A Latent Profile Analysis of Perceived Family Reactions to Youth LGBTQ Identity

Abstract

Family members’ reactions to youth identity disclosure are important predictors of well-being for lesbian, gay, bisexual, transgender, queer or questioning (LGBTQ) youth. To better understand potential variation within and across families’ current reactions, this study established latent profiles of family-level reaction patterns and examined predictors and outcomes associated with these patterns. In 2011-2012, LGBTQ youth (N = 447, M_age = 18.8) rated their mother’s, father’s, brother’s, and sister’s reactions to their LGBTQ identity, and reported their own depressive symptoms and self-esteem. Latent profile analysis tested patterns of family members’ reactions. Most participants reported either moderately positive reactions (49.2%) or very positive reactions (34.0%) from all family members, though 16.8% of youth reported negative reactions from all family members. Youth social positions and demographic factors predicted profile membership: transgender youth, youth assigned male at birth, older age at first disclosure predicted membership in the negative family reaction profile, whereas gay youth, having a parent and/or sibling with an LGBTQ identity, co-residence with either mothers, fathers, or siblings, and more years since first disclosure predicted membership in the very positive family reaction profile. Multiracial youth and younger youth were more likely to be in the moderately positive family reaction profile. Youth in families characterized by negative reactions had higher depressive symptoms and lower self-esteem compared to those who reported moderately positive or very positive reactions. Findings underscore the interconnectedness of family members’ reactions and suggest that interventions for LGBTQ youth with rejecting and/or less accepting family members may need to target the entire family system.

Many LGBTQ youth disclose their LGBTQ identities to their parents and siblings during adolescence (Grafsky, 2018). Although disclosure is a significant milestone that can foster greater emotional closeness between LGBTQ youth and their families (e.g., Hilton & Szymanski, 2014), disclosure occurs in the context of a heterosexist and cissexist society, and positive reactions from family members are not assured. Findings from studies over the past few decades indicate that many LGBTQ youth report intolerant or rejecting reactions from family members (D’Augelli et al., 1998; 2008). Furthermore, perceptions of family connectedness have declined for lesbian, gay, and bisexual youth, relative to heterosexual youth, from 1998 to 2013 (Watson et al., 2019). Family members’ reactions to LGBTQ youths’ identity have important implications for that young person’s well-being, such that positive reactions predict better physical and emotional well-being (Ryan et al., 2010) whereas negative reactions are associated with greater incidence of attempted suicide, substance use, unprotected sex, higher levels of depression, and lower self-esteem (Ryan et al., 2009; Meanley et al., 2021).

Despite a growing body of work on family reactions, relatively little is known about how individual family members’ reactions cohere or differ from one another. Most of the extant literature measuring family reactions has asked youth to report overall family reactions, despite potential differences between individual family members’ reactions. Yet family systems perspectives would suggest that considering and integrating multiple family members’ reactions is needed to truly understand “family” reactions (Heatherington & Lavner, 2008). Differences in family members’ reactions to youth identity disclosure have been suggested by previous research—for example, that siblings’ reactions may be more positive in part because these relationships are more equal and less hierarchical relative to those with parents (Grafsky et al., 2018). It is also likely that there are subgroups of youth with different patterns of family reactions: some youth may perceive positive reactions from their siblings but negative reactions from their parents; others may perceive positive reactions from their mother, negative reactions from their father, and more moderate reactions from their siblings; and still others may perceive consistent reactions across family members (either positive or negative). Profile-based (i.e., person-centered) methods can be used to identify these different types of patterns based on mother, father, and sibling reactions, but have yet to be applied to the study of reactions to LGBTQ identity. The purpose of this study was to investigate profiles of family members’ current reactions to youth LGBTQ identity, examine the sociodemographic predictors of these profiles, and investigate how family reaction profiles are associated with youth well-being.

Perceived Family Reactions to Youth LGBTQ Identity

Questions about family reactions to youth identity have been investigated since there has been attention in the social sciences to the lives of LGBTQ youth.¹ An early study showed that among LGB youth whose family members knew their identity, 51% of mothers, 27% of fathers, and 57% of siblings were perceived as fully accepting, while 26% of fathers, 10% of mothers, and 15% of siblings had rejecting reactions (D’Augelli et al., 1998); the remaining family members were perceived to have intolerant reactions (9% of mothers and siblings; 16% of fathers). Later work has shown that LGBQ youth perceive few negative reactions from siblings after disclosure, but when negative reactions do occur, they are more likely to come from brothers than from sisters (D’Augelli et al., 2008). Among transgender and gender diverse youth, one study found that 54% of mothers and 63% of fathers reacted negatively to initial disclosure, a percentage that remained quite high over an average of 3 years post-disclosure (50% for mothers, 44% for fathers) (Grossman et al., 2005). More recent research indicates that although many youth perceive positive reactions, negative reactions are especially likely to occur among male family members (e.g., fathers, brothers) in reaction to youth disclosing pansexual, asexual, transgender, or gender diverse identities (Gamarel et al., 2020).

Average differences between family members’ reactions have also been described, though the results are not consistent across studies. For instance, recent Italian studies have found no differences between perceptions of mothers’ and fathers’ reactions to youth disclosure (Baiocco et al., 2015; Pistella et al., 2020), whereas some U.S. studies have found that mothers have more accepting reactions than fathers (Newcomb et al., 2018). Other work has shown that fathers react more positively than mothers and report fewer worries, anger, and guilt (Willoughby et al., 2006). Mothers’ and fathers’ reactions have been found to be significantly less positive than brothers’ and sisters’ reactions (Pistella et al., 2020).

Existing research comparing percentages or average differences (e.g., average mother reaction versus average father reaction) cannot address how, within a given family, family members may differ in their reactions. The prevailing focus on sample-level comparisons obscures the possibility that there might be subgroups of LGBTQ youth with distinct patterns of family reactions, as described previously. Identifying such patterns would address questions such as the degree to which families are likely to be characterized by highly positive sibling reactions but negative parent reactions, or whether the discordance between family members’ average affective reactions (e.g., mothers’ and siblings’ reactions to youth identity are more positive, on average, than fathers’ reactions) is actually common for most families.

