Hoeksema, 2001), and thus likely attributable to gender differences rather than to a
difference between having a lesbian or gay identity.
Identity.
Sexual identity dimensions: measured at both wave 1 and wave 2 were used in our
analyses. They were calculated on the basis of responses to an Assessment of Multiple
Identities (AMI). Participants reported up to 12 personal, relational, and collective identities
in response to the question, “Who am I?” (Kuhn & McPartland, 1954). Among these 12
identities, participants were asked to specify their gender, racial/ethnic, and sexual identities.
la Roi et al. Page 6
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

While participants were asked to enter an identity term describing these areas, they could
enter the specific identity term they preferred (e.g., homosexual, bisexual, queer). Following
the elicitation of identities, participants rated each identity on a set of 70 descriptive
attributes, which were derived from the five-factor model of personality (Costa & McCrae,
1992). The attribute list included terms such as talented, guilty, unhappy, attractive, and
dependable, and participants indicated whether each attribute does not apply (0), applies to
some extent (1), or applies to a great extent (2) to an identity.
Dimensions of sexual identity were derived from the above AMI inventory, by conducting a
Hierarchical Classes Analysis (HICLAS; De Boeck & Rosenberg, 1988). HICLAS software
(De Boeck & Rosenberg, 1988) was used to analyze the identities and identity attribute
ratings provided by the participants. Previous research suggests that the 70-item attribute list
is a valid alternative to the free-response format for rating identities to which the HICLAS-
software was originally applied (Gara, Woolfolk, & Allen, 2002). A free-response format
for rating identities would have taken hours to complete and thus was not optimal for
administration as part of this study (Rosenberg & Gara, 1985).
The software follows an iterative process of differentiation to identify clusters of identities
and their corresponding attributes. Each level of differentiation is termed a “Rank.” At the
lowest level of differentiation (Rank 1), all identities and attributes are combined into a
single, unified cluster. Each successive increase in Rank breaks the identities and attributes
into increasingly differentiated and hierarchical sets of clusters. The dimensions reported
here employ HICLAS results at Rank 4, which has previously demonstrated good
psychometric properties for the modeling of identity interrelationships (Allen et al., 1996;
Woolfolk, Gara, & Ambrose, 1999).
Prominence refers to the location of sexual orientation within the Rank 4 HICLAS model of
identity interrelationships. Sexual orientation could be positioned at different tiers within the
model, depending on the degree to which it is elaborated by attributes. Identities
characterized by a greater number of attributes were located at higher tiers within the
hierarchical model. The prominence of sexual orientation was coded 0 (indicating that the
target identity was dropped from the model) to 4 (indicating that the identity was at the
highest possible level within the model).
Valence was defined as the proportion of positive attributes from the attributes respondents
chose to describe their sexual identity.
Integration was defined as the degree to which a target identity showed overlap with other
identities within the identity model (that is, they shared precisely the same attributes).
Integration was defined as a proportion, calculated as the number of identities that showed
overlap with a target identity divided by the total number of identities respondents used to
describe themselves.
Global self-complexity was defined as the total number of identity and attribute clusters
within the identity model. Identity models may contain a highly complex structure with
many clusters, or a simpler structure with fewer clusters. Complexity could range from 1 (a
la Roi et al. Page 7
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

single identity/attribute cluster) to 70 (70 individual and separate attributes), but a Rank 4
HICLAS commonly produces 5 to 15 clusters (Woolfolk et al., 1999).
Minority stressors.
Outness to family: was measured on a 4-point scale: out to all, out to most, out to some, out
to none. We used outness to family (rather than friends, co-workers, or health workers) as a
proxy for stress related to concealment (Meyer, 2003).
Chronic strain: was measured using 28 items referring to sources of strain from 9 areas of
life: general or ambient problems, financial issues, work, relationships, parenting, family,
social life, residence, and health (Wheaton, 1999) (wave 1 α = .73; wave 2 α = .72). On a
scale of 1–3, respondents were asked to indicate whether statements such as “You’re trying
to take on too many things at once” were not true, somewhat true, or very true for them at
the time of the interview.
Everyday discrimination: was assessed using the Williams scale (1997). The 8-item scale
assessed how often these experiences occurred over respondents’ lifetimes on a 4-point scale
(1 = often through 4 = never): treated with less courtesy, less respect, and receiving poorer
service than others, as well as being threatened or harassed, called names or insulted. It
showed good reliability (wave 1 α = .84; wave 2 α = .85). We modified the scale to ask, for
each item, whether it was related to sexual orientation, gender, ethnicity, race, age, religion,
physical appearance, income level/social class, or some other form of discrimination. We
used the total number of instances of discrimination related to sexual orientation.
Internalized homophobia: was measured using an 8-item scale, adapted by Frost and
Meyer (2009) from Martin and Dean (1992). Items include statements such as “How often
have you wished you weren’t gay?” Response options ranged from 1 = often to 4 = never.
The scale had good reliability (wave 1 α = .86; wave 2 α = .85).
