PLACE:319 Snedecor Hall

SPEAKER:
Mike Elliot
University of  Michigan

TITLE:

Model-Based Alternatives to Trimming Survey Weights

ABSTRACT:

In sample surveys with unequal probabilities of inclusion, units are often weighted by the inverse of the probability of inclusion to avoid biased estimates of population quantities such as means (Horvitz and Thompson 1952). Highly disproportional sample designs yield large weights, which can result in weighted estimates that have a high variance. Weight trimming (Potter 1990, Kish 1992) reduces large weights to a fixed cutpoint value and adjusts weights below this value to maintain the untrimmed weight sum.  This approach reduces variance at the cost of introducing
some bias. An alternative approach (Holt and Smith 1979, Little 1991, Lazzeroni and Little 1998) uses random-effects models to induce shrinkage across weight strata. I compare these two approaches, and introduce extensions of each: a compound weight pooling model that allows model averaging over estimators based on different trimming points, and a weight smoothing model based on a non-parametric spline function for the underlying weight stratum means. The latter method performs well in simulations when compared with alternative estimators.  Methods are also applied to estimates of depression frequency and severity using data from the National Comorbidity Survey.
 
 

COFFEE: 3:45 p.m., 104 Snedecor Hall