Predicting Perceived Family Reaction Profiles

Youth social positions and demographics (e.g., sexual orientation, gender identity, age, sexual identity development milestones, having LGBTQ family member(s), living arrangement, race, ethnicity) may be significantly associated with profiles of family reactions. Differences in family reactions by sexual orientation and gender identity have yielded somewhat inconsistent results. Shilo and Savaya (2011) found that bisexual youth reported significantly less family acceptance of their sexual orientation compared to gay or lesbian youth, though Ryan et al. (2010) found that family acceptance did not significantly vary by sexual orientation, gender identity, or transgender identity. In contrast, perceived family rejection does typically differ by youth sexual orientation and gender identity. For instance, sexual minority men report significantly more rejecting family reactions than sexual minority women (Ryan et al., 2009). Pansexual, asexual, and youth who report another sexual orientation experience more family rejection than gay or lesbian, queer, or questioning youth; transgender boys and non-binary assigned female at birth youth also experience more family rejection compared to cisgender boys and transgender girls (Gamarel et al., 2020).

Youth age has been found to predict parental reaction such that younger youth report more negative reactions (Baiocco et al., 2015). Regarding sexual identity development, milestones for LGBQ youth can include becoming aware of lesbian, gay, bisexual, and queer sexual identity labels, identifying themselves as such, and first disclosure of their LGBQ identity to another person. Given that these sexual identity milestones can occur before disclosure to families, these milestones can be investigated as predictors of family reactions to determine whether longer (or shorter) times since the milestone occurred are associated with more or less positive family reactions. Other family members (parents, siblings) may also hold LGBTQ identities, which may impact the positivity of responses from family members when youth disclose their own LGBTQ identity. Stone et al. (2022) note that the presence of parents or siblings who hold LGBTQ identities may help youth navigate family life and provide needed camaraderie in a context that is often isolating and hostile. Co-residence with parents and siblings is likely an outcome related to family reactions which may differ by reaction profile; youth perceiving hostile reactions may be most likely to be living apart from family members given their negative reactions.

Youth race and ethnicity is also relevant to family reactions to youth LGBTQ identity disclosure. Regarding youth race, Abreu et al. (2022) found that LGBTQ youth of color report less LGBTQ-specific support from parents relative to LGBTQ White youth. Such findings should be contextualized within intersecting racist, homophobic, and transphobic systems of oppression and the ensuing complexity of parents’ fears and concerns as they prepare their children to navigate these hostile systems. Abreu et al. (citing work by Furman et al., 2018) suggest that racism within LGBTQ spaces means that LGBTQ-specific support and advocacy systems provide more resources to White parents of LGBTQ youth than to parents of color. Regarding youth ethnicity, Hispanic or Latina/o/x LGBTQ youth report more negative family reactions relative to non-Hispanic or Latina/o/x LGBTQ youth (Ryan et al., 2009). Gattamorta et al. (2019) found that, although Hispanic parents of LGB youth reported initial negative reactions to youth identity disclosure, cultural factors (particularly familismo) helped to explain changes in reactions that eventually led all parents in the sample to accept their LGB child.

Outcomes Associated with Family Member Reactions

Patterns of family member reactions may ultimately be associated with youth well-being (depressive symptoms and self-esteem). Family members’ reactions to youth LGBTQ identity have been found to predict a range of important youth outcomes. Family reactions that are positive (assessed through youth report of parent/caregiver behavior) predict greater self-esteem, social support, and better physical health for LGBT young adults (Meanley et al., 2021; Ryan et al., 2010), whereas negative reactions are associated with greater incidence of attempted suicide, illegal substance use, unprotected sex, and higher levels of depression for LGB young adults (Ryan et al., 2009), as well as lower self-esteem (Meanley et al., 2021). Similar results have been found for transgender and gender diverse youth; youth who perceive their parents as supportive of their identity report higher life satisfaction, self-esteem, and fewer depressive symptoms (de Vries et al., 2016). Nonetheless, it is somewhat difficult to assess the contributions of whole family systems to youth well-being because most measures of family reactions are aggregated and do not examine variation in family member responses.

Current Study

The current study extends research on family members’ perceived reactions to youth LGBTQ identity by identifying unique family-level profiles based on four family members’ (father, mother, sister, brother) current reactions to youth LGBTQ identity disclosure, and the predictors and outcomes associated with these profiles. As noted previously, much of the research detailing patterns of family reactions has aged, and updated information about family members’ reactions is needed. This study uses latent profile analysis to address the following exploratory research questions:

1. Are there distinct profiles of family members’ current reactions to youth LGBTQ identity and what is the prevalence of such profiles?

2. Which sociodemographic characteristics (sexual orientation, gender identity, age, sexual identity development milestones, having LGBTQ family member(s), living arrangement, race, ethnicity) predict latent profile membership?

3. How do youth well-being outcomes (self-esteem, depressive symptoms) differ across profiles?

No a priori predictions were made about the number, prevalence, or potential correlates of the profiles.

Method

Data for this study come from the first wave of a longitudinal study of 1,061 LGBTQ youth that aimed to investigate risk and protective factors related to suicidal ideation, collected from November 2011 – October 2012. We report how we determined our sample size, all data exclusions (if any), all manipulations, and all measures in the study, which was not preregistered. Data and analysis code are available from the corresponding author upon request. Institutional Review Boards of two participating universities approved all study procedures. Participants were recruited from three community-based organizations and college groups in the United States located in the Northeast, Southwest, and West Coast. Participants were eligible for study inclusion if they identified as LGBTQ and were between the ages of 15 and 21 years—data cleaning was applied to ensure that all participants met eligibility criteria. The principal investigators received a federal certificate of confidentiality that allowed youth to participate without requiring parental consent. Youth under 18 years met with a youth advocate to receive more information about the study before they provided informed consent to participate. After the initial screening, eligible participants contacted site coordinators to confirm an appointment to complete a survey packet. Participants completed the survey packet at the selected study site, which took between 40 – 80 minutes. Participants also received a cash incentive for their participation ($30 at the Northeast and West Coast participation sites; $20 at the Southwest site).

