Changes in sexual identity and family support: A cohort analysis

Abstract

Family support is critical for the well-being of sexual minority (SM) people. While research has identified that family support is not uniform across SM individuals with different sexual identities, it is unclear whether family support is impacted when individuals change sexual identity. This study explored associations between changes in sexual identity and trajectories of family support among U.S. SM individuals from three age cohorts, and whether these associations were moderated by cohort and directionality of identity change. Data come from a longitudinal study of a national probability sample of SM individuals. Latent Growth Curve Models were conducted to compare trajectories of family support of individuals who changed sexual identity to those who had consistent sexual identities. Multiple-group analysis examined whether these associations differed by cohort membership. Follow-up analyses examined how the direction of identity change influences the associations between changes in sexual identity and family support. Results showed that patterns of stability and change in sexual identity were not equally associated with trajectories of family support across cohort groups. Individuals in the middle cohort who changed sexual identities experienced less family support, especially when they changed from a monosexual to a plurisexual identity. Among individuals from the older cohort, sexual identity change was associated with levels of family support, but the effects were mixed depending on the direction of identity change. Changes in family support followed by changes in sexual identity vary based on age and/or cohort. More studies examining how identity change is associated with trajectories of social support are needed.

While extant research has generally conceptualized sexual identity as fixed, youths and adults often report changes in sexual identity over time (Rosario et al., 2006; Srivastava et al., 2022). Prior work has also described these changes in sexual identity as fluidity or mobility of sexual identity (Campbell et al., 2022; Katz-Wise et al., 2017; Ott et al., 2011). Data from a U.S. nationally representative sample indicated that about 6.0% of the population reported a change in sexual identity over four years of follow-up (Evans-Polce et al., 2023). Other research has documented that changes in sexual identity are more common among youth (approximately 6.0% to 30.0%; Srivastava et al., 2022), women (Everett, 2015; Srivastava et al., 2022) and sexual minorities (Dickson et al., 2013; Evans-Polce et al., 2023; Hu & Denier, 2023), and can occur more than once (Diamond, 2008; Feinstein et al., 2019; Lilly et al., 2024; Phillips et al., 2020). When studying changes in sexual identity, it is critical to distinguish sexual identity changes that reflect responses to natural changes in sexual attractions and behaviors from changes that may be a result of sexual orientation change efforts (also known as “conversion therapy”). Historically, sexual orientation change efforts aim to coercively change in sexual minority (SM) individuals the sexual orientation away from SM orientations. Multiple studies have shown that being subject to such efforts as a minor is associated with compromised mental health in adults (Blosnich et al., 2020; Green et al., 2020; Jones et al., 2022), and have been criticized by scholars and major professional organizations (AMA, 2022; APA, 2021; Fish & Russell, 2020).

Theories of sexual identity development initially hypothesized that individuals typically have a linear trajectory in which individuals first become aware of their sexual attractions, and after a period of confusion, individuals recognize and disclose their sexual identity to others (Cass, 1979). In this model of sexual identity development, individuals would usually have stable patterns of sexual attractions and identity. However, further research considered and empirically demonstrated that for some individuals, sexual identity was instead fluid (i.e., a capacity for changes in sexual identity; Campbell et al., 2021; Diamond, 2006; Lilly et al., 2024; Rosario et al., 2006). Following that perspective, sexual identity development could be a continuous process for some individuals, influenced by contextual factors, and could reflect either short or long-term changes (Diamond, 2008, 2016).

Scholars have suggested that individuals may experience a loss of support when they change sexual identities (Everett, 2015; Ott et al., 2013; Rosario et al., 2011). This hypothesis was originally driven by prior studies that revealed that youth often experience negative initial reactions from their family when disclosing their SM identity (D’Augelli et al., 1998, 2005). In other words, a sexual identity change from an assumed heterosexual identity to a SM identity is often not initially welcomed by family members. Such a negative initial reaction to a change from heterosexual to SM identity reflects the ingrained societal heteronormative expectations within families; when individuals deviate from this expectation, family members respond with rejection and marginalization. However, these studies were cross-sectional and could be subject to retrospective bias. In a related topic, in a recent longitudinal study, Real & Russell (2025b) found a similar pattern where SM youth who changed gender identities from cisgender to a transgender or gender-diverse identity perceived decreasing family acceptance over time. Although that study examined changes in gender identity, this evidence indicates that a change in identity can in fact lead to loss of support from family members. That is, when individuals adopt a more stigmatized identity, they may experience more marginalization and discrimination (Real et al., 2024), including from family. While there is some evidence that family may be less supportive when individuals change from a heterosexual identity to a sexual minority, there is a lack of literature examining patterns of family reaction beyond changes from heterosexual to sexual minority, which can be relatively common (Krueger et al., 2024).

Although SM people’s experiences are often treated as being similar across different subgroups, family members’ views on sexual identity may differ for SM people based on their identities. For example, a systematic review showed that plurisexual individuals tend to experience more discrimination from family than monosexual individuals (McGeough & Sterzing, 2018). What is missing from the literature is the consideration that individuals may change from a SM identity to another SM identity (e.g., changing from a plurisexual identity to a gay or lesbian identity). On one hand, one might argue that because these individuals already identify as a SM, they may not experience differences in minority stress when they change to another SM identity. On the other hand, these individuals may still have a heightened exposure to rejection, given that identity change itself confronts heteronormative expectations (Katz-Wise & Todd, 2022), and some identities may be more stigmatized than others (McGeough & Sterzing, 2018; Ross et al., 2018). Recent work found small differences in perceived social support and support network sizes across SM identities, such that plurisexual individuals perceived less social support than monosexual individuals, and participants who changed sexual identities presented levels of social support and support network sizes similar to consistently monosexual individuals (la Roi et al., 2022). However, since prior evidence has shown that the direction of identity change impacts family relationships (Real & Russell, 2025), it is likely that associations between sexual identity change and levels of support may differ depending on the direction of the change, rather than assuming that all changes in sexual identity are the same. Notably, there is a lack of longitudinal studies that empirically test whether reported changes in sexual identity are associated with changes in perceived family support. To address this gap in the literature, we tested whether patterns of stability and change of sexual identity are associated with trajectories of family support.