Stigma: was measured using a scale that assessed expectations of rejection and
discrimination based on one’s sexual minority identity (adapted from Link, 1987). Items
included for example, “Most people would willingly accept someone like me as a close
friend”. Participants responded to a 6-item measure that utilized a 4-point scale ranging from
(1) agree strongly to (4) disagree strongly. The measure was internally consistent (wave 1 α
= .88; wave 2 α = .88). Responses were coded so that higher scores reflected more stigma.
Prejudiced events.: This measure was adapted from the Structured Event Probe and
Narrative Rating scale (Dohrenwend, Raphael, Schwartz, Stueve, & Skodol, 1993).
Assessment at step 1 was conducted by the interviewer, who recorded narratives of each life
event that happened to the respondent. At step 2, each narrative was coded by two
independent raters on several dimensions (e.g., magnitude) including whether or not
prejudice was involved. To assess reliability, we evaluated the consistency between the
raters. Of all possible event ratings (N = 77,085), only 2% were discrepant, indicating a high
degree of reliability. Inconsistent ratings were evaluated in a consensus meeting among 3 or
more raters (for more on this measure, see Kman, Palmetto, & Frost, 2006). The variable
la Roi et al. Page 8
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

used in this study was calculated as the total number of life events that were classified as
involving prejudice.
Mental health indicators.—Both wave 1 and wave 2 mental health indicators were used
in the analyses.
Depressive symptoms: were measured using the Center for Epidemiological Studies
Depression scale (CES-D; Radloff, 1977). The CES-D is a 20-item self-report measure on
depressive symptoms experienced over a 1-week period prior to the interview. The scale
showed high internal consistency in this sample (wave 1 α = .92; wave 2 α = .91).
Psychological well-being: was measured using an 18-item self-administered assessment
(Ryff & Keyes, 1995). This measure assesses psychological well-being across six domains:
self-acceptance, positive relations with others, autonomy, environmental mastery, purpose in
life, and personal growth. The internal consistency of the scale was sufficient (wave 1 α = .
75; wave 2 α = .77).
Social well-being: was measured using a 15-item self-administered assessment (Keyes,
1998). The construct examines respondents’ perception of their social environment and
includes five dimensions of social well-being in this regard: social coherence, social
integration, acceptance, contribution, and actualization. Internal consistency of the scale was
sufficient (wave 1 α = .78; wave 2 α = .75).
Control variables.—We used the wave 1 measurements of our control variables.
Education was operationalized as a dichotomous variable that distinguished participants
who had obtained less than or equal to a high school diploma from the rest of the sample.
We created an unemployment category, defined as non-student individuals who were
seeking work. We assessed net worth by asking participants to calculate how much money
they would have or owe after converting all assets to cash and paying all debts (Conger et
al., 2002). Responses were then coded into a dichotomous net worth variable, with 1
indicating negative net worth. Furthermore, respondents were classified in three age
categories: 18–29 years of age, 30–44, and 45–59. These categories approximate periods of
post adolescent entry into and exploration of the LGB community; the subsequent and
greater assumption of social roles related to partnership, child-care responsibilities, work, or
community activities in young adulthood; and the deepening or broadening commitment of
these roles in midlife, particularly as they relate to the wellbeing of future generations.
Gender. Respondents were either male or female. No transgender respondents were included
in the sample. Ethnicity. Respondents were part of one of three ethnic groups: White/
Caucasian, Black/African American, or Hispanic/Latinx.
Analysis Plan
A schematic overview of our analyses can be found in Figure 1. First, we tested whether
bisexual respondents differed from other sexual minority respondents in terms of sexual
identity dimensions. To ascertain the stability of our findings, both wave 1 and wave 2
sexual identity dimensions were regressed on sexual identity status.
la Roi et al. Page 9
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Using path analysis, we consequently examined whether sexual identity dimensions had
direct associations with mental health (path 2a, Figure 1) estimating paths between each of
four sexual identity dimensions and each of three mental health outcomes. We then tested
whether a bisexual identity status was indirectly associated with mental health via sexual
identity dimensions (paths 1 and 2a, Figure 1). Inference regarding the statistical
significance of indirect effects was based on 95% bias-corrected confidence intervals (95%
BC CI) of the specific paths using non-parametric bootstrap analyses (Preacher & Hayes,
2008) with 4,000 bootstrapped resamples of size = n. Bootstrap analyses were preferred over
other methods for inference on the indirect effects, as bootstrapping is robust against non-
normality of the sampling distribution of the indirect effect (Montoya & Hayes, 2016). We
estimated both cross-sectional and longitudinal models. Baseline data was used in the cross-
sectional analyses. In the longitudinal models, we estimated the effect of baseline sexual
identity dimensions on mental health measured at wave 2, controlling for mental health
measured at wave 1. Models were estimated using Full Information Maximum Likelihood
(FIML) in order to compensate for missing data (Allison, 2003). Item missingness was low,
ranging between 0% (most wave 1 measures) and 7.6% (psychological well-being wave 2).
We allowed for correlated errors between sexual identity dimension and between mental
health measures, thereby acknowledging potential inter-correlation amongst sexual identity
dimensions and mental health measures respectively. In all analyses, we controlled for
education, unemployment, negative net worth, age, gender, and ethnicity.
