Abstract:
|
Biased sampling problems appear in many areas of research, including, medicine, epidemiology and public health, social sciences and economics. Since Owen's (1988) seminal work on empirical likelihood, it has been getting popular in statistical and econometrical researches. Qin and Lawless (1994) showed that the empirical likelihood method for over-identified parameter problems is asymptotically equivalent to Hansen's (1982) generalized method of moments. In this talk we will use examples, including case and control studies with secondary outcome problems, utilization of auxiliary information in case and control studies, semiparametric exponential tilting genetic mixture models etc to show the applications of empirical likelihood to the biased sampling problems can produce very fruitful result.
|