Large public surveys such as the Youth Risk Behavior Survey conducted by the Centers for Disease Control XXXXXXXXXXoften provide weights to be used in an analysis. Subpopulations may be oversampled to...

Large public surveys such as the Youth Risk Behavior Survey conducted by the Centers for Disease Control (2013) often provide weights to be used in an analysis.

Subpopulations may be oversampled to insure that the number of participants in a particular subpopulation is large enough to get estimates of the desired precision. For example, if estimates are required for each state separately, then states with smaller population would need to be oversampled relative to states with a large population to get the same precision. In combining data over subpopulations, should larger weight be given to observations from the oversampled subpopulation to those not in the oversampled subpopulation?