We then tested associations between sexual identity and mental health. In path 2b (Figure 1),
we tested whether dimensions of sexual identity functioned as effect moderators of the
association between minority stressors and mental health by running OLS regressions. Both
cross-sectional (using baseline data) and longitudinal models were estimated. In the
longitudinal models, we regressed mental health at wave 2 on minority stressors at wave 2,
sexual identity dimensions at baseline, and the interaction between minority stressors and
sexual identity dimensions, controlling for mental health at baseline. Three mental health
indicators were analyzed as outcomes: social well-being, psychological well-being, and
depressive symptoms. This resulted in testing a total of 6 * 3 = 18 possible interaction
effects per sexual identity dimension. The large number of tests conducted necessitated the
usage of a multiple test procedure so that the False Discovery Rate (FDR) was controlled.
The FDR refers to the rate of significant findings that are actually null findings (Storey &
Tibshirani, 2003). To this end, p-values of the interaction effects were transformed into q-
values, using the method of Benjamini and Hochberg (1995). Q-values were calculated per
set of 18 parameter estimates of the interaction effects of each sexual identity dimension.
Only parameter estimates with a q<.05 were interpreted as significant, leading to an
expected value of 5% for the FDR (Benjamini & Hochberg, 1995; Storey & Tibshirani,
2003).
la Roi et al. Page 10
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Results
Minority Stress and Mental Health of Bisexual Individuals Compared with Other Sexual
Minority Individuals
Compared with other sexual minority respondents, fewer bisexual women and men were out
to all their family members, and they had higher levels of internalized homophobia (Table
1). Bisexual respondents experienced less discrimination and fewer prejudice events
compared with other respondents. At wave 1, bisexual respondents had lower social well-
being than other sexual minority respondents; at wave 2, they had lower social and
psychological well-being and more depressive symptoms than other sexual minority
respondents (Table 1).
Identity Dimensions in Bisexual Respondents
Table 1 also shows that, at both wave 1 and wave 2, bisexual and other sexual minority
respondents rated their sexual identities as important, with no differences in identity
prominence between the groups. However, in both waves, bisexual individuals had lower
identity valence, meaning they viewed their bisexual identity less positively than other
sexual minority individuals. Bisexual respondents also had lower identity integration,
showing that their bisexual identity was less integrated with their other personal and social
identities in comparison to the other sexual minority respondents in our study. At wave 2,
bisexual respondents also had higher identity structure complexity, meaning that they had
comparatively less integrated identity hierarchy structures, which consisted of more clusters
(similar differences in identity complexity were found at wave 1, yet these were not
statistically significant).
Direct Effect of Identity Dimensions on Mental Health
In cross-sectional and longitudinal models, we assessed whether identity valence and
identity integration were positively related to mental health, and whether self-complexity
was negatively related to mental health. Results of the cross-sectional path analysis are
depicted in Figure 2 (all path coefficients for both the cross-sectional and longitudinal path
analysis can be found in Appendix A). The model fitted the data well (χ2(20) = 17.4,p = .
625; RMSEA = .000, 90% CI [.000, .037]; CFI = 1.000). Results were consistent across the
three mental health outcomes: higher identity valence was related to better mental health,
whereas higher identity complexity was related to poorer mental health. The longitudinal
model (Figure 3) fitted the data well too (χ2(26) = 28.2, p = .349; RMSEA = .015, 90% CI [.
000, .043]; CFI = 0.998) but only the effect of identity valence on depressive symptoms was
significant.
Identity Dimensions as a Mediator of Mental Health disparities between Bisexual and Other
Sexual Minority Individuals
Bootstrap analyses on the cross-sectional path model showed indirect effects of bisexual
identity on social well-being (β = −.07, 95% BC CI [−0.14, −0.03]), psychological well-
being (β = −.09, 95% BC CI [−0.17, −0.03]), and depressive symptoms (β=.06, 95% BC CI
[0.02, 0.11]), running via identity valence. These results show that lower valence of sexual
identity explained, in part, bisexual individuals’ lower mental health (Figure 2).
la Roi et al. Page 11
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Longitudinal analyses showed a significant indirect path from bisexual identity status to
depressive symptoms via identity valence (β = .03, 95% BC Cl [0.003, 0.066]) but the
results for psychological and social well-being were not significant in the longitudinal
analysis.
To understand the findings regarding well-being we assessed variability in well-being over
the one year follow-up. Results from the longitudinal path analysis showed that both social
well-being and psychological well-being were relatively stable over time (stability
coefficients: β = .51, p< 01, 95% CI [.43, .60]; β = .61, p< 01, 95% CI [.53, .70],
respectively) compared with depressive symptoms (stability coefficient: β = .33, p< 01, 95%
CI [.23, .43]), leaving little variation in social and psychological well-being to be explained
by sexual identity dimensions.
To further probe the relationships between valence and depressive symptoms, we assessed
the possibility of reversed causation by estimating a cross-lagged model (Figure 4). Results
point to bi-directional effects: an effect of valence (at wave 1) on depressive symptoms at
wave 2 (β = −.11, p<.05, 95% CI [−.21, −.02]), but also an effect of depressive symptoms at
wave 1 on valence at wave 2(β = −.21, p<.01, 95% CI [.32, .49]). The path coefficients of
depressive symptoms on valence and valence on depressive symptoms were not significantly
different from each other (χ2(1) = 1.8, p =.18).
