Analyses here are those described in Wood, P. K. (2011). Developmental models for children’s temperament: Alternatives to chronometric polynomial curves, Infant and Child Development, 20, 194-212.
A longitudinal data set composed of children's difficult temperament across five years is analyzed.
Simulated Data Only
Results of this analysis will not exactly match those reported in the paper, because those estimates are based on FIML estimates based on the raw data which cannot, for human subjects reasons, be released here. As a result, the analyses based here on simulated data which reproduce the variance/covariance matrix and means in the paper are only approximately equal to those reported in the article.
Orthogonal Linear Model in Mplus
In Mplus, it is possible to specify the linear model in its orthogonal form (i.e., rotated relative to the aperture of growth) using nonlinear constraints. Doing this in Amos would involve the use of phantom variables to accomplish the same result.