Participants

The current study included a subset of youth with four family members (mother, father, sister, brother) who (a) knew their LGBTQ identity and (b) reported current reactions for these individuals (n = 452) based on responses to the Family and Friends Knowledge and Reactions Scale described in the Measures section. This excluded youth who either did not report on these four family members,² such as youth who did not have two different-gender siblings (n = 280) or who reported that one or more family members did not know their LGBTQ identity (n = 277).³ Chi-square and Fisher’s exact tests indicated that youth included versus not included in the sample significantly differed on sexual orientation, race, ethnicity, and sex assigned at birth. Specifically, monosexual youth (lesbian, gay, heterosexual), Black or African American, Native American or Alaska Native, multiracial youth, Hispanic or Latina/o/x youth, and youth assigned male at birth were more likely to be in the analytic sample. There were no differences by gender identity or age (determined using Welch two-sample t-tests) among youth who were included or excluded from the analytic sample.

Youth who reported that a family member currently has “no reaction” to their identity were recoded using a follow-up question: “What do you think the person was feeling?” Participants who responded that their family member had “no reaction” and did not provide a response to the follow-up question but did have a response for another family member (n = 10) were included in the sample, with full information maximum likelihood used to treat missing data. A final sample of 447 participants was included in focal analyses.

In the final sample, the average age of participants was 18.85 years (SD = 1.79, range = 15-21). Youth reported their sexual orientation as gay (n = 155, 34.9%), bisexual (n = 152, 34.2%), lesbian (n = 100, 22.5%), questioning (n = 20, 4.5%), and heterosexual (n = 17, 3.8%). Eighty-six percent (n = 383) of youth were cisgender and 14.3% (n = 64) were transgender.⁴ Almost 16% of youth (n = 71) had a parent or sibling who also held an LGBTQ identity (4% of mothers; 0.89% of fathers; 7.4% of sisters; 6.3% of brothers). Youth reported their sex assigned at birth as male (51.6%) and female (48.4%). Thirty-two percent (n = 143) of participants reported their race as Black or African American, 23.0% (n = 103) as multiracial or another race, 16.1% (n = 72) as White or European American, 3.1% (n = 14) as Native American or Alaskan Native, 3.1% (n = 14) as Asian American or Pacific Islander, and this information was not reported by 22.6% (n = 101) of participants. Forty-one percent (n = 184) of participants reported their ethnicity as Hispanic or Latina/o/x; 44.7% reported their ethnicity as not Hispanic or Latina/o/x (n = 200); 14.1% did not report this information. Fifty-three percent of youth participated at the Northeast site, 31.5% participated at the West Coast site, and 14.9% participated at the Southwest site.

Measures

Family Member Knowledge and Current Reaction to LGBTQ Identity

Youth reported on family members’ current reactions to their LGBTQ identity using items on the Family and Friends Knowledge and Reactions Scale (D’Augelli et al., 2008). Youth responded to the question, “When you first told, or he/she/they first found out that you were LGBTQ, his/her/their reaction was…” followed by the question, “What is the person’s reaction to you now?” This latter question was utilized in this study as a measure of current reaction to LGBTQ identity. Participants reported on mother, father, brother/stepbrother, and sister/stepsister reactions to LGBTQ identity. Responses included 1 = Very negative, 2 = Negative, 3 = Positive, 4 = Very positive, 5 = He/She doesn’t know,⁵ and 6 = No reaction. Participants who selected that their family member had “no reaction” to their identity responded to a follow-up question, “If you selected ‘no reaction,’ what do you think the person was feeling?” to which participants were asked to rate their perceptions of reaction from 1 = Very negative to 4 = Very positive. The responses to this follow-up question were substituted for participants who selected “no reaction” to the current reaction question.

Sexual Identity Development Milestones

Milestones in youth sexual identity development were assessed using three questions: (1) “How old were you when you first became aware that other people called themselves lesbian/gay/bisexual/queer?”, (2) “How old were you when you first labeled yourself lesbian/gay/bisexual/queer?”, and (3) “How old were you when you first told someone that you were lesbian/gay/bisexual/queer?” Participants responded by writing their age at the milestone in a text box. Participant responses were coded to reflect ages numerically (e.g., “17 or 18” was recoded to “17.5”; “All my life,” “Birth,” or “Always,” recoded to “0”). Time since sexual identity development milestones was calculated as age at milestone subtracted from youth current age, yielding three variables: (1) years since awareness (i.e., participant was aware that others called themselves LGBQ), (2) years since participant first labeled themselves LGBQ, and (3) years since participant first told someone else that they were LGBQ.

Sexual Orientation

Youth sexual orientation was assessed through the question, “How would you describe your sexual identity?” Responses included the following categories: “gay,” “lesbian,” “bisexual, but mostly gay or lesbian,” “bisexual, equally gay/lesbian and heterosexual/straight,” “bisexual, but mostly heterosexual/straight,” “heterosexual/straight,” “questioning/uncertain, don’t know for sure.” The three bisexual categories were combined to yield five total sexual orientation categories.

Gender Identity

Youth gender identity was assessed through two questions “What is your birth sex?” (male, female, or intersex) and “What is your gender identity?” (man, woman, queer, trans woman, trans man, or another identity). Using responses to both questions (National Academies of Sciences, Engineering, and Medicine, 2022), participants were categorized as “transgender” if they selected a transgender or gender diverse category (or if sex assigned at birth and gender identity differed) and as “cisgender” if assigned sex was the same as gender identity.

Sex Assigned at Birth

Sex assigned at birth was assessed through the question, “What is your birth sex?” Response options included male, female, and intersex.⁶

Race and Ethnicity

Participants reported their race (Asian or Pacific Islander; Black or African American; Native American or Alaska Native; multiracial; White or European American), and missing responses were coded as “missing,” yielding six categories. Participants reported their ethnicity (Hispanic or Latina/o/x; not Hispanic or Latina/o/x), and missing responses on this variable were coded as “missing,” yielding three categories.

Living with Family Members

Participants responded to the prompt, “With whom do you live…” and reported whether they lived with their mother(s), father(s), and sibling(s). Responses were dichotomously coded for each family member; e.g., 1 = Living with mother, and 0 = Not living with mother.