Interaction Effect of Identity Dimensions on the Relationships Between Stress and Mental
Health
We tested the hypothesis that sexual identity dimensions moderated the association between
minority stressors and mental health by assessing whether sexual identity dimensions
measured at baseline moderated any of 144 possible associations between minority stressors
and mental health at baseline (4 identity dimensions times 18 possible moderations) and at
wave 2 (4 identity dimensions times 18 possible moderations). Only one interaction was
statistically significant at q < .05 (identity complexity increased the impact of discriminatory
events on depressive symptoms at baseline). Therefore we conclude that our results do not
support the moderation hypothesis in this study.
Discussion
Consistent with literature on health disparities, we found that bisexual individuals had
poorer social and psychological well-being and more depressive symptoms than other sexual
minority individuals. We assessed how bisexual individuals differed from other sexual
minority individuals in terms of sexual identity dimensions and whether these differences
explained health disparities between these groups. We found that sexual identity was
prominent for both groups of respondents, but bisexual respondents reported lower valence
and integration. In turn, identity valence had a significant impact on depressive symptoms,
and differences in valence between bisexual and other sexual minority individuals partially
explained disparities in depressive symptoms between the groups. The hypothesis that
sexual identity dimensions moderate the impact of minority stress on mental health was not
supported.
la Roi et al. Page 12
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

The findings are significant for showing the impact of identity on depressive symptoms and
for beginning to explain disparities among sexual minority subgroups. The use of an
extensive and flexible identity measure, which allowed each respondent to use her or his
preferred identity term as a referent, is especially significant. Typically, studies of sexual
minority individuals, including measures of identity, make little distinctions among
subgroups, potentially missing important divergences (Balsam & Mohr, 2007; Dyar et al.,
2015).
That identity valence is the only dimension that proved robust in our various assessments is
interesting, if not surprising—identity valence has been discussed as an important aspect of
identity development for decades now (Cass, 1984)—but work on bisexual identity
development is lagging (Eliason & Schope, 2007). Our work does not inform about the
causes of lower identity valence and integration among bisexual individuals, but some
hypotheses can be offered. Structural constraints are significant and include greater stigma
related to bisexuality, fewer role models, and fewer resources in the LGBT community. All
of these may leave bisexual individuals less able than other sexual minority individuals to
counter social biphobia and adopt positive bisexual identities (Roberts et al., 2015; Rust,
2002).
Our findings underscore the importance of longitudinal analyses for assessing the
directionality of associations. Whereas our cross-sectional analyses suggested that both
identity complexity and valence were substantially related to mental health, only a bi-
directional association between valence and depressive symptoms was preserved when
analyzing these associations over time. This suggests that results from cross-sectional
research might overstate the impact of social identity on mental health.
This study is novel for being the first to investigate whether sexual identity moderated the
association between minority stress and mental health. Although this was suggested in the
theoretical paper that introduced the minority stress framework (Meyer, 2003), it had not yet
been empirically tested. As this is the first study to test this hypothesis, which was not
supported here, our results should be considered with some caution. Future researchers
ought to test this intriguing hypothesis, for example using measures of other types of
stressors than the mostly distal stressors we tested.
Limitations and Suggestions for Future Research
Several limitations are notable. First, the data was collected a little over ten years ago. To
the extent that bisexual identity is now more positively held by bisexual people, our findings
may be biased. However, we have no reason to believe that significant social changes in the
status of bisexuality and in how bisexual identity is viewed have occurred, as observers
continue to note that bisexual identity is stigmatized both within and outside the LGBT
community (Dodge et al., 2012), and bisexual individuals continue to evidence lower mental
health when compared with other sexual minority individuals. This is perhaps most clearly
portrayed in a recently published meta-analysis by Ross and colleagues (2017), who
summarized health disparities between bisexual and lesbian or gay individuals in studies
using data collected between the 1990’s and 2014. Health disparities between the bisexual
and lesbian/gay group in studies using collected data collected in 2010 or later appeared to
la Roi et al. Page 13
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

be at least as large as disparities reported in studies using data from before 2010 (see Ross et
al., 2017, Figure 2, authors’ calculations). Similarly, since our data were collected, there has
been an expansion of available labels for people with a non-binary sexual orientation (e.g.,
pansexual). The implications of our findings to other identity labels remain to be explored.
However, despite the expansion of labels, bisexuality remains an important category.
Furthermore, studies on individuals with plurisexual identities have shown that people with
bisexual and other labels often show many commonalities in terms of sexual identity
(Galupo et al., 2017).
A second limitation of this study is that, although we used a diverse sample of sexual
minority individuals, it was recruited from sources in the LGBT community in New York
City, raising concerns about generalizability. For example, connection to the LGBT
community was probably higher in this sample than in the general population of bisexual
individuals who are less connected to the LGBT community. This might have attenuated the
differences we could find between bisexual and other sexual minority individuals. Studies
that use probability sampling would be insightful to understanding how bisexual individuals
who are not connected with the LGBT community might differ.