In addition to differences among SM people based on their identities, there may be differences among SM people that have to do with sociohistorical timing, aging and life stage, or cohort. The last few decades have seen substantial sociocultural changes in the awareness and acceptance of SM identities; thus, SM people who enter adulthood in changing sociohistorical times may have distinct experiences with societal support. The 1970s and 1980s saw the emergence of gay (and ultimately LGBT) rights movements; many SM people were not out, or were among the first generations of LGB people to come out to their families. In the 1990s, HIV dramatically changed the societal landscape for SM people, resulting in both more visibility but also more stigma for the SM community. In the 2000s, although SM people experienced growing visibility and acceptance, several U.S. states banned marriage for same-sex couples (Bishop et al., 2021). In the 2010s, however, SM identities gained extensive visibility and social acceptance, as evidenced in the Gallup polls (Gallup Inc., 2022): while 53% of the population believed that same-sex relationships were not morally acceptable in 2001, this rate decreased to 34% in 2015. Societal changes were also reflected in federal legal changes, as seen in the legalization of marriage for same-sex couples in 2015 (Ogolsky et al., 2019).

Sociocultural contexts can influence the way individuals perceive, accept, and report sexual identities (Charlton et al., 2016; Fish & Russell, 2022). More accepting societal environments may facilitate the exploration of different sexual identities as well as the acceptance of one’s own and others’ sexual orientations, attractions, and identities. That is, living as an SM individual in different societal contexts may shape social networks and interactions with peers and family. For example, as a form of dealing with the rejection from their families when they disclose their SM identity, many SM people develop chosen families – non-biologically related individuals who provide support (Dewaele et al., 2011) and many SM people include friends in their descriptions of families (Soler et al., 2018). Recent research showed that whereas middle-aged and older SM people tend to rely more on friends as a source of support for major issues, younger SM people rely more on family members (la Roi et al., 2022). Further, family members from older generations may resist accepting SM identities; even when considering that the overall social climate toward SM people has improved over the years, older individuals present the lowest rates of acceptance toward SM people (Pew Research Center, 2020). Thus, it is also possible that families from older generations may be even less supportive when family members report changes in sexual identity. Ultimately, the improvement in climates toward SM people may not have had the same implications for the family dynamics of middle-aged and older SM people; at the same time, such patterns may simply reflect age-related differences, since younger adults may be closer to family members to whom they may have stronger or more recent emotional or financial ties.

The Current Study

Although previous studies have documented that a non-negligible proportion of individuals change sexual identities, little is known about the impact of changes in sexual identity on family relationships, despite knowledge that positive family relationships are critical for the well-being of SM people (Real & Russell, 2025; Ryan et al., 2010; Watson et al., 2019). The proposed study leverages data from a unique, longitudinal study of SM people from three distinct cohorts who came of age in distinct sociohistorical contexts to examine whether changes in reports of sexual identities were associated with the final levels (i.e., intercept) and rates of change (i.e., slope) over time of family support (Paths A1-A2 in the conceptual model in Figure 1).

Figure 1. Conceptual Model.

Note: Latent Growth Curve Models examining the trajectory of family support and whether patterns of sexual identity change and stability predicts intercept and slope of family support. Models are controlled for age at Wave 1, sex at birth, gender identity at Wave 1, race/ethnicity, cohort, income level, education, and urbanicity and adjusted for sampling weights.

First, given theoretical assumptions that changes in sexual identity are associated with loss of family support (D’Augelli et al., 2005; Everett, 2015), we hypothesized that individuals who changed sexual identities during the study period would report declines and lower levels of family support at the end of the study. Follow-up analyses also tested whether the direction of identity change influences the links between sexual identity change and family support. In this regard, we expected that individuals who reported a sexual identity change toward a more stigmatized identity (i.e., changed from a monosexual to a plurisexual identity) would report declines and lower levels of family support at the end of the study. Second, given that exposure to different social climates may reflect differences in how individuals experience support from others after a sexual identity change (Fish & Russell, 2022), the next research question explored whether the association between sexual identity change and trajectories of family support differed (were moderated by) cohort membership (Path B1 in Figure 1). We anticipated that cohort membership moderated the relationship between sexual identity change and family support, such that individuals from the middle and older cohorts who changed sexual identity would experience steeper declines in family support and lower levels of family support at the end of the study compared to participants from the younger cohort. As noted, direction of changes in sexual identity may matter for family support: analyses also examine associations between direction of identity change and family support for different cohorts. These analyses were exploratory, given the lack of empirical evidence examining how the direction of identity changes across different age cohorts.

Methods

Participants and procedures

Data come from a three-wave U.S. national probability study of lesbian, gay, and bisexual people from three generations (The Generations Study) (Meyer et al., 2020). Participants were recruited based on the Gallup Daily Tracking Survey, a telephone interview conducted with a national probability sample of over 366,000 participants. In the initial survey, 3.5% (n = 12,837) identified as Lesbian, Gay, Bisexual, or Transgender. Since the original study aimed to examine the health of SM people, participants were invited for this study if they identified as a SM (transgender participants were screened to participate in the TransPop Study, a sibling study to the Generations study). Additional inclusion criteria were being aged within the range of three cohorts (18–25, “younger cohort”; 34–41, “middle cohort”; and 52–59 “older cohort”), spoke English, completed sixth grade, and were White, Black, or Latino, or multiracial if included one of the races mentioned. A total of 3,525 non-transgender SM people were eligible; those who agreed to participate had the surveys sent via mail or email. Participants were compensated with a $25 gift certificate. Enrollment for Wave 1 participation was between March 2016 and March 2017. Participants were invited to participate in two follow-up surveys one year apart from the previous wave of data collection.

The baseline survey was completed by 1518 participants. Follow-up assessments were completed by 894 participants at Wave 2 (67% retention) and by 707 participants at Wave 3 (53% retention); 615 participants (41%) participated in all three waves of data, and 369 participants (24%) had data for two waves. Because identity change is assessed by examining differences in sexual identity across waves, we only include participants with data in at least two waves (n=985). One participant was excluded from the present analysis because they did not identify as SM during the study period. Twenty-three participants who identified at Wave 1 with sexual identities other than monosexual and plurisexual (e.g., asexual) were also excluded from this study due to small cell sizes that would not allow for the examination of the moderating effects of cohort membership. The final sample consisted of 961 participants, of which 499 were assigned female at birth and 462 were assigned male at birth (the large majority were cisgender, 93.2%; 6.8% identified as non-binary) who were from a diversity of educational backgrounds (see Table 1). At baseline, 49.9% of participants were monosexual, and 50.1% were plurisexual. Ages ranged from 18 to 60 (mean 38.0 years); 56.2% were from the younger cohort, 22.3% from the middle cohort, and 21.5% from the older cohort. The majority of participants were from Urban locations (88.9%), White (69.7%), and had an income 300% above the federal poverty line (51.5%). Sampling weights provide population-representative estimates and compensate for selection probabilities and disproportionalities in nonresponse (Meyer et al., 2020). All study procedures were reviewed and approved by UCLA and Gallup Institutional Review Boards. Data and methodology are available at https://www.generations-study.com/methods. Detailed information about the Generations study can be found elsewhere (Meyer et al., 2020).

