Josie Booth posted on Wednesday, July 28, 2010 - 8:14 am
I'm running a mimic model and I have a question about the interpretation.
I have included a categorical predictor of two latent variables. The resulting path coefficients are negative. From my reading this seems to suggest that the group coded as 0 has a higher latent mean than the group coded as 1, however this doesn't seem to make sense as performance on the indicator variables are higher for the group coded 1.
Please could you clarify whether I am interpreting this correctly and if so why this might be.