Abstract:
|
There are many options available for the analysis of ordinal outcomes. The Proportional Odds Model, based on the logit link, is, perhaps, still the most often seen. Recently, a comparable model has been suggested based on the log link. Using the log link introduces constraints on the parameter space. Other constraints continue to translate from any link. We present results 1) on conditions for the uniqueness of the Maximum Likelihood Estimate (MLE) based on the rank of certain matrices, 2) about using constrOptim in R to determine the MLE 3) and some other fundamental results. We also offer some closed form expressions for the MLE and we offer some results with hypothesis tests in constrained spaces.
|