For females, change in depressive symptom levels was described by a strong positive slope and a negative quadratic whilst for males change consisted of a more moderate positive slope and a positive quadratic. For both sexes there was strong evidence of variation in symptom levels at baseline [female intercept variance = 0.74 (SE = 0.048), male intercept variance = 0.52 (SE = 0.045)], and also of variation in symptom change, described by the slope variance [female slope variance = 3.22 (SE = 0.373), male slope variance = 2.65 (SE = 0.330)]. Tests of parameter differences between males and females were carried out using a single-sample analysis, stratified by gender. In terms of the growth factor covariance matrix, there was strong evidence (p < 0.001) of greater intercept variance for females but little evidence (p ~ 0.2) for a difference in either slope variance or intercept/slope covariance. There was strong evidence (p < 0.001) for gender differences in all three growth factor means, as is apparent from the trajectory forms in Fig. 2. Results were consistent when varying the amount of missing data permitted.