Table 1.

Sample descriptives.

	Full Sample (n=961)		Changed sexual identity (n = 92)		Plurisexual to monosexual (n=41)		Monosexual to Plurisexual (n=39)		Changed to other sexual identities (n=14)
	N (SD)	Weighted %	N (SD)	Weighted %	N (SD)	Weighted %	N (SD)	Weighted %	N (SD)	Weighted %
Sex assigned at birth
Female	499	56.8%	58	71.3%	24	67.6%	24	71.4%	10	82.4%
Male	462	43.2%	34	28.7%	17	32.4%	15	28.6%	2	17.6%
Gender identity at Wave 1
Woman	463	52.5%	52	64.4%	20	63.3%	23	63.7%	9	70.5%
Man	444	40.7%	31	26.7%	14	29.6%	15	26.6%	2	17.6%
Non-binary	54	6.8%	9	8.9%	5	7.2%	3	9.6%	1	11.8%
Sexual identity at Wave 1
Monosexual	561	49.9%	33	40.0%	4	74.7%	27	51.1%	2	24.8%
Plurisexual	400	50.1%	59	60.0%	37	25.3%	12	48.9%	10	75.2%
Age	38.0 (14.9)		33.3 (14.0)		31.9 (13.0)		33.8 (15.2)		36.3 (13.4)
Race-ethinicity
Non-Latino White	687	69.7%	60	65.1%	26	60.5%	25	67.1%	9	73.1%
Non-Latino Black	71	7.8%	7	6.5%	3	8.1%	4	6.8%	0	0.0%
Latino	76	8.6%	6	5.1%	3	4.2%	3	7.3%	0	0.0%
Multiracial	127	14.0%	19	23.3%	7	27.2%	9	18.7%	3	26.9%
Urbanicity
Non-urban	108	11.1%	14	13.1%	7	13.1%	6	13.1%	1	13.1%
Urban	853	88.9%	78	86.9%	32	86.9%	35	86.9%	11	86.9%
Education
High School or Less	164	37.9%	18	42.6%	10	44.1%	7	49.3%	1	14.2%
Some college	295	32.5%	34	35.9%	14	28.6%	14	33.9%	6	65.9%
College	291	17.8%	24	13.0%	8	16.5%	12	8.8%	4	16.6%
More than college	211	11.8%	16	8.6%	7	10.8%	8	8.0%	1	3.3%
Cohort
Younger (18–26 years)	387	56.2%	47	65.0%	21	63.0%	22	72.9%	4	43.5%
Middle (32–43 years)	230	22.3%	25	25.3%	8	29.8%	12	16.0%	5	43.9%
Older (52–60 years)	344	21.5%	20	9.7%	10	7.2%	7	11.1%	3	12.7%
Income Level
<100% FPL	101	14.4%	9	15.3%	4	12.5%	4	16.0%	1	21.3%
100–299% FPL	161	21.4%	19	18.5%	6	10.1%	7	18.8%	6	43.5%
200–299% FPL	120	12.6%	13	15.5%	5	22.0%	6	9.1%	2	16.8%
300% + FPL	563	51.5%	48	50.7%	23	55.4%	22	56.1%	3	18.4%
Religion
Agnostic or Atheist	411	46.3%	42	46.0%	19	43.2%	19	50.0%	4	40.4%
Reported a religion	536	53.7%	49	54.0%	21	56.8%	20	50.0%	8	59.6%

Note: NL = Non-Latino. Examples of other sexual identities include asexual, and heterosexual. FPL = Federal poverty line.

^*Row percentages based on the valid percent (excludes missing values).

Measures

Sexual identity stability and change.

At each wave, participants reported their sexual identity (“Which of the following best describes your current sexual orientation?”). Response options included: straight/heterosexual, lesbian, gay, bisexual, queer, same-gender loving, and other. Participants who reported “other” wrote in their sexual identity. Given that some identities had small numbers, we combined participants’ reported sexual identities into three categories: monosexual and same-sex attracted (e.g., gay, lesbian); plurisexual (e.g., bisexual, queer, pansexual); and other sexual identities (e.g., asexual, anti-label). As the main predictor for this study, a change in sexual identity was indicated when participants’ sexual identity did not match their sexual identity from the previous wave. Specifically, identity change was considered when participants changed from monosexual to plurisexual (or other identity), and from plurisexual to monosexual (or other identity). Based on the patterns of stability and change of sexual identities, and because it is known that family support differs between participants who are monosexual compared to plurisexual, four groups were created: participants who were consistently monosexual (n = 528), participants who were consistently plurisexual (n = 341), participants who changed at Wave 2 (n = 45), and participants who changed at Wave 3 (n = 47). For participants who changed sexual identities twice in the study (n = 17), the last identity change was considered (i.e., participants were allocated in the group who changed sexual identities at Wave 3). In the supplementary materials, Table S1 describes participants’ sexual identity at wave 1 and at the last wave of participation.

Family support.

The Multidimensional Scale of Perceived Social Supports (MSPSS) is a 12-item scale that assesses perceptions of social support from different sources (i.e., significant other, friends, and family). MSPSS items were introduced by the following sentence: “Please rate your level of agreement with the following items.” We used four items of the subscale assessing support from family. These item were: “I can talk about my problems with my family”, “My family tries to help me”, “My family is willing to make decisions”, and “I get the emotional help and support I need from my family”. Items were rated on a 7-point Likert scale from 0 (Strongly disagree) to 7 (Strongly agree). In waves which participants had data but had missing scale items, items were imputed by the Generations study team using a single imputation by chained equations and predictive mean matching. A mean score was calculated for each wave. The MSPSS scale has been validated among the SM population (Kler et al., 2023), and the family support subscale has strong reliability across waves (Wave 1 α=.92, Wave 2 α=.92, and Wave 3 α=.93).

Covariates.

Several demographic characteristics were used in the models as key control variables, including participants’ age, sex at birth (Male [reference group], Female), gender identity at Wave 1 (Man, Woman, Non-binary), race/ethnicity (Black, Latino, Multiracial, White [reference group]), cohort (younger [reference group], middle, older), income level (below 100% federal poverty line [FPL], 100–199% FPL, 200–299% FPL, and above 300% FPL), education (High school or less [reference group], Some college, College, More than college) and urbanicity (Non-urban [reference group], Urban). These variables were considered as control variables because of their association with differences in family support and relationships (Frost et al., 2016; Real & Russell, 2025; Swank et al., 2012; Swendener & Woodell, 2017). The study team imputed some covariates with missing values using a single imputation by chained equations (fully conditional specification), using predictive mean matching (Morris et al., 2014). Of note, sixteen participants had missing data in income level because they did not have enough data for such calculation. Moreover, because religiosity plays a role in family reactions to SM identities (Miller et al., 2020), we also included participants’ religiosity as a control variable. Religiosity was measured with one item: “What is your present religion, if any?” Responses were dichotomized between participants who reported any religion and participants who reported being either atheist or agnostic.