Family Members‘ LGBTQ Identities

Participants responded to the prompt, “Are any of your relatives LGBTQ?” and were asked to specify which family members had LGBTQ identities. Responses were dichotomously coded for parents and siblings, 1 = One or more parent or sibling LGBTQ, and 0 = Parents and siblings not LGBTQ.

Depressive Symptoms

Youth depressive symptoms were measured using the 20-item Beck Depression Inventory-Youth (BDI-Y; Beck, 1996; sample items: “I have trouble doing things,” “I have trouble sleeping”). Response options ranged from 0 = Never to 3 = Always. Items were averaged, with higher scores reflecting more depressive symptoms (in the current study, α = 0.95).

Self-Esteem

The 10-item Rosenberg Self-Esteem Scale (Rosenberg, 1979) was used to assess youth self-esteem (sample items: “On the whole, I am satisfied with myself,” “I can do just about anything I really set my mind to”). Response options ranged from 1 = Strongly disagree to 4 = Strongly agree. Items were averaged, with higher scores reflecting higher self-esteem (α = 0.82).

Plan of Analysis

We conducted descriptive analysis with four binary indicators consisting of knowledge of youth LGBTQ identity from mothers, fathers, brothers, and sisters to create the subset of youth who reported that all family members knew their identity (see Table S1 in supplemental materials). Next, we conducted a latent profile analysis with four continuous, standardized indicators which assessed youth perceptions of family members’ current reactions to youth identity. Data were managed in R (R Core Team, 2021) and latent profile models were analyzed in Mplus 8.8 (Muthén & Muthén, 1998 – 2017). A one-profile model was fit first and successive profiles were added to determine the number of profiles that best fit the data; the model was estimated with up to five profiles, at which point the model failed to converge. An optimal profile solution was determined through evaluation of multiple indicators of approximate fit including: the Akaike information criterion (AIC), Bayesian information criterion (BIC), and sample size-adjusted BIC (SABIC). Two likelihood ratio tests were also considered: the Vuong-Lo-Mendell-Rubin likelihood ratio test (VLMR-LRT) and bootstrapped likelihood ratio test (BLRT) compare the current profile solution (k) to the solution with one less profile (k − 1). For both ratio tests, a statistically significant test at p < .05 indicates that the previous solution, k − 1, was a significantly worse fit than the current profile solution, k.

Classification diagnostics—entropy and average posterior probabilities (AvePP)—were consulted to ensure that the solutions distinguished distinct profiles. Entropy values greater than 0.80 indicate adequate accuracy of profile assignment (Clark & Muthén, 2009; Wang et al., 2017). AvePP indicates profile-specific classification accuracy and values greater than 0.70 indicate adequate separation of profiles (Nagin, 2005). Substantive meaningfulness, parsimony of profiles, and size of profile solutions (≥ 5% of the sample) were also considered in determining the optimal solution. Practicality is an important consideration in latent profile modeling, as these analyses often yield incongruence between fit statistics (Masyn, 2013).

Once the optimal profile solution was determined, the indicator variables for each of the latent classes were compared using the manual BCH procedure in order to examine between- and within-profile differences in family members’ reactions. Omnibus Wald tests in MPlus were used to examine between-profile differences (e.g., how mothers in Profile 1 differed from mothers in Profile 2) and within-profile differences (e.g., how mothers and fathers in Profile 1 differed from one another); when significant, pairwise comparisons were also conducted.

To examine individual and family predictors of profile membership, demographic variables (sexual orientation, gender identity, age, sex assigned at birth, sexual identity development milestones, LGBTQ family members, co-residence with family members, race, ethnicity, , and site of participation) were examined to test how these variables were associated with the probability of profile membership. These variables were included using the three-step BCH approach (Asparouhov & Muthén, 2014; Bolck et al., 2004), which offers an advantage in that latent profile estimation and the relationships with auxiliary variables are carried out in separate steps, rather than in a single step, which avoids biasing profile formation. Additionally, the three-step BCH approach allows uncertainty in profile assignment to stay in the model when auxiliary variables (covariates and distal outcomes) are added. In the first step, the LPA is estimated using only the selected indicator variables; second, the participants’ individual profile probabilities are used to create a “most likely profile” variable, accounting for individuals’ uncertainty in classification; third, training variables (BCH weights) were used to create the latent profiles, which was regressed onto auxiliary variables using multinomial logistic regression, while accounting for potential misclassification. Finally, to examine whether youth well-being varied by family reaction profile membership, profile-specific means (intercepts) of each outcome variable (depressive symptoms, self-esteem) were estimated in Mplus, controlling for demographic covariates. Omnibus Wald tests were used to examine whether means significantly differed by profile membership.

Results

Fit statistics AIC, BIC, and SABIC indicated that the 4-profile solution fit the data best (Table 1). These values were plotted relative to number of profile solutions, which revealed diminishing returns with the addition of profiles greater than 3 (see Figure S1 in supplementary material). The BLRT indicated that each successive profile fit significantly better than the preceding solution and was therefore non-informative. The VLMR-LRT indicated that the 2-profile solution had superior fit to that of the 1-profile solution, but the addition of subsequent profiles did not improve fit. Solutions with greater than 3 profiles yielded categories with small percentages of participants (< 5%), which may be an anomalous result and not a stable, observed pattern. Entropy and AvePP values also declined from the 3- to 4-profile solution, indicating that the overall and profile-specific classifications became less accurate or distinct with the addition of more profiles. The 3-profile solution provided a parsimonious and interpretable solution, had a robust number of participants in each profile, acceptable entropy, and excellent AvePP (> .90). Therefore, the 3-profile solution was selected as the optimal fit for the data.

Table 1.