Conclusions
Despite these limitations, this study is important for showing that bisexual individuals stand
apart as a distinct subgroup within the population of sexual minority individuals. Bisexual
Americans comprise about half of the sexual minority population (Gates, 2011), but they
receive little specialized attention in research and policy analysis. Researchers, policy
makers, and public health workers ought to pay greater attention to understanding and
addressing the issues that impact bisexual health. Further research could help in developing
services and support specific to bisexual individuals and assess what interventions may
improve bisexual identity valence. As our study suggests, improvement in valuation of
bisexual identity may lead to improved mental health among bisexual individuals.
Supplementary Material
Refer to Web version on PubMed Central for supplementary material.
Acknowledgements:
Data employed in this study received funding from the National Institutes of health, Grant Number
5R01MH066058.
References
Allen LA, Woolfolk RL, Gara M, & Apter JT (1996). Possible selves in major depression. Journal of
Nervous & Mental Disease, 184(12), 739–745. http://doi.org/10.1097/00005053-199612000-00004
Allison PD (2003). Missing data techniques for structural equation modeling. Journal of Abnormal
Psychology, 112(4), 545–557. http://doi.org/10.1037/0021-843X.112.4.545 [PubMed: 14674868]
Ashmore RD, Deaux K, & McLaughlin-Volpe T (2004). An organizing framework for collective
identity: Articulation and significance of multidimensionality. Psychological Bulletin, 130(1), 80–
114. http://doi.org/10.1037/0033-2909.130.1.80 [PubMed: 14717651]
la Roi et al. Page 14
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Balsam KF, & Mohr JJ (2007). Adaptation to sexual orientation stigma: A comparison of bisexual and
lesbian/gay adults. Journal of Counseling Psychology, 54(3), 306–319. http://doi.org/
10.1037/0022-0167.54.3.306
Benjamini Y, & Hochberg Y (1995). Controlling the false discovery rate: A practical and powerful
approach to multiple testing. Journal of the Royal Statistical Society, 57(1), 289–300. http://doi.org/
10.2307/2346101
Bostwick WB, Boyd CJ, Hughes TL, & McCabe SE (2010). Dimensions of sexual orientation and the
prevalence of mood and anxiety disorders in the United States. American Journal of Public Health,
100(3), 468–475. http://doi.org/10.2105/AJPH.2008.152942 [PubMed: 19696380]
Brewster ME, & Moradi B (2010). Perceived experiences of anti-bisexual prejudice: Instrument
development and evaluation. Journal of Counseling Psychology, 57(4), 451–468. http://doi.org/
10.1037/a0021116
Cass V (1984). Homosexual identity formation: Testing a theoretical model. Journal of Sex Research,
20(2), 143–167. http://doi.org/10.1080/00224498409551214
Conger RD, Wallace LE, Sun Y, Simons RL, McLoyd VC, & Brody GH (2002). Economic pressure in
African American families: A replication and extension of the family stress model. Developmental
Psychology, 38(2), 179–193. http://doi.org/10.1037/0012-1649.38.2.179 [PubMed: 11881755]
Costa PT,J, & McCrae RR (1992). Revised NE0 Personality Inventory (NEO-PI-R) and NE0 Five-
Factor Inventory (NEO-FFI) professional manual. Odessa, FL: Psychological Assessment
Resources.
Cox N, Berghe W, Dewaele A, & Vincke J (2010). Acculturation strategies and mental health in gay,
lesbian, and bisexual youth. Journal of Youth and Adolescence, 39(10), 1199–1210. http://doi.org/
10.1007/s10964-009-9435-7 [PubMed: 20689983]
De Boeck P, & Rosenberg S (1988). Hierarchical classes: Model and data analysis. Psychometrika,
53(3), 361–381. http://doi.org/10.1007/BF02294218
Dodge B, Schnarrs PW, Reece M, Martinez O, Goncalves G, Malebranche D, ... Fortenberry JD
(2012). Individual and social factors related to mental health concerns among bisexual men in the
midwestern United States. Journal of Bisexuality, 12(2), 223–245. http://doi.org/
10.1080/15299716.2012.674862 [PubMed: 22745591]
Dohrenwend BP, Raphael KG, Schwartz S, Stueve A, & Skodol A (1993). The structured event probe
and narrative rating method for measuring stressful life events. In Goldberger L & Breznitz S
(Eds.), Handbook of Stress: Theoretical and Clinical Aspects (2nd ed., pp. 174–199). New York:
Free Press.
Durso LE, & Meyer IH (2013). Patterns and predictors of disclosure of sexual orientation to healthcare
providers among lesbians, gay men, and bisexuals. Sexuality Research and Social Policy, 10(1),
35–42. http://doi.org/10.1007/sl3178-012-0105-2 [PubMed: 23463442]
Dyar C, Feinstein BA, & London B (2015). Mediators of differences between lesbians and bisexual
women in sexual identity and minority stress. Psychology of Sexual Orientation and Gender
Diversity, 2(1), 43–51. http://doi.org/10.1037/sgd0000090
Eliason MJ, & Schope R (2007). Shifting sands or solid foundation? Lesbian, gay, bisexual, and
transgender identity formation. In Meyer IH & Northridge ME (Eds.), The Health of Sexual
Minorities (pp. 3–26). Boston, MA: Springer US.
