Hello Everyone, I am familiar with the use of 2SLS to mimic SEM. I even used this approach once to estimate a model with a latent interaction term.
Does anyone know how to extend this approach to nested data? I have cancer patients with measures of cancer symptoms, functional impairment, and psychological distress taken at multiple points in time. The number of measures varies by patient and the intervals between measurements are uneven. In some ways, it doesn't seem like extending the 2SLS approach should be too difficult. One of my uncertainties though is how to assess model fit. For example, one of the rules of thumb I learned is that the R2 from all first stage regressions should be at least .10. It's not clear to me though what the correspnding criterion would be if I were running a mixed model. I also used a Basman F-statistic to assess model fit when a did a 2SLS. I no longer rember the details of its calculation, but wonder if it translates well to mixed models (or if there are other fit statistics that do). If anyone has knowledge of/experience with this and would be willing to point me in the right direction, that would be greatly appreciated. Thanks, Paul ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
