The Factorial Invariance Across Gender of Three Well-Supported Models

Abstract

Three well-supported latent models of posttraumatic stress disorder (PTSD; i.e., Emotional Numbing, Dysphoria, and Dysphoric Arousal [DA] models) were examined using factorial invariance across gender. There is a notable lack of studies that have investigated the factorial invariance of the PTSD models across gender, and, to date, no study has assessed the factorial invariance of the DA model across gender. The current sample consisted of 571 children and adolescent earthquake survivors (297 male and 274 female participants) from China. The results indicated that the 17 symptoms of PTSD have equivalent factor loadings between the male and female participants. Furthermore, the female participants evidenced more severe manifestations of PTSD. Moreover, the newly proposed five-factor DA model provided superior fit to the data compared with the Emotional Numbing and Dysphoria models. The implication of these results is discussed.

Elucidating dimensions of posttraumatic stress symptoms and their functional correlates in disaster-exposed adolescents

“…This model has provided an optimal representation of PTSD symptoms in trauma-exposed adults (Harpaz-Rotem et al, 2014; Pietrzak et al, 2012) and youth (Elhai et al, 2013). To date, virtually all research on the dysphoric arousal model in disaster-exposed youth has been conducted in Chinese adolescent earthquake survivors (e.g., Wang, Armour et al, 2013; Wang, Long et al, 2011). Our results extend work on the dysphoric arousal model in disaster-exposed youth to both a new cultural context and natural disaster.…”

“…We examined whether the best-fitting model was invariant across adolescent gender and younger (12–14 years) and older (15–17 years) age using a comprehensive invariance testing approach (Meredith and Teresi, 2006; Wang, Armour et al, 2013). We tested a series of models that imposed increasingly stringent restrictions across groups: 1) Configural invariance model (same factor structure across groups but all parameters allowed to vary); 2) Weak factorial/metric invariance model (factor loadings constrained to be equal across groups); 3) Strong factorial/scalar invariance model (observed variable thresholds constrained to be equal across groups to assess equivalence in item severity); 4) Strict factorial invariance model (residual error variances constrained to be equal across groups to assess equivalence in measurement error across groups); 5) Invariance of factor variances and covariances (factor variances and covariances constrained to be equal across groups); and 6) Invariance of factor means (factor means constrained to be equal across groups).…”

“…Findings are consistent with the adult literature and have supported the dysphoria and numbing models over the DSM-IV model (Ayer et al, 2011; Bennett et al, 2014; Elhai et al, 2013). Of the studies comparing the 4- and 5-factor models in youth, all have supported the dysphoric arousal model as the best-fitting model (Bennett et al, 2014; Elhai et al, 2013; Wang et al, 2012; Wang, Armour et al, 2013; Wang, Li et al, 2011; Wang, Long et al, 2011; Wang, Wang et al, 2013). …”

“…First, virtually all studies of the dysphoric arousal model in disaster-exposed youth have been conducted in Chinese adolescent earthquake survivors (e.g., Wang, Armour et al, 2013; Wang, Long et al, 2011). It is unknown whether extant findings apply to youth from other cultural backgrounds exposed to other types of natural disasters.…”

“…However, a recent study found evidence for some gender differences in factor loadings and factor variances/covariances of the dysphoric arousal model in a sample of juvenile-justice-involved youth (Bennett et al, 2014). Only one study each has examined measurement invariance in PTSD symptom structure in disaster-exposed youth across gender (Wang, Armour et al, 2013) and age (Anthony et al, 1999), with evidence for at least some degree of invariance across these characteristics. Third, understanding of how different PTSD dimensions relate to measures of functioning is limited.…”