Ellemers N, & Haslam SA (2012). Social identity theory. In Van Lange PAM, Kruglanski AW, &
Higgins ET (Eds.), Handbook of Theories of Social Psychology (pp. 378–398). Los Angeles, CA:
Sage.
Frost DM, & Meyer IH (2009). Internalized homophobia and relationship quality among lesbians, gay
men, and bisexuals. Journal of Counseling Psychology, 56(1), 97–109. http://doi.org/10.1037/
a0012844 [PubMed: 20047016]
Galupo MP, Ramirez JL, & Pulice-Farrow L (2017). “Regardless of their gender”: Descriptions of
sexual identity among bisexual, pansexual, and queer identified individuals. Journal of
Bisexuality, 17(1), 108–124. http://doi.org/10.1080/15299716.2016.1228491
Gara MA, Woolfolk RL, & Allen LA (2002). Social cognitive complexity and depression: Cognitive
complexity moderates the correlation between depression self-ratings and global self-evaluation.
la Roi et al. Page 15
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

The Journal of Nervous and Mental Disease, 190(10), 670–676. http://doi.org/
10.1097/00005053-200210000-00003 [PubMed: 12409860]
Gates GJ (2011). How many people are lesbian, gay, bisexual, and transgender ? The Williams
Institute, (April), 1–8. http://doi.org/10.1080/13691058.2012.673639
Hequembourg AL, & Brallier SA (2009). An exploration of sexual minority stress across the lines of
gender and sexual identity. Journal of Homosexuality, 56(3), 273–298. http://doi.org/
10.1080/00918360902728517 [PubMed: 19319738]
Hughes M, Kiecolt KJ, Keith VM, & Demo DH (2015). Racial identity and well-being among African
Americans. Social Psychology Quarterly, 78(1), 25–48. http://doi.org/10.1177/0190272514554043
Institute of Medicine (Ed.). (2011). The health of lesbian, gay, bisexual, and transgender people:
Building a foundation for better understanding. Washington, DC: The National Academies Press.
Kertzner RM, Meyer IH, Frost DM, & Stirratt MJ (2009). Social and psychological well-being in
lesbians, gay men, and bisexuals: The effects of race, gender, age, and sexual identity. American
Journal of Orthopsychiatry, 79(4), 500–510. http://doi.org/10.1037/a0016848 [PubMed:
20099941]
Keyes CLM (1998). Social well-being. Social Psychology Quarterly, 61(2), 121. http://doi.org/
10.2307/2787065
Kman S, Palmetto N, & Frost D (2006). A rater’s guide to stressful life events. Unpublished
manuscript. Retrieved from http://www.columbia.edu/~im15/life-events/stressmanual.pdf
Kuhn MH, & McPartland TS (1954). An empirical investigation of self-attitudes. American
Sociological Review, 19(1), 68–76. http://doi.org/10.2307/2088175
Kuyper L, & Fokkema T (2011). Minority stress and mental health among Dutch LGBs: Examination
of differences between sex and sexual orientation. Journal of Counseling Psychology, 58(2), 222–
33. http://doi.org/10.1037/a0022688 [PubMed: 21401219]
Kwon P (2013). Resilience in lesbian, gay, and bisexual individuals. Personality and Social
Psychology Review, 17(4), 371–383. http://doi.org/10.1177/1088868313490248 [PubMed:
23904452]
Levitt HM, Horne SG, Herbitter C, Ippolito M, Reeves T, Baggett LR, ... Geiss M (2016). Resilience
in the face of sexual minority stress: “Choices” between authenticity and self-determination.
Journal of Gay & Lesbian Social Services, 28(1), 67–91. http://doi.org/
10.1080/10538720.2016.1126212
Link BG (1987). Understanding labeling effects in the area of mental disorders: An assessment of the
effects of expectations of rejection. American Sociological Review, 52(1), 96–112. http://doi.org/
10.2307/2095395
Major B, & O’Brien LT (2005). The social psychology of stigma. Annual Review of Psychology, 56,
393–421. http://doi.org/10.2307/3089335 [PubMed: 15709941]
Marshal MP, Dietz L, Friedman M, Stall R, Smith H, McGinley J, ... Brent D (2011). Suicidality and
depression disparities between sexual minority and heterosexual youth: A meta-analytic review.
Journal of Adolescent Health, 49(2), 115–123. http://doi.org/10.1016/j.jadohealth.2011.02.005
[PubMed: 21783042]
Martin J, & Dean L (1992). Summary of measures: Mental health effects of AIDS on at-risk
homosexual men. Unpublished manuscript. Columbia University; Mailman School of Public
Health.
Meyer IH (2003). Prejudice, social stress, and mental health in lesbian, gay, and bisexual populations:
Conceptual issues and research evidence. Psychological Bulletin, 129(5), 674–697. http://doi.org/
10.1037/0033-2909.129.5.674 [PubMed: 12956539]
Meyer IH (2015). Resilience in the study of minority stress and health of sexual and gender minorities.