Fit Statistics for Family Members’ Current Reaction Latent Profile Analysis (n = 447)

K	AIC	BIC	SABIC	df	VLMR-LRT	BLRT	Entropy
1	6094.733	6130.145	6104.746	8	-	-	-
2	5394.910	5452.455	5411.182	13	p < .001	p < .001	0.918
3	4749.740	4829.416	4772.270	18	p = .006	p < .001	0.918
4	4557.456	4659.265	4586.244	23	p = .379	p < .001	0.929

Note. K = number of profiles. AIC = Akaike information criterion. BIC = Bayesian information criterion. SABIC = Sample size-adjusted Bayesian information criterion. VLMR-LRT = Vuong-Lo-Mendell-Rubin likelihood ratio test. BLRT = Bootstrapped likelihood ratio test. Selected solution is bolded for emphasis.

Profile Descriptions

Standardized item response means were plotted by profile for interpretation (Figure 1). Between-profile (Table 2) and within-profile (Table S2) comparisons were also examined.

Figure 1. Perceived Family Member Reaction by Profile.

Note. Scores are standardized. Profile 1: n = 75; Profile 2: n = 220; Profile 3: n = 152. Higher scores reflect more positive reactions.

Table 2.

Between-Profile Comparisons of Family Member Reactions

Family member	Profile 1 (n = 75) M (SD)	Profile 2 (n = 220) M (SD)	Profile 3 (n = 152) M (SD)	Overall group comparison1	p-value
Mother	1.69 (0.73) A	2.86 (0.54) B	3.61 (0.59) C	269.59	< .001
Father	1.76 (0.76) A	2.71 (0.52) B	3.66 (0.53) C	300.16	< .001
Brother	1.53 (0.23) A	2.99 (0.01) B	4.00 (0.00) C	14139.68	< .001
Sister	1.97 (0.79) A	3.09 (0.21) B	3.96 (0.05) C	830.19	< .001

Note.

¹Wald Test of Parameter Constraints (df = 2).

Significant group comparisons were probed using pairwise comparisons. Groups with different subscripts differed significantly on that variable p < .05. Scores presented in this table are raw scores. Profile 1 = Negative Family Reaction. Profile 2 = Moderately Positive Family Reaction. Profile 3 = Very Positive Family Reactions.

Negative Reaction (Profile 1)

The “Negative Reaction” profile (n = 75, 16.8%) characterized youth who reported negative reactions from all family members. Between-profile differences in family member reaction scores indicated that all family members had significantly more negative reactions than those in Profiles 2 and 3; standardized indicator scores indicated medium to large effect sizes for all between-profile comparisons (Cohen’s d ranged from 0.57 for sisters to 1.91 for mothers in Profiles 1 vs. 3). Within-profile differences in family members’ reactions indicated significant differences. Brothers’ reactions were rated as the most negative, followed by sisters’ reactions, mothers’ reactions, and fathers’ reactions (least negative). The differences between brothers’ and sisters’ reactions, brothers’ and fathers’ reactions, brothers’ and mothers’ reactions, and fathers’ and sisters’ reactions, were statistically significant. Mothers’ and fathers’ reactions did not significantly differ, nor did sisters’ and mothers’ reactions.

Moderately Positive Reaction (Profile 2)

The “Moderately Positive Reaction” profile (n = 220, 49.2%), which was the largest, was characterized by youth whose family members reported somewhat positive reactions. Tests of between-profile differences indicated that reactions from all four family members were significantly more positive than family members in Profile 1 (d range: 0.94 for fathers to 1.69 for brothers) and significantly less positive than family members in Profile 3 (d range: 0.74 for mothers to 1.02 for sisters). Within-profile differences indicated that family members’ reactions did not significantly differ from one another.

Very Positive Reaction (Profile 3)

The “Very Positive Reaction” profile (n = 152, 34.0%) included youth who perceived very positive reactions from all family members. Between-profile difference tests revealed that all four family members in this profile had significantly more positive reactions than those in Profiles 1 (d range: 0.57 for sisters to 1.91 for mothers) and 2 (d range: 0.74 for mothers to 1.02 for sisters). Brothers’ reactions were rated as the most positive, followed by sisters’ reactions, fathers’ reactions, and mothers’ reactions (least positive). Mothers’ and fathers’ reactions did not statistically differ from one another, but all other within-profile comparisons were significant.

Predictors of Family Reaction Profiles

Univariate multinomial logistic regression was used to test whether profile membership systematically differed by individual and family sociodemographic factors. Categorical covariates (i.e., sexual orientation, gender identity, age, sex assigned at birth, race, ethnicity, living with family members, site of participation, having an LGBTQ family member) were effect coded and continuous covariates (i.e., age, milestones) were grand-mean centered. Given the exploratory nature of these analyses, predictors were entered individually. Odds ratios and 95% confidence intervals are reported in Table 3 for these analyses. Gay youth were more likely than bisexual youth to be in Profile 3 (Very Positive Reaction) relative to Profile 2 (Moderately Positive Reaction; OR = 1.64, p = .031), and less likely to be in either Profile 1 (Negative Reaction; OR = 0.55, p = .041) or Profile 2 (Moderately Positive Reaction; OR = 0.61, p =.031) relative to Profile 3 (Very Positive Reaction). Transgender youth were more likely than cisgender youth to be in Profile 1 (Negative Reaction) than in Profile 3 (Very Positive Reaction; OR = 1.53, p = .025). Youth assigned male at birth were more likely than youth assigned female at birth to be in Profile 1 (Negative Reaction) relative to Profile 2 (Moderately Positive Reaction; OR = 1.33, p = .039). Multiracial youth were more likely than White youth to be in Profile 2 (Moderately Positive Reaction) relative to Profile 1 (Negative Reaction; OR = 2.36, p = .015). Given an older age by one year, the chance of membership in Profile 1 (Negative Reaction) increased relative to Profile 2 (Moderately Positive Reaction; OR = 1.18, p = .031).

Table 3.

