Eins,
Your depependent variable consists of a weighted linear composite of several [presumably correlated] variables.
By employing a linear mixed model, you can test if there is a significant (1) difference in group centroids at baseline, (2) change from baseline to post-intervention per group centroid, and (3) difference in change in group centroids from baseline to post-intervention, after controlling for whatever variables deemed warranted.
Ryan
On Tue, Nov 27, 2012 at 1:12 AM, E. Bernardo
<[hidden email]> wrote:
There are two groups (experimental and control groups), each with n=100. Subjects are randomly assigned to experimental and control group. Pre-test was administered to each group, while posttest (same questionnaire used in the pretest) was administered after the intervention after several months. The test scores are latent variables. Considering that it is only a two-wave data, can we use latent growth curve(LCG) analysis for testing the following: (1) relationship between pretest score (intercept) and rate of change (slope) in scores(from pretest to posttest); (2) effect of the intervention program (experimental versus control) on the slope; and (3) effect of some demographic characteristics (e.g. gender, civil status) on the relationship between intervention program and
slope.
If LGC is not appropriate for two-wave data, please explain why?
Comments are welcome.
Eins
Locked