Attrition

We compared participants’ demographic characteristics and levels of family support at Wave 1 of those who participated in the follow-up waves to those who just participated in Wave 1, and those who had at least two waves of data to those who completed all waves of data. Results from these analyses are displayed in Tables S2–S4 in the Supplementary Materials. Given that missing data were linked to several covariates, missing data were considered missing at random (Little, 2024). Missing data were handled with full-information maximum likelihood (FIML); FIML uses all existing information without imputing missing data.

Analytic plan

Data were managed using Stata 18.0 (StataCorp, 2023) and analyzed using Mplus 8.1 (Muthén & Muthén, 2017) in a structural equation modeling framework. Trajectories of family support were examined using Latent Growth Curve Modeling (LGCM) using maximum likelihood estimation with robust standard errors (MLR), in which a latent intercept and a slope were estimated for all models. Given that sexual identity change is measured as occurring during the study period, the intercept was set at Wave 3 to indicate the final levels of the family support trajectory. The slope estimated the rate of change in family support across waves. Slope and intercept variances were freely estimated. Only a linear slope was considered because the Generations study only has three waves of data. Factor loadings indicated the years between waves (i.e., −2, −1, and 0; the negative signs represent assessments prior to the intercept, which is here considered as the growth curve end point). In order to test whether sexual identity change is associated with final levels or changes over time in perceived family support (intercept and slope), we first conducted an unconditional LGCM to estimate perceived family support means for the intercept and slope. Then, we ran a conditional LGCM to test whether participants who had a sexual identity change had a different slope and intercept of family support (paths A1-A2 in Figure 1) compared to participants who had consistent monosexual and plurisexual identities, controlling for demographic characteristics. Models examining the distinct effects of sexual identity change in Wave 2 and Wave 3 on the intercept and slope were tested to verify whether the effects were similar (i.e., constrained to be equal in both waves) or varied at each wave (i.e., estimated freely at each wave). Given prior evidence that the direction of identity change is associated with different outcomes (Everett, 2015; Krueger et al., 2024), follow-up models examined whether the direction of change contributes to differences in the associations between identity change and the trajectories of family support. Because participants who changed to sexual identities other than monosexual and plurisexual had cell sizes that were too small (e.g., participants who changed to an asexual, heterosexual, or other identity), these participants were excluded from such follow-up analyses (n = 12). When participants changed sexual identities twice, the last direction of sexual identity change was considered.

Next, we examine whether these associations are moderated by individuals’ cohort membership (path B1 in Figure 1) by conducting multiple group analysis. For this, a base model was estimated where model parameters were freely estimated across groups (i.e., younger, middle, older). Then, a more restricted model was tested where cohort groups were constrained to have an equal association between sexual identity change and family support. A chi-square difference test with Satorra-Bentler corrections compared the fit of the more restrictive model to the less restrictive model and examined the degree to which the two models were significantly different from each other. A worse model fit in the more restrictive model suggests that there were significant differences between cohort groups in terms of the association between sexual identity change and family support (i.e., moderation effects). When a worse model fit was found, we conducted a follow-up analysis to estimate whether cohort group differences were in their final levels (i.e., intercept) or in the rate of change (i.e., slope) of family support. Similarly to the first set of analyses, we also examined whether the associations between the direction of sexual identity change and the trajectories of family support were similar or different for the three generational groups.

For each model tested, comparisons were made using participants who were consistently monosexual and participants who were consistently plurisexual across waves as reference groups. All models control for age, sex at birth, gender identity at Wave 1, race/ethnicity, cohort, income level, education, religiosity and urbanicity and are adjusted using sampling weights to provide population-based estimates and to correct for nonresponse and attrition. Model fit was assessed with a root-mean-square error of approximation (RMSEA <.08 for a good fit), comparative fit index (CFI >.95 for a good fit, >.90 for acceptable fit), Tucker-Lewis Index (TLI) (good at >.95, acceptable at > .90) and chi-square test (p-value > .05 for a good fit) (Kline, 2023).

Results

Table 2 displays descriptive statistics for the key study variables. Across waves, participants’ family support scores were moderately correlated with each other (r_range = .56 - .70), and weighted averages ranged between 4.73 and 4.87, suggesting that, on average, participants perceived mid-levels of support from their families. Figure 2 depicts Wave 3 family support weighted means and 95% confidence intervals stratified by participants’ cohort membership and pattern of stability and change of sexual identities. Among participants who had at least two waves of data, 92 of 961 participants (10.8% weighted; when considering participants who had other sexual identities at baseline, the prevalence of sexual identity change was 12.5%) reported a different sexual identity at least once during the study period. Sexual identity change was not correlated to family support in any wave.

Table 2.

Pairwise correlations, means, standard errors, chronbach α, and ranges of study variables.

	Variables	1	2	3	4	5
1.	W1 Family Support	╲
2.	W2 Family Support	0.63 *	╲
3.	W3 Family Support	0.56 *	0.70 *	╲
4.	Sexual Identity change W2	−0.04	0.01	−0.05	╲
5.	Sexual Identity change W3	0.02	−0.03	0.02	−0.05	╲
	N	961	871	689	961	961
	Mean	4.78	4.70	4.87	-	-
	Standard error	0.07	0.07	0.08	-	-
	α	0.92	0.92	0.93	-	-
	Range	1–7	1–7	1–7	0–1	0–1

Note:

^*p <.05.

W2 = Wave 2. W3 = Wave 3.

Figure 2 –

Means (accounting for sampling weights) and 95% confidence intervals of family support at Wave 3 stratified by participants’ cohort membership and pattern of stability and change of sexual identities.

We first estimated an unconditional LGCM for family support to examine the trajectories of family support among participants. The model with a linear trajectory of family support fitted the data well (χ²(1) = 7.004, p = .001; RMSEA = .079 [CI 90%: .032–.138]; CFI = 0.981; TLI = 0.942). The mean for the final levels of perceptions of family support (i.e., intercept; b = 4.88, p<.001) and for the slope (b = 0.07, p = .059) indicates that participants perceived their family as relatively supportive at the end of the study and that perceptions of family support were not significantly different over time. Family support intercept and slope variances were significant (respectively, b = 2.05, p <.001, and b = 0.30, p < .001), indicating heterogeneity between participants in final levels and rate of change of family support.