Demographic Predictors of Profile Membership

	Profile 1 (reference)				Profile 2 (reference)				Profile 3 (reference)
	Profile 2		Profile 3		Profile 1		Profile 3		Profile 1		Profile 2
Covariate	OR	[95% CI]	OR	[95% CI]	OR	[95% CI]	OR	[95% CI]	OR	[95% CI]	OR	[95% CI]
Sexual orientation (ref: Bisexual)
Lesbian	1.48	0.78, 2.81	1.75	0.89, 3.45	0.67	0.36, 1.28	1.18	0.73, 1.92	0.57	0.29, 1.12	0.85	0.52, 1.37
Gay	1.07	0.62, 1.86	1.83	1.03, 3.24	0.93	0.54, 1.61	1.70	1.11, 2.61	0.55	0.31, 0.97	0.59	0.38, 0.90
Heterosexual	0.83	0.29, 2.40	0.50	0.13, 1.88	1.21	0.42, 3.51	0.60	0.19, 1.89	2.00	0.53, 7.55	1.66	0.53, 5.20
Questioning	0.73	0.28, 1.96	0.67	0.22, 2.04	1.36	0.51, 3.64	0.91	0.35, 2.36	1.51	0.49, 4.62	1.11	0.42, 2.88
Gender identity (ref: Cisgender)
Transgender	0.51	0.29, 0.89	0.42	0.23, 0.79	1.97	1.12, 3.49	0.84	0.49, 1.43	2.36	1.26, 4.40	1.19	0.70, 2.04
Sex assigned at birth (ref: Female)
Male	0.53	0.33, 0.84	0.77	0.47, 1.25	1.90	1.20, 3.02	1.46	1.02, 2.08	1.30	0.80, 2.12	0.69	0.48, 0.98
Race (ref: White)
API	0.32	0.12, 0.84	0.49	0.20, 1.22	3.18	1.20, 8.45	1.56	0.60, 4.03	2.04	0.82, 5.09	0.64	0.25, 1.66
Black	1.08	0.69, 1.71	0.79	0.49, 1.26	0.92	0.59, 1.46	0.73	0.50, 1.06	1.27	0.80, 2.04	1.38	0.95, 2.01
NAAN	1.20	0.38, 3.85	1.48	0.47, 4.71	0.83	0.26, 2.67	1.23	0.54, 2.82	0.67	0.21, 2.14	0.81	0.36, 1.85
Multiracial	2.36	1.32, 4.21	1.75	0.97, 3.16	0.42	0.24, 0.76	0.74	0.50, 1.11	0.57	0.32, 1.04	1.35	0.91, 2.01
Missing	1.02	0.62, 1.70	1.01	0.61, 1.69	0.98	0.59, 1.63	0.99	0.66, 1.49	0.99	0.59, 1.65	1.01	0.67, 1.51
Ethnicity (ref: Not Latina/o/x)
Latina/o/x	0.75	0.55, 1.03	1.01	0.73, 1.41	1.33	0.98, 1.82	1.35	1.04, 1.75	0.99	0.71, 1.38	0.74	0.57, 0.97
Missing	1.01	0.56, 1.23	0.63	0.41, 0.98	1.20	0.81, 1.78	0.76	0.53, 1.08	1.59	1.03, 2.47	1.32	0.92, 1.89
Participation site (ref: Northeast)
Southwest	1.00	0.51, 1.99	1.64	0.83, 3.26	0.99	0.50, 1.97	1.64	0.99, 2.70	0.61	0.31, 1.21	0.61	0.37, 1.01
West Coast	1.17	0.71, 1.95	1.21	0.71, 2.06	0.85	0.51, 1.41	1.03	0.69, 1.53	0.83	0.48, 1.41	0.97	0.66, 1.44
Age	0.85	0.75, 0.96	0.89	0.78, 1.01	1.16	1.04, 1.34	1.04	0.95, 1.15	1.13	0.99, 1.29	0.96	0.87, 1.06
Age aware of LGBQ people	0.99	0.92, 1.07	0.94	0.87, 1.01	1.01	0.94, 1.09	0.95	0.90, 1.00	1.07	0.99, 1.15	1.06	1.00, 1.11
Years since awareness	0.97	0.91, 1.03	1.03	0.96, 1.10	1.03	0.97, 1.10	1.06	1.01, 1.12	0.97	0.91, 1.04	0.94	0.90, 0.99
Age self-identify as LGBQ	1.00	0.92, 1.10	0.96	0.88, 1.06	1.00	0.91, 1.09	0.96	0.91, 1.02	1.04	0.94, 1.14	1.04	0.99, 1.10
Years since self-identification	0.94	0.87, 1.02	0.99	0.92, 1.08	1.07	0.98, 1.15	1.06	1.00, 1.13	1.01	0.93, 1.09	0.94	0.89, 1.00
Age first disclosure as LGBQ	0.93	0.85, 1.03	0.87	0.78, 0.96	1.07	0.97, 1.18	0.93	0.87, 0.99	1.15	1.04, 1.28	1.08	1.01, 1.15
Years since first disclosure	1.00	0.91, 1.09	1.09	0.99, 1.20	1.01	0.92, 1.10	1.09	1.03, 1.17	0.92	0.84, 1.01	0.91	0.86, 0.98
Living with family members
Living with mother	1.84	1.15, 2.92	2.54	1.56, 4.14	0.55	0.34, 0.87	1.39	0.98, 1.97	0.39	0.24, 0.64	0.72	0.51, 1.02
Living with father	3.57	1.74, 7.36	3.20	1.52, 6.72	0.28	0.14, 0.58	0.90	0.60, 1.34	0.31	0.15, 0.66	1.12	0.75, 1.67
Living with sibling	2.27	1.24, 4.15	2.48	1.33, 4.61	0.44	0.24, 0.81	1.09	0.75, 1.61	0.40	0.22, 0.75	0.91	0.62, 1.34
LGBTQ parent or sibling (ref: No)	2.13	0.97, 4.66	2.98	1.36, 6.56	0.47	0.21, 1.03	1.40	0.89, 2.20	0.34	0.15, 0.74	0.72	0.45, 1.12

Note. OR = Odds ratio. Profile 1 = Negative Family Reaction. Profile 2 = Moderately Positive Family Reaction. Profile 3 = Very Positive Family Reactions.

“API” = Asian or Pacific Islander; “NAAN” = Native American or Alaska Native. Significant predictors (estimates p < .05) are bolded for emphasis.

