Sluggish Cognitive Tempo in Psychiatrically Hospitalized Children: Factor Structure and Relations to Internalizing Symptoms, Social Problems, and Observed Behavioral Dysregulation

Measurement invariance was tested to investigate if CFA results were consistent between boys and girls. At the outset, our model demonstrated adequate configural invariance (χ ² = 1426.41, p < 0.001, χ ²/df = 1.75; CFI = 0.89; TLI = 0.88; RMSEA = 0.04, 90%CI 0.04, 0.05), suggesting that the pattern of free and fixed loadings was invariant across genders. Using the configural model as a baseline, we tested weak invariance of factor loadings across participant gender. Given our focus on SCT, we first tested whether constraining the factor loadings of the three SCT items to be equal across gender significantly worsened fit compared to a model in which these loadings were allowed to freely vary (i.e., partial invariance: Bryne et al. 1989). The constrained model was not significantly different from the unconstrained model (Δχ²(3) = 0.12, n.s.; ΔCFI = 0.00; RMSEA = 0.04), suggesting invariance for the SCT items. We next tested weak invariance across all factor loadings. Although the chi-square difference test was significant (Δχ²(30) = 45.27, p = 0.03), such tests may be too strict for testing full invariance given the large sample size and other indices suggesting invariance (ΔCFI = 0.00; RMSEA = 0.05 falls within the 95 % CI of the configural model). Examination of specific items indicated that item 24 (“Doesn’t eat well”) loaded on the DEP factor for boys (λ = 0.27) but not for girls (λ = 0.12). As a check, allowing this item to freely vary while constraining all others to be equal did, in fact, indicate weak invariance (Δχ²(29) = 41.37, n.s.; ΔCFI = 0.00; RMSEA = 0.05). Finally, tests of strong factorial invariance were conducted, with results suggesting invariant factor intercepts across gender (Δχ²(6) = 7.05, n.s.; ΔCFI = 0.00; RMSEA = 0.05). Given these results, the full model collapsed across gender is presented.