Sexual Identity Change and Family Support

We first tested whether the effects of identity change on the final levels and slope of family support were similar or varied in each wave. This model-fitting process led us to a model where the effects of identity change on family support were constrained to be equal across waves. The LGCM examining whether sexual identity change had significant differences in the trajectory of family support compared to participants who had consistent monosexual and plurisexual identities, including covariates in the model presented good model fit (χ²(20) = 30.078, p = .090; RMSEA = .021 [90% CI: .000–.037]; CFI = 0.985; TLI = 0.959; Table 3). Results indicated that the final levels and the slope of family support for participants who changed sexual identity during the study period were not significantly different than participants who were consistently monosexual (b_intercept = −0.21, p = .366, b_slope = 0.07, p = .488) and participants who were consistently plurisexual (b_intercept = 0.08, p = .728, b_slope = 0.08, p = .411). The comparison of the final levels of family support between participants who had consistent identities approached significance, such that participants who were consistently plurisexual reported lower final levels of family support compared to those who were consistently monosexual (b_intercept = −0.29, p = .074). No difference in the slope was found between such groups (b_slope = −0.01, p = .922).

Table 3.

Latent Growth Curve Models examining the effects of any sexual identity change on the trajectories family support

	Intercept			Slope
	b	SE	p	b	SE	p
Consistently monosexual	Ref
Consistently plurisexual	−0.29†	0.16	.074	−0.01	0.08	.922
Sexual orientation identity change	−0.21	0.23	.366	0.07	0.10	.488
Consistently monosexual	0.29†	0.16	.074	0.01	0.08	.922
Consistently plurisexual	Ref
Sexual orientation identity change	0.08	0.23	.728	0.08	0.09	.411
Sex Assigned at Birth (Ref = Male)
Female	−0.47	0.63	.453	0.06	0.31	.846
Gender Identity W1 (Ref = Man)
Woman	0.49	0.65	.445	−0.09	0.32	.769
Non-binary	0.08	0.56	.883	−0.04	0.29	.893
Income Level (Ref = Below 100% FPL)¥	0.19**	0.07	.008	0.02	0.04	.684
Urbanicity (Ref= Non-urban)
Urban	−0.25	0.19	.192	−0.06	0.09	.514
Education (Ref = High School or less)
Some college	0.15	0.18	.423	0.13	0.09	.146
College	0.15	0.19	.433	0.13	0.10	.206
More than college	0.29	0.21	.180	0.15	0.11	.181
Race/Ethnicity (Ref = NL White)
NL Black	0.10	0.28	.727	0.08	0.16	.625
Latina/x/o	0.22	0.25	.382	0.09	0.14	.522
NL Multiracial	0.04	0.20	.842	0.04	0.09	.679
Cohort (Ref = Younger)
Middle	−0.28	0.49	.564	0.08	0.24	.724
Older	−0.27	1.05	.797	0.42	0.55	.441
Age	−0.01	0.03	.701	−0.01	0.02	.385
Religion (Ref=Agnostic/Atheist)
Reported a religion	0.08	0.14	.572	−0.15*	0.07	.024

Note:

^¥Income level was treated as a continuous variable. FPL = Federal poverty line. NL = Non-Latino.

Next, we examined contributions of the direction of change to the trajectories of family support (Table 4). Participants who changed from a monosexual to a plurisexual identity (vs. consistently monosexual: b_intercept = −0.34, p = .313, b_slope = −0.07, p = .523; vs. consistently plurisexual: b_intercept = −0.03, p = .933, b_slope = −0.06, p = .527), and participants who moved from a plurisexual to a monosexual identity (vs. consistently monosexual: b_intercept = −0.24, p = .426, b_slope = 0.15, p = .295; vs. consistently plurisexual: b_intercept = 0.07, p = .796, b_slope = 0.16, p = .270) did not have differences in the intercept and the slope compared to the groups with consistent sexual identities.

Table 4.

Latent Growth Curve Models examining the effects of the direction of sexual identity change on the trajectories family support

	Intercept			Slope
	b	SE	p	b	SE	p
Consistently monosexual	Ref
Consistently plurisexual	−0.31†	0.16	.059	−0.01	0.08	.938
Monosexual to Plurisexual	−0.34	0.33	.313	−0.07	0.11	.523
Plurisexual to Monosexual	−0.24	0.30	.426	0.15	0.15	.295
Consistently monosexual	0.31†	0.16	.059	0.01	0.08	.938
Consistently plurisexual	Ref
Monosexual to Plurisexual	−0.03	0.34	.933	−0.06	0.10	.527
Plurisexual to Monosexual	0.07	0.29	.796	0.16	0.15	.270
Sex Assigned at Birth (Ref = Male)
Female	−0.51	0.63	.414	0.03	0.31	.932
Gender Identity W1 (Ref = Man)
Woman	0.56	0.65	.391	−0.05	0.32	.868
Non-binary	0.11	0.56	.846	−0.03	0.29	.916
Income Level (Ref = Below 100% FPL)¥	0.18 *	0.07	.011	0.02	0.04	.539
Urbanicity (Ref= Non-urban)
Urban	−0.25	0.19	.188	−0.07	0.09	.440
Education (Ref = High School or less)
Some college	0.17	0.18	.354	0.13	0.09	.149
College	0.14	0.19	.454	0.12	0.10	.246
More than college	0.29	0.21	.173	0.15	0.11	.192
Race/Ethnicity (Ref = NL White)
NL Black	0.09	0.28	.750	0.08	0.15	.601
Latina/x/o	0.21	0.25	.402	0.09	0.14	.507
NL Multiracial	0.00	0.20	.999	0.03	0.09	.754
Cohort (Ref = Younger)
Middle	−0.16	0.48	.738	0.10	0.24	.686
Older	−0.10	1.05	.924	0.45	0.55	.419
Age	−0.02	0.03	.573	−0.02	0.02	.364
Religion (Ref=Agnostic/Atheist)
Reported a religion	0.09	0.14	.497	−0.15*	0.07	.028

Note:

^¥Income level was treated as a continuous variable. FPL = Federal poverty line. NL = Non-Latino.

^†p < .10;

^*p < 0.05,

^**p < 0.01,

^***p < .001.