Sexual identity development milestone variables were also associated with profile membership. Given an older age at first disclosure as LGBQ by one year, the chance of membership in Profile 1 (Negative Reaction) was greater relative to Profile 3 (Very Positive Reaction; OR = 1.15, p = .022). For every year since first disclosure, the chance of membership in Profile 3 (Very Positive Reaction) was greater relative to Profile 2 (Moderately Positive Reaction; OR = 1.09, p = .024). Youth who were currently co-residing with either their mother, father, or sibling(s) were more likely than those not residing with these family members to be in Profile 2 (Moderately Positive Reaction; for mothers, OR = 1.84, p = .032; fathers, OR = 3.57, p = .004; siblings, OR = 2.27, p = .026) or Profile 3 (Very Positive Reaction; for mothers, OR = 2.54, p = .002; fathers, OR = 3.20, p = .010; siblings, OR = 2.48, p = .016) relative to Profile 1 (Very Negative Reaction). Youth who had an LGBTQ parent or sibling were more likely than those without any LGBTQ family members to be in Profile 3 (Very Positive Reaction; OR = 2.98, p = .023) relative to Profile 1 (Very Negative Reaction).

Mean Differences in Youth Wellbeing by Family Reaction Profile

Mean differences in youth depressive symptoms and self-esteem were estimated and compared for each profile (see Supplemental Figure S2)—outcome scores were standardized. All variables significantly associated with profile membership were included as control variables. Means differed significantly across profiles for both depression and self-esteem. Pairwise comparisons were conducted to compare pairs of profiles for mean differences for each outcome (Table 4). Results indicated that youth in Profile 1 (Negative Reaction) had the highest average depressive symptoms; depressive symptoms in Profile 2 (Moderately Positive Reaction) and Profile 3 (Very Positive Reaction) did not differ. Average self-esteem was significantly highest in Profile 3 (Very Positive Reaction), followed by Profile 2 (Moderately Positive Reaction) and Profile 1 (Negative Reaction).

Table 4.

Mean Differences in Youth Well-being Between Family Reaction Profiles

	Profile 1 M (SD)	Profile 2 M (SD)	Profile 3 M (SD)	Overall group comparison1	p-value
Depressive symptoms	1.02(0.51)A	0.82(0.35)B	0.72(0.25)B	9.59	.008
Self-esteem	2.91(0.28)A	3.12(0.25)B	3.24(0.24)C	17.02	< .001

Note. Values for depressive symptoms and self-esteem for each profile are raw scores. Response scale for depression ranges from 0 – 3; scale for self-esteem ranges from 1 – 4.

¹Wald Test of Parameter Constraints (df = 2).

Significant group comparisons were probed using pairwise comparisons. Groups with different subscripts significantly differed on that variable p < .05. Profile 1 = Negative Family Reaction. Profile 2 = Moderately Positive Family Reaction. Profile 3 = Very Positive Family Reactions.

Discussion

Families are important to LGBTQ youths’ well-being, and family member reactions that are negative or rejecting of youths’ sexual orientation and gender identities can have a detrimental effect. The current study used a profile-based approach to reveal distinct family reaction patterns, as well as the sociodemographic predictors and youth well-being outcomes associated with such patterns. Findings from a latent profile analysis of perceived family members’ reactions yielded a three-profile model comprised of Very Positive, Moderately Positive, and Negative family reaction profiles. These results are consistent with known variability across families, yet also indicate that reactions tended to cohere within families, such that family members were perceived as generally being similar to one another. Importantly, sensitivity analyses with different configurations of family members (e.g., mother, father, sister; mother, sister, brother) were consistent with the three-profile model.

Variability and Coherence Among Family Reactions of LGBTQ Youth

Perceived reactions to youth LGBTQ identity varied considerably, though some family reactions were more common than others. Almost half (49.2%) of youth were in the Moderately Positive profile, followed in prevalence by the Very Positive profile (34.0%) and the Negative profile (16.8%). Between-profile comparisons confirmed that the profiles were distinct. Youth in the Moderately Positive profile perceived reactions short of very positive, but rather somewhere in-between; youth in the Very Positive profile perceived significantly more positive reactions for each family member compared to the youth in the other profiles; and youth in the Negative profile reported significantly less positive reactions for each family member compared to those in the other profiles.

Despite overall coherence across family members, within-profile analysis revealed some discrepancies in family members’ reactions. The particularly negative reactions from brothers and sisters in the Negative profile are especially striking, as most literature suggests that siblings are an important source of support for LGBTQ youth (Pistella et al., 2020; Toomey & Richardson, 2009), as in the Very Positive profile showing more positive sibling reactions compared to parent reactions. Our findings indicate that in families characterized by highly negative reactions, siblings are unlikely to be perceived as sources of support. These findings underscore the importance of considering different family members’ responses even within larger family patterns of support (Very Positive) or rejection (Negative). Youth with unsupportive parents may not be able to rely on siblings for support—indeed, siblings may be even less supportive than parents in rejecting family environments. Notably, there was less variation in family member reactions within the Moderately Positive profile, where all family members tended to be perceived uniformly as having lukewarm reactions. Within-profile tests also showed that in all three profiles, mother and father reactions did not significantly differ from one another, indicating similarity in youth-perceived parent reactions.

Predictors Associated with Profile Membership

Youth social positions, sexual identity development milestones, living arrangement, and having a parent or sibling with an LGBTQ identity significantly predicted profile membership and helped illuminate the characteristics of youth in each profile. Gay youth, cisgender youth, youth with a parent and/or sibling who held an LGBTQ identity, and youth who were younger at age of first disclosure as LGBQ were more likely to be in the Very Positive profile relative to the Negative profile; more years since first disclosure as LGBQ also distinguished the Very Positive profile from the Moderately Positive profile. Bisexual youth, youth assigned female at birth, and younger youth were more likely to be in the Moderately Positive profile relative to other profiles. Transgender youth, older youth, and youth who did not currently live with their mother, father, or siblings were more likely to be in the Negative profile relative to either of the other profiles; youth assigned male at birth and White youth (relative to multiracial youth) were more likely to be in the Negative profile relative to the Moderately Positive profile.