Psychology of Sexual Orientation and Gender Diversity, 2(3), 209–213. http://doi.org/10.1037/
sgd0000132
Mohr JJ, Jackson SD, & Sheets RL (2017). Sexual orientation self-presentation among bisexual-
identified women and men: Patterns and predictors. Archives of Sexual Behavior, 46(5), 1465–
1479. http://doi.org/10.1007/s10508-016-0808-1 [PubMed: 27502348]
la Roi et al. Page 16
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Mohr JJ, & Kendra MS (2011). Revision and extension of a multidimensional measure of sexual
minority identity: The Lesbian, Gay, and Bisexual Identity Scale. Journal of Counseling
Psychology, 58(2), 234–245. http://doi.org/10.1037/a0022858 [PubMed: 21319899]
Montoya AK, & Hayes AF (2016). Two-condition within-participant statistical mediation analysis: A
path-analytic framework. Psychological Methods, 22(1), 6–27. http://doi.org/10.1037/met0000086
[PubMed: 27362267]
Nolen-Hoeksema S (2001). Gender differences in depression. Current Directions in Psychological
Science, 10(5), 173–176. http://doi.org/10.1111/1467-8721.00142
Preacher KJ, & Hayes AF (2008). Asymptotic and resampling strategies for assessing and comparing
indirect effects in multiple mediator models. Behavior Research Methods, 40(3), 879–891.
[PubMed: 18697684]
Radloff LS (1977). The CES-D scale: A self-report depression scale for research in the general
population. Applied Psychological Measurement, 1(3), 385–401. http://doi.org/
10.1177/014662167700100306
Riggle EDB, Rostosky SS, Black WW, & Rosenkrantz DE (2017). Outness, concealment, and
authenticity: Associations with LGB individuals’ psychological distress and well-being.
Psychology of Sexual Orientation and Gender Diversity, 4(1), 54–62. http://doi.org/10.1037/
sgd0000202
Roberts TS, Horne SG, & Hoyt WT (2015). Between a gay and a straight place: Bisexual individuals’
experiences with monosexism. Journal of Bisexuality, 15(4), 554–569. http://doi.org/
10.1080/15299716.2015.1111183
Roccas S, & Brewer MB (2002). Social identity complexity. Personality and Social Psychology
Review, 6(2), 88–106. http://doi.org/10.1207/S15327957PSPR0602_01
Rosenberg S, & Gara M (1985). The multiplicity of personal identity. Review of Personality and
Social Psychology, 6, 87–113.
Ross LE, Salway T, Tarasoff LA, MacKay JM, Hawkins BW, & Fehr CP (2017). Prevalence of
depression and anxiety among bisexual people compared to gay, lesbian, and heterosexual
individuals: A systematic review and meta-analysis. The Journal of Sex Research, 1–22. http://
doi.org/10.1080/00224499.2017.1387755
Rust PC (2002). Bisexuality: The state of the union. Annual Review of Sex Research, 13(1), 180–240.
[PubMed: 12836732]
Ryff CD, & Keyes CLM (1995). The structure of psychological well-being revisited. Journal of
Personality and Social Psychology, 69(4), 719–727. http://doi.org/10.1037/0022-3514.69.4.719
[PubMed: 7473027]
Sarno E, & Wright AJ (2013). Homonegative microaggressions and identity in bisexual men and
women. Journal of Bisexuality, 13(1), 63–81. http://doi.org/10.1080/15299716.2013.756677
Simmel G, & Wolff K (1964). The sociology of Georg Simmel. New York: Collier-Macmillan.
Stirratt MJ, Meyer IH, Ouellette SC, & Gara MA (2008). Measuring identity multiplicity and
intersectionality: Hierarchical Classes Analysis (HICLAS) of sexual, racial, and gender identities.
Self and Identity, 7(1), 89–111.
Storey JD, & Tibshirani R (2003). Statistical significance for genomewide studies. Proceedings of the
National Academy of Sciences of the United States of America, 100(16), 9440–5. http://doi.org/
10.1073/pnas.1530509100 [PubMed: 12883005]
Thoits PA (1991). On merging identity theory and stress research. Social Psychology Quarterly, 54(2),
101. http://doi.org/10.2307/2786929
Troiden RR (1989). The formation of homosexual identities. Journal of Homosexuality, 17(1–2), 43–
73. http://doi.org/10.1300/J082v17n01_02 [PubMed: 2668403]
Weinberg M, Williams C, & Pryor D (1994). Dual attraction: Understanding bisexuality. New York:
Oxford University Press.
Wheaton B (1999). The nature of stressors. In Horwitz A & Scheid T (Eds.), A handbook for the study
of mental health: Social contexts, theories, and systems (pp. 176–197). Cambridge, UK:
Cambridge University Press.
la Roi et al. Page 17
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Williams DR, Yan Yu Y, Jackson JS, & Anderson NB (1997). Racial differences in physical and
mental health: Socio-economic status, stress and discrimination. Journal of Health Psychology,
2(3), 335–351. http://doi.org/10.1177/135910539700200305 [PubMed: 22013026]
Woolfolk R, Gara M, & Ambrose T (1999). Self-complexity and the persistence of depression. Journal
of Nervous & Mental Disease, 187(7), 393–399.
la Roi et al. Page 18
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Public significance statement:
This study suggests that bisexual individuals differ from other sexual minority
individuals in notable ways. Bisexual individuals thought less positively about their
sexual orientation and their sexual orientation was less well integrated with other parts of
their identity. Furthermore, the fact that bisexual individuals thought less positive about
their sexual orientation than other sexual minority individuals partly explained mental
health disparities between these groups.
la Roi et al. Page 19
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Figure 1.
