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