Anne Chan posted on Monday, April 26, 2010 - 6:16 am
Hello. I applied LGC in examining how 2 groups of students differed in their achievement (I used multigroup LGC). These are the analysis results:
(1) Both the slopes of their latent growth curve are insignificant. (I get this result from the LGC model) (2) The means of the achievement scores of the 2 groups are significantly different. The mean difference is 5 at time one, 3.5 at time two, 2 at time three and 1 at time four. (I get these results by comparing the chi-square of models with and without holding the means of the comparing pairs constant)
It seems that the gap between the 2 groups became narrower across time, but the slope of their latent growth curve in itself is not significant. So should I conclude the gap between the 2 groups became narrower (based on the mean comparison) or should I say there is no change regarding the gap between the 2 groups across time (based on the insignificant slope)?
It sounds like the mean of the slope growth factor is not siginficant for either group indicating that the development is flat in both groups. This is a finding about development. This can occur even if the outcome means are significantly different across time, for example, one group started significantly higher and stayed there and the other group started lower and stayed there. Note that the intercepts of the outcomes are estimated as part of the model not the means. If you want to compare means, you need to use MODEL CONSTRAINT to define them and MODEL TEST to test the differences between them. The mean of the outcome at time 5 with timescore 4 is: