Naila Smith posted on Wednesday, August 09, 2017 - 6:25 am
My colleagues and I are working on some analyses using GMM and GGMM. In our analyses, we rescaled the response scale of our key measure, which was originally continuous (i.e., 0-4) to a binary response scale (0 or 1). We then conduct GMM using this measure. In a previous post (http://www.statmodel.com/discussion/messages/14/976.html?1449626922), you advise against standardizing variables (z-scores) before conducting LGCM. We want to be extra cautious and find out the following:
1. Does your advice against standardization of variables extend to rescaling scores to a binary variable and then running a GMM? 2. Would you say the approach described above is statistically appropriate given a defensible theoretical rationale?