Being Out Matters for the Measurement Equivalence of Family Acceptance and Rejection among LGBTQ+ Youth

• In Event: 1-101 - Measurement Validation and Equivalence in LGBTQ Youth Research: Examples and Recommendations

Integrative Statement

Family support is important for the healthy development of young people (Snapp et al., 2015). However, sexual and gender minority youth (SGMY) experience deficits in family support (Fish & Russell, 2018), particularly with regard to their sexual and gender identities (D’Augelli et al., 1998; Savin-Williams & Ream, 2003). Family acceptance and rejection of SGMY is associated with positive and negative outcomes, respectively (Ryan et al., 2009; Snapp et al., 2016). Many of these studies have focused on SGMY who have disclosed their identity to their parents or do not account for youth’s degree of disclosure. Youth who are out to family members report greater rejection than those who are not out (e.g., D’Augelli et al., 1998). Thus, depending on disclosure status, family experiences may be qualitatively and quantitatively different for SGMY and influence how SGMY respond to family support measures, which could lead to biased estimates of associations with health outcomes. In the current study, we conducted measurement invariance tests to determine whether measures of family rejection and acceptance systematically differ between those who are and who are not out to parents.

Data are from a larger online national sample (N=17,112) of 13-17-year-old SGMY who lived in the U.S. Youth were included if they were currently living with a parent. Parents knowledge of SGMY’s sexual or gender identity was assessed on a four-point scale from “none” to “all” (Mohr & Fassinger, 2000); we recoded responses to categorize SGMY as not out (n = 2825) or out (n = 7217). Family rejection and acceptance were two separate latent constructs consisting of four items (Table 1) adapted from items from the Family Acceptance Project (see Ryan et al., 2010). These constructs were simultaneously modeled and allowed to covary. We examined configural (unconstrained; overall patterns of factor loadings freely estimated across groups), metric (factor loadings constrained to equality), and scalar (intercepts constrained to equality) invariance between youth who were out or not out. We determined whether models passed measurement invariance based on model CFI (ΔCFI < .01) rather than chi-square differences due to large sample size (Chueng & Rensvold, 2002).

Overall model fit of the two-group model was acceptable (Table 2). The two-group model also passed metric equivalence, indicating that 1-unit increases in the latent constructs correspond to a similar increase in observed indicators for both groups. However, the model indicated a statistical decrease in model fit (ΔCFI = .016) when intercepts were constrained, suggesting that these measures do not accurately capture mean level experiences of family rejection and acceptance for youth who are not out relative to youth who are out to family. Family rejection intercepts were higher and acceptance intercepts lower for those not out to parents, despite parental unawareness of SGMY status. Results suggest that rewording or creating new family rejection and acceptance measures might be necessary for accurate comparisons between out and not out SMGY. Another solution might be partial invariance, in which some item intercepts or loadings are freely estimated to improve model fit, and will be examined.

Authors

• Amanda M Pollitt, University of Texas at Austin
  Presenting Author

• Jessica N. Fish, University of Maryland College Park
  Non-Presenting Author

• Ryan J Watson, University of Connecticut
  Non-Presenting Author