Cohort Differences in Sexual Identity Change and Family Support

In order to explore possible moderation effects of cohort membership on the association between sexual identity change and trajectories of family support, LGCM models where the paths between sexual identity change and the intercept and the slope of family support were free to vary between cohort groups were estimated and compared to LGCM models where these paths were estimated to be equal across groups. A significantly worse model fit was observed in the constrained model compared to the model in which the intercept and the slope were allowed to vary between groups (Satorra-Bentler Δχ² (4) = 14.452, p = .001) (see Table S5 in the supplemental materials), suggesting that there were meaningful differences in the association of history of sexual identity change and the trajectories of family support between cohort groups.

Follow-up analysis freeing separately only the association between sexual identity change and the intercept, and (separately) sexual identity change and the slope, aimed to investigate whether differences held for the intercept, the slope, or both. This process revealed that the cohort group differences were held in the association between sexual identity change and the intercept of family support (Satorra-Bentler Δχ² (2) = 12.688, p < .001). Specifically, we found that participants from the middle cohort who changed sexual identities in the study period reported lower final levels of family support than those who were consistently monosexual and those who were consistently plurisexual (vs. consistently monosexual: b_intercept = −0.95, p = .034; vs. consistently plurisexual: b_intercept = −1.09, p = .018). Having had a sexual identity change was not associated with final levels of family support among the younger and the older cohorts compared to the two groups with consistent identities within the same cohorts (see Table 5). There were no significant differences in the slope of family support across participants who changed sexual identities and those who had consistent sexual identities for the three cohort groups. Of note, among participants from the older cohort, those who were consistently plurisexual reported lower final levels of family support compared to those who were consistently monosexual (b_intercept = −0.73, p = .014).

Table 5.

Multigroup analysis examining moderation effects of cohort membership on the association between sexual identity change and trajectories family support

	Younger				Middle				Older
	Intercept		Slope		Intercept		Slope		Intercept		Slope
	b	SE	b	SE	b	SE	b	SE	b	SE	b	SE
Consistently monosexual	(Ref)				(Ref)				(Ref)
Consistently plurisexual	−0.21	0.23	0.02	0.11	0.13	0.31	0.14	0.19	−0.73	0.30	−0.01	0.13
Sexual orientation identity change	0.19	0.28	0.19	0.14	−0.95*	0.45	−0.11	0.21	0.14	0.41	0.04	0.17
Consistently monosexual	0.21	0.23	−0.02	0.11	−0.13	0.31	−0.14	0.19	0.73	0.30	0.01	0.13
Consistently plurisexual	(Ref)				(Ref)				(Ref)
Sexual orientation identity change	0.40	0.27	0.17	0.12	−1.09*	0.46	−0.25	0.19	0.87†	0.48	0.06	0.21
Sex Assigned at Birth (Ref = Male)
Female	−0.56	0.78	0.17	0.37	−0.66	1.65	−0.85	0.90	−0.35	0.80	0.27	0.42
Gender Identity W1 (Ref = Man)
Woman	0.34	0.82	−0.34	0.39	0.96	1.66	0.97	0.91	0.56	0.83	−0.28	0.43
Non-binary	−0.19	0.72	−0.23	0.35	0.93	1.57	0.89	0.87	0.55	0.51	−0.15	0.36
Income Level (Ref = Below 100% FPL)¥	0.08	0.09	0.00	0.05	0.54***	0.14	0.02	0.08	0.08	0.13	−0.05	0.06
Urbanicity (Ref = Non-urban)
Urban	−0.12	0.28	0.08	0.12	−0.80*	0.38	−0.55†	0.29	0.22	0.33	−0.02	0.14
Education (Ref = High School or less)
Some college	0.15	0.24	0.08	0.12	−0.21	0.37	0.60***	0.17	0.34	0.33	0.03	0.14
College	0.08	0.29	0.06	0.15	−0.34	0.38	0.38*	0.18	0.52	0.34	0.31*	0.14
More than college	−0.54	0.51	−0.31	0.24	−0.13	0.39	0.61***	0.19	0.70	0.34	0.26†	0.14
Race/Ethnicity (Ref = NL White)
NL Black	0.02	0.39	0.17	0.20	−0.40	0.42	−0.05	0.29	0.96	0.58	0.28	0.19
Latina/x/o	−0.01	0.35	−0.01	0.20	0.76*	0.33	0.59**	0.22	0.42	0.48	0.00	0.17
NL Multiracial	0.07	0.23	0.01	0.12	0.17	0.43	0.32	0.22	0.28	0.43	0.18	0.18
Age	−0.01	0.05	−0.01	0.02	−0.04	0.06	0.00	0.04	0.00	0.05	−0.04	0.02
Religion (Ref=Agnostic/Atheist)
Reported a religion	0.02	0.20	−0.30**	0.10	0.02	0.24	0.01	0.12	0.43†	0.24	0.11	0.10

Note:

^¥Income level was treated as a continuous variable. FPL = Federal poverty line. NL = Non-Latino. Results from model where intercept and slope varied freely between cohort groups. Intercept was set at Wave 3. SE = Standard deviation.

^†p < .10;

^*p < 0.05,

^**p < 0.01,

^***p < .001.

Notably, we also found cohort membership differences in the associations between the direction of sexual identity change and trajectories of family support (see Table 6). This model revealed that differences in final levels of family support for participants who changed sexual identities (vs. participants with consistent monosexual, and consistent plurisexual identities) in the middle cohort were mainly driven by participants who changed from a monosexual to a plurisexual identity. Specifically, compared to participants who had consistent identities, those who changed from a monosexual to a plurisexual identity reported lower final levels of family support (vs. consistently monosexual: b_intercept = −1.95, p = .007; vs. consistently plurisexual: b_intercept = −1.99, p = .006). Participants who changed from a monosexual to a plurisexual identity also had decreasing levels of family support over time compared to those with a consistent plurisexual identity (b_slope = −0.51, p = .060). Among participants from the older cohort, a mixed pattern was found: (1) participants who changed from a monosexual to a plurisexual identity reported lower final levels of family support than those with a consistent monosexual identity (b_intercept = −1.06, p = .060), and (2) participants who changed from a monosexual to a plurisexual identity reported higher final levels of family support than participants with consistent plurisexual identities (b_intercept = 1.06, p = .022).

Table 6.