Schematic overview of analyses
la Roi et al. Page 20
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Figure 2. Summary of cross-sectional path analysis
Notes: 1. * p < 0.05, ** p < 0.01, two-sided.
2. Standardized effects. SE between parentheses. Significant effects in black, non-significant
effects in gray.
3. Additional controls: education, unemployment, negative net worth, age, gender, and
ethnicity.
4. Method = FIML, N = 396. χ2(20) = 17.4, p = .625; RMSEA = .000 , 90% CI [.000, .037];
CFI = 1.000.
la Roi et al. Page 21
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Figure 3. Summary of longitudinal path analysis
Notes: 1. * p < 0.05, ** p < 0.01, two-sided.
2. Standardized effects. SE between parentheses. Significant effects in black, non-significant
effects in gray
3. Additional controls: education, unemployment, negative net worth, age, gender, and
ethnicity
4. Method = FIML, N = 396. χ2(26) = 28.2, p = .349; RMSEA = .015, 90% CI [.000, .043];
CFI = 0.998.
la Roi et al. Page 22
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Figure 4. Cross-lagged path analysis valence and depressive symptoms
Notes: 1. * p < 0.05, ** p < 0.01, two-sided.
2. Standardized effects. SE between parentheses.
3. Controlled for effects of education, unemployment, negative net worth, age, gender,
ethnicity. and sexual orientation on Wave 1 and Wave 2 valence and depressive symptoms.
4. Method = FIML, N = 396. χ2(0) = 0; RMSEA = .000; CFI = 1.000 (fully identified
model).
la Roi et al. Page 23
Am J Orthopsychiatry. Author manuscript.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Author Manuscript Author Manuscript Author Manuscript Author Manuscript
la Roi et al. Page 24
Table 1.
Descriptive statistics by sexual orientation group
Mean (SD) / Proportion
Variable Bisexual (n = 71)
Other sexual
minority
(n = 325)
Covariates
High school education or less 31% 20% z = 2.09*
Negative net worth (wave 1) 64% 54% z = 1.55
Unemployed (wave 1) 15% 25% z = 2.32*
Age at wave 1: 18-29 52% 49% z = −0.79
30-44 46% 45%
45-59 01% 06%
Female 61% 48% z = 1.97*
White 20% 37% χ2(2) = −7.94*
Black/African American 42% 31%
Hispanic/Latinx 38% 32%
Minority stressors (wave 1)
Outness to family: out to none 24% 7% χ2(2) = −24.46**
Out to some 28% 21%
Out to most 20% 21%
Out to all 28% 51%
Chronic strain 1.47 (0.25) 1.43 (0.22) t = 1.41
Everyday discrimination 1.61 (2.02) 2.54 (2.04) t = −3.48**
Internalized homophobia 1.83 (0.78) 1.31 (0.47) t = 7.41**
Stigma 2.12 (0.77) 2.05 (0.77) t = 0.68
Number of prejudiced events 0.45 (0.71) 0.83 (1.07) t = −2.86*
Wave 1 mental health
Social well-being 4.47 (0.78) 4.85 (0.88) t = −3.28**
Psychological well-being 5.27 (0.75) 5.40 (0.77) t = −1.31
Depressive symptoms 0.82 (0.56) 0.69 (0.56) t = 1.75
Wave 2 mental health
Social well-being 4.46 (0.76) 4.91 (0.81) t = −4.06**
Psychological well-being 5.23 (0.86) 5.51 (0.77) t = −2.67**
Depressive symptoms 0.88 (0.58) 0.68 (0.52) t = 2.86**
Wave 1 sexual identity
Prominence 3.50 (0.93) 3.56 (0.77) t = −0.54
Valence 0.64 (0.23) 0.73 (0.19) t = −3.50**
Integration 0.22 (0.20) 0.29 (0.23) t = −2.36*
Global self-complexity 10.29 (3.21) 9.83 (3.55) t = 0.98
Wave 2 sexual identity
Am J Orthopsychiatry. Author manuscript.

Author Manuscript Author Manuscript Author Manuscript Author Manuscript
la Roi et al. Page 25
Mean (SD) / Proportion
Variable Bisexual (n = 71)
Other sexual
minority
(n = 325)
Prominence 3.58 (0.76) 3.46 (0.83) t = 1.09
Valence 0.69 (0.21) 0.78 (0.19) t = −3.45**
Integration 0.22 (0.20) 0.31 (0.25) t = −2.77**
Global self-complexity 10.88 (2.88) 9.42 (3.42) t = 3.24**
Notes:
1.
*p < 0.05
**p < 0.01, two-sided.
2. Group differences in sexual identity dimensions remained significant when controlling for education, unemployment, negative net worth, age,
gender, and ethnicity.
Am J Orthopsychiatry. Author manuscript.