The result of step 3, the test for added components, indicated that the hypothetical model
was too parsimonious, i.e., important variables or parts were missing (Δ2 = 142.6 (cut-off =
18.3 with df = 10 and p = .05)). Free covariances between the latent variable disturbances
were therefore included in subsequent analyses to control for the influence of unmeasured
variables. Correlations at T2; T3; T4 between PC and SC disturbances were: .49; .55; .44 (all
p