Multigroup analysis examining moderation effects of cohort membership on the association between the direction of sexual identity change and trajectories family support

	Younger				Middle				Older
	Intercept		Slope		Intercept		Slope		Intercept		Slope
	b	SE	b	SE	b	SE	b	SE	b	SE	b	SE
Consistently monosexual	(Ref)				(Ref)				(Ref)
Consistently plurisexual	−0.21	0.24	0.02	0.11	0.05	0.30	0.12	0.19	−0.72*	0.30	−0.01	0.13
Monosexual to Plurisexual	0.00	0.39	0.00	0.15	−1.95**	0.72	−0.38	0.28	0.44	0.47	0.17	0.19
Plurisexual to Monosexual	0.17	0.29	0.31	0.19	−0.54	0.60	0.05	0.29	−1.06†	0.57	−0.27	0.32
Consistently monosexual	0.21	0.24	−0.02	0.11	−0.05	0.30	−0.12	0.19	0.72	0.30	0.01	0.13
Consistently plurisexual	(Ref)				(Ref)				(Ref)
Monosexual to Plurisexual	0.21	0.38	−0.02	0.13	−1.99**	0.72	−0.51†	0.27	1.16	0.52	0.18	0.22
Plurisexual to Monosexual	0.37	0.28	0.29	0.18	−0.59	0.60	−0.08	0.28	−0.34	0.63	−0.26	0.34
Sex Assigned at Birth (Ref = Male)
Female	−0.61	0.79	0.11	0.37	−0.72	1.34	−0.83	0.78	−0.42	0.82	0.25	0.42
Gender Identity W1 (Ref = Man)
Woman	0.40	0.82	−0.28	0.38	1.05	1.36	0.98	0.78	0.66	0.85	−0.25	0.44
Non-binary	−0.17	0.72	−0.22	0.35	1.27	1.24	0.98	0.75	0.59	0.52	−0.15	0.36
Income Level (Ref = Below 100% FPL)¥	0.09	0.09	0.00	0.05	0.47**	0.14	0.03	0.09	0.10	0.13	−0.04	0.06
Urbanicity (Ref = Non-urban)
Urban	−0.17	0.27	0.05	0.12	−0.81*	0.37	−0.54	0.29	0.20	0.34	−0.02	0.14
Education (Ref = High School or less)
Some college	0.15	0.24	0.08	0.12	−0.12	0.37	0.59*	0.17	0.33	0.33	0.03	0.14
College	0.09	0.29	0.05	0.15	−0.33	0.37	0.32	0.18	0.52	0.34	0.31*	0.14
More than college	−0.52	0.51	−0.30	0.24	−0.06	0.37	0.57*	0.19	0.72*	0.34	0.27†	0.14
Race/Ethnicity (Ref = NL White)
NL Black	0.05	0.38	0.18	0.20	−0.45	0.43	−0.05	0.29	1.03†	0.60	0.29	0.19
Latina/x/o	0.02	0.35	0.00	0.20	0.74*	0.32	0.58**	0.21	0.43	0.48	0.00	0.17
NL Multiracial	0.05	0.23	−0.02	0.12	0.06	0.45	0.39	0.23	0.26	0.43	0.17	0.19
Age	−0.01	0.05	−0.01	0.02	−0.05	0.06	0.01	0.04	−0.01	0.05	−0.04†	0.02
Religion (Ref=Agnostic/Atheist)
Reported a religion	0.02	0.20	−0.29**	0.10	0.05	0.24	0.00	0.12	0.40†	0.24	0.10	0.10

Note:

^¥Income level was treated as a continuous variable. FPL = Federal poverty line. NL = Non-Latino. Results from model where intercept and slope varied freely between cohort groups. Intercept was set at Wave 3. SE = Standard deviation.

^†p < .10;

^*p < 0.05,

^**p < 0.01,

^***p < .001.

Discussion

Prior cross-sectional work documented that when individuals (in particular, youth) change from a heterosexual identity to a SM identity, they often experience initial negative reactions from family members (D’Augelli et al., 2005; Rosario et al., 2009). Using longitudinal data from a U.S. national probability sample of SM people, this study examined whether changes in sexual identity between different SM identities are associated with trajectories of family support. While any type of change in sexual identity was not associated with final levels (i.e., intercept) or the rates of change (i.e., slope) of family support for this sample, regardless of their cohort membership, we found that the associations between sexual identity change and the trajectories of family support differed across generational cohorts. First, our findings suggest that middle-aged adults (i.e., 34–43 years) perceived families as less supportive after changing sexual identity (in particular, when they changed from a monosexual to a plurisexual identity) compared to their peers who had a consistent sexual identity over time. Second, among participants from the older cohort, we also found that identity change was associated with more or less family support depending on the direction of sexual identity change. While participants who changed from a plurisexual to a monosexual identity reported lower levels of family support compared to participants who were consistently monosexual, participants who changed from a monosexual to a plurisexual identity reported higher levels of family support compared to participants who were consistently plurisexual.

Findings from this study indicate that associations between sexual identity changes differ based on individuals’ life stage or generational cohort. Specifically, individuals from the middle cohort who changed from a monosexual to a plurisexual identity perceived their family as less supportive compared to individuals who did not change sexual identities during the study. Different from the younger cohort, adults in the middle cohort are at an age where stability in sexual identity is more expected. Additionally, families may be more resistant to accepting and supporting identity change among middle-aged adults than they are when it occurs among youth. In addition to the age, a cohort effect may also come into play: these families may be less supportive of individuals who report sexual identity changes because prior generations are more likely to have negative views of sexual diversity (Fish & Russell, 2022), including changes in sexual identity.

While any change in sexual identity was not associated with differences in family support among adults in the older cohort, mixed findings emerged when examining associations between the direction of identity change and family support trajectories. First, we found that participants who changed from a plurisexual to a monosexual identity reported lower final levels of family support than those who had consistently monosexual identities. Although, in this case, participants changed from an identity that tends to be more stigmatized to an identity that tends to be less stigmatized, it is possible that these participants may have received less support just because families may consider identity change itself as non-normative at that life stage. Moreover, a change in identity could be a manifestation of a change in sexual attractions and romantic relationships (Rosario et al., 2006). Thus, it is possible that these participants identified as plurisexual and were in different-gender relationships, and started to adopt a monosexual identity after being in a same-gender relationship, which may have led to loss of support. In contrast and contrary to our expectations, participants from the older cohort who changed from a monosexual to a plurisexual reported higher levels of family support compared to those who were consistently plurisexual. This finding was surprising given that adults in the older cohort tend to have family members who are also older, and older individuals are more likely to have more negative views of sexual diversity (Fish & Russell, 2022). At the same time, these patterns are based on a small number of cases in the older cohort, for whom sexual identity change was least common; the reasons explored here are speculative and cannot be confirmed based on the available data. Future studies should continue to investigate how the direction of identity change is associated with social support for older adults.

