Chunk #16 — Materials and Methods — Statistical analysis (see Supplementary Methods for details of data preparation) — Multigroup second‐order factor model (CT Study sample)
To be certain that the same trauma constructs were being measured in women and men, we tested whether the structure of the factors demonstrated sufficient measurement invariance (MI) with respect to sex (Meredith 1993; Lubke and Muthen 2004). Following the framework of Muthen and Muthen (Muthen and Muthen 1998), we fit a baseline model that allowed the factors' measurement structures to differ across the female and male groups. Specifically, the baseline model imposed the following constraints across the female and male groups: (i) thresholds and factor loadings were freely estimated for the first‐order factors; (ii) the loadings for the second‐order factor were also freely estimated; (iii) scale factors were set to one in both groups; (iv); intercepts (means) for the first‐order factors were set to zero in both groups and their variances were set to one; and (v) the mean and variance of the second‐order factor were set to zero and one, respectively. Then, we fit more restrictive (nested) models that equated the measurement structures across male and female groups (factor loadings and thresholds for the first‐order factors; factor loadings
