How are
(semi)parametric assumptions really used in statistical models?
•Choice of functional for comparisons
•Formula for computing
the estimate of the functional
•Distributional family for the estimate
•Mean-variance relationship across alternatives
•Shape of distribution for data