We also found marginal differences in family support between participants who were consistently monosexual and participants who were consistently plurisexual. It was notable, however, that such a difference was evident among participants in the older generation, such that those who were consistently plurisexual reported less family support at the end of the study than those who were consistently monosexual. Whereas these findings highlight the fact that plurisexual identities tend to be more stigmatized than monosexual identities (Ross et al., 2018), our findings also reveal that older adults may be more vulnerable to exclusion from family relationships. Because older SM people are also typically surrounded by other older people, this discrepancy in family support corroborates the idea that older generations tend to be less accepting of SM individuals, which seems to be amplified for plurisexual individuals. Aligned with this, recent work has shown that older SM adults tend to rely less on family members as a source of support for major needs compared to younger SM adults (la Roi et al., 2022).

In the past decade, although growing attention has been given to the associations between sexual identity change and compromised well-being (Everett, 2015; Feinstein et al., 2019; Fish & Pasley, 2015; Katz-Wise et al., 2017; Xu & Rahman, 2024), studies that examined possible mechanisms for these associations are still scarce. A mechanism through which sexual identity changes may be associated with compromised mental health is a possible loss of social support from others (Everett, 2015; Katz-Wise et al., 2019; Rosario et al., 2011). Although there is evidence that sexual identity change is associated with compromised mental health when individuals change between SM identities (Krueger et al., 2024), the hypothesis that loss of family support is a mechanism to compromised mental health was based on research that mainly considers changes from a heterosexual identity to a SM identity (D’Augelli et al., 2005; Everett, 2015). In this sense, associations between identity change and family support may be dependent on the direction of identity change. Given the importance of family support for SM people’s well-being (la Roi et al., 2022; Real & Russell, 2025; Ryan et al., 2010), our study suggests that family support could be a potential mechanism in which identity change may lead to compromised mental health, but these associations may not only be dependent on the direction of change but also on an individual’s life stage/generational cohort. More studies with larger sample sizes should consider how the direction of identity change (including changes between SM identities) influences family relationships for individuals in different life stages. Future research should apply mediation models to empirically test whether identity change leads to compromised mental health through family support.

Finally, we did not find differences in family support based on patterns of stability and change of sexual identity for SM individuals from the younger cohort. There are several possible explanations for this finding. Younger cohorts have younger families, which in turn may be more accepting of sexual diversity and changes in sexual identity (Fish & Russell, 2022). Further, families may consider sexual identity exploration developmentally appropriate for younger adults (Lilly et al., 2024; Xu et al., 2021), particularly in younger cohorts. However, despite expectations that family support among younger SM individuals would be less impacted by sexual identity change compared to those in the middle and the older cohorts, it was somewhat surprising that such identity changes were not associated with the trajectories of family support for younger individuals. This is particularly noteworthy because the majority of the studies that have found associations between sexual identity changes and psychopathology utilize samples of youths (Everett, 2015; Srivastava et al., 2022, 2023; Xu & Rahman, 2024). Again, considerations of the direction of identity change may bring insights on this topic, since these studies consider changes from a heterosexual to a SM identity, rather than changes between different SM identities as we do in the present study. It may be that associations between changes in family support after sexual identity change may be distinctly salient among those who change from a heterosexual to a SM identity, and less salient among people who have already assumed a SM identity.

Limitations

Despite the strengths of the current study, including a diverse national probability sample, the longitudinal design, and a novel investigation of associations between sexual identity change and trajectories of family support across cohorts, this study has limitations. First, it is not possible to determine whether differences in family support are an effect of generational differences or normative aging. For example, the finding that participants who had changed sexual identities were not different than those with consistent identities in terms of family support can be explained by an increase in acceptance of sexual diversity in society, especially for younger generations (Fish & Russell, 2022; la Roi et al., 2022). On the other hand, age-related expectations for sexual identity consistency may shape the patterns of family support; individuals from the middle cohort may experience less family support because identity change is not expected in this stage of life. Second, the number of individuals changing sexual identities over time was small (10.8%), especially among the middle and older cohorts. However, the use of robust standard errors seeks to compensate for possible bias and provide reliable parameter estimates. Furthermore, because our sample only includes participants who initially identified as SM, we were not able to examine how the changes from a heterosexual to a SM identity impact the trajectories of family support. Prospective studies with larger samples should examine the impacts of multiple directions of sexual identity change in different sources of support. Third, participants were followed for only two years; studies with longer follow-up periods may elucidate possible long-term implications of sexual identity changes for family support. Prior cross-sectional work considering change from a (assumed) heterosexual identity to a SM identity showed that although family reactions are initially negative, they tend to improve over time (D’Augelli et al., 2005).

Fourth, this study measures self-reported perceptions of family support, and meanings of family support may vary across the ages and cohorts included here. As patterns of family life have changed for SM people over time (Ophir et al., 2023), it could be that differences in perceptions of family support have to do with definitions of who is understood as family, including the fact that more recent cohorts are more likely to include chosen family when describing their families (Hull & Ortyl, 2019). The differences that we observe are a reflection of individuals’ own perceptions of rejection from others when they change sexual identities, rather than actual family views. Moreover, although the data utilized in this study is from a probability sample, generalizability of findings may be limited to samples outside of the U.S., especially considering the specificities of the sociopolitical climate in the U.S. Lastly, despite including participants from a diversity of backgrounds, individuals who accept being part of a study of SM people may not represent the experiences of family support of all individuals who change identities; for example, participant retention for such a longitudinal study may be biased toward those with more family support. Future research should consider including participants with a diversity of sexual identities at baseline to capture possible differences between these groups.

Conclusions

By examining the associations of changes in sexual identity and trajectories of family support in a U.S. national probability sample from three different cohorts, the current study suggests that family support following a sexual identity change is not equal across individuals from different cohorts. While younger SM individuals who had changed sexual identities experience the same levels of family support as those with a consistent sexual identity over time, middle-aged individuals perceived their family as less supportive. Among older SM adults, depending on the direction of identity change, individuals had less or more family support. Altogether, findings indicate that some SM individuals experience marginalization and social rejection when they change sexual identities, despite significant improvements in societal climate toward SM individuals. These results support a need for attention to family reactions to changes in sexual identity in systems of support and care. For example, mental health and healthcare providers may screen for family reactions when individuals report a change in their sexual identity, regardless of their age, and provide additional support when needed. Continued studies are needed to explore experiences, the implications on individuals’ lives as they change their sexual identity, and possible moderators of these associations.

Public Significance Statement:

This study suggests that sexual minority individuals who report a change in sexual identities may observe changes in levels of family support over time. These changes in family support were particularly observed among participants from middle and older cohorts and varied based on the direction of the sexual identity change. Understanding individuals’ sexual identity over time is important to provide adequate support when families react negatively.

Availability of data and material:

Data and methodology can be found at: https://www.icpsr.umich.edu/web/DSDR/studies/37166.

Code availability:

The final code used for the analyses reported in the published version of the manuscript will be made available upon request by emailing the corresponding author.