Results are largely consistent with previous work, which shows that among sexual and gender diverse youth, gay or lesbian youth and cisgender youth report the lowest levels of family rejection (Gamarel et al., 2020). Many transgender and gender diverse youth perceive rejecting reactions from mothers and fathers, which may persist over time (Grossman et al., 2005; 2021). Findings regarding youth age of disclosure differ somewhat from previous research—typically, younger age at disclosure is related to greater negative reactions from family members (Baiocco et al., 2015). However, our results also show that in addition to younger age at first disclosure, more time had passed since this disclosure (for youth in the Very Positive profile). Family reactions can change over time, with most changes trending toward greater acceptance in the years post-disclosure (Grossman et al., 2021; Hilton & Szymanski, 2014). Thus, it is possible that younger age at first disclosure does not always spell negative family reactions in the long-term, though longitudinal work is needed to establish temporal sequence. Relatedly, youth in the Negative profile were older and had disclosed their identities at older ages, which could indicate that youth may wait to disclose to family members because they expect negative reactions—and are confirmed in this expectation. Findings regarding differences between White and multiracial youth are novel. Although inconsistent with prior research findings that LGBTQ youth of color report more negative reactions from parents than White youth (Abreu et al., 2022), findings from the current study indicate that multiracial LGBTQ youth may experience family environments characterized by modestly more positive reactions than White LGBTQ youth. Youth who did not reside with family members were more likely to be in the Negative profile than any other profile. Living arrangement could be related to the negative reactions from family members, as previous research documents that family rejection is linked with neglect and parental abandonment, contributing to overrepresentation of LGBTQ youth in unstable living or housing circumstances (Baams et al., 2019). Finally, youth who reported that at least one other family member was LGBTQ were almost 3 times as likely to experience very positive family reactions than negative reactions. This finding joins a growing body of literature indicating the benefits of ties between LGBTQ people within families, who may provide support and camaraderie for youth as they navigate family relationships (Stone et al., 2022).

Family Reactions and Youth Well-Being

Profile membership was associated with significant differences in youth well-being. Previous work on this topic suggests that positive family reactions are protective against negative well-being outcomes and boost youth positive self-evaluations (Ryan et al., 2010), whereas family rejection predicts worse mental health and self-esteem (Ryan et al., 2009). Analysis of profile membership yielded largely consistent results in the current sample. Youth in the Very Positive profile had significantly higher average self-esteem than the other profiles, and youth in the Moderately Positive profile had higher average self-esteem than youth in the Negative profile. Additionally, youth in the Negative profile had significantly higher depressive symptoms than the youth in the Moderately Positive and Very Positive profiles, who did not significantly differ. It is possible that even lukewarm family reactions can protect against increased depressive symptoms, even if self-esteem is relatively lower. These findings are particularly striking given that sociodemographic characteristics were controlled in these analyses. Although directionality and causality cannot be implied from this cross-sectional study, results suggest that negative family reactions might negatively affect LGBTQ youth well-being, above and beyond the effects of other sociodemographic factors.

Limitations and Future Directions

The current study provides a much-needed update to the literature on family reactions to youth LGBTQ identity. Much of the research on this topic dates from the 1990s to the early 2000s, and it is possible that family reactions, or the meaning and implications of such reactions, have changed for more recent cohorts of youth. Thus, the current study provides a snapshot of youth perceptions of family reactions to their LGBTQ using data from a large, diverse sample of LGBTQ youth collected in 2011-2012. Another strength is the person-centered, profile-based analytic approach, which allowed for the creation of family-level reaction patterns based on four family members. This offered a more nuanced analysis of reactions than have previously been captured by aggregates or averages.

However, there are several methodological limitations. First, family members’ reactions were reported by the LGBTQ youth, not by the family member themselves. Though not uncommon in the literature (e.g., D’Augelli et al., 1998; 2008), the findings from the study thus reflect youth-perceived reactions, which may differ from what family members might report. Second, the study did not include a household roster detailing the number or age of siblings, or whether youth’s household included more than one mother or more than one father. Therefore, the analysis might not have captured the reactions of all siblings or parents. Third, although sensitivity analyses with different configurations of family members yielded the same three-profile model as those with four family members (mother, father, brother, sister), youth with other family configurations are still not represented in these patterns. An implication of the sampling approach is the limited generalizability of these findings to the entire population of LGBTQ youth, and more research is needed to understand the family experiences of LGBTQ youth living in family configurations that are not represented in these analyses (e.g., family reaction patterns for youth who do not have siblings, or have extended family members in their household). Fourth, youth included in the current study reported that all family members knew their LGBTQ identity, and the study implications cannot be extended to youth whose family members do not know their identities, or who have some family members who know and some who do not know. Finally, it will be important to replicate these findings with more contemporary samples to determine whether different patterns emerge. It is possible that family members’ acceptance of minoritized sexual identities has changed over time at the population level, perhaps as a result of changing attitudes towards and awareness of LGBTQ people. For instance, Watson et al. (2019) found that average parental acceptance and family connectedness sexual minority youth had decreased from 1998 – 2013; similar historical decreases may be relevant in the current study, and changes in patterns of acceptance are also possible. In spite of this limitation, the current study provides a much-needed update to the current literature.

Conclusion

The current study provides a novel contribution by investigating patterns—rather than an aggregate—of family members’ reactions to youths’ LGBTQ identity and how these reactions contribute to LGBTQ youths’ well-being. Results indicated that family members’ perceived reactions to tended to cohere within families such that youth perceived either negative, moderately positive, or very positive reactions from all members. Sociodemographic variables predicted profile membership and well-being outcomes significantly varied by profile. Results are limited in generalizability to LGBTQ youth who report on four family members (mother, father, brother, and sister), and more research is needed to understand the experiences of LGBTQ youth in more expansive family configurations. Despite this, results underscore the interconnectedness of family members and support established links between family members’ reactions to youth LGBTQ identity and well-being outcomes. These findings suggest that clinicians working with LGBTQ youth with less accepting families should be aware that these youth may face compounded stress from multiple members’ negative reactions and work with these youth to develop positive social relationships outside the family. They also underscore the need for evidence-based, effective family-based interventions to foster acceptance within families (Parker et al., 2018) and ultimately promote greater well-being for LGBTQ youth.