Abstract:
|
The Monthly Wholesale Trade Survey (MWTS) provides estimates of change in inventories and sales for wholesale businesses in the United States. Previously, we generated an artificial population with frame data and two years of MWTS sales and inventory data, and saw how the current MWTS Horvitz-Thompson and random group estimators performed over repeated samples with no missing data. Inferential procedures (complete-data point estimates, variance estimates and interval estimates) did not behave as large-sample normal theory suggests they should. The sample estimates were highly skewed with poor confidence interval coverage. Here, we update our simulation study with a more realistic representation of the wholesale trade population and additional years of MWTS data and pursue a Bayesian method for estimating change in inventories and sales. This method makes better use of information obtained from covariates on the sampling frame, does not require large-sample normal approximations, and allows us to skip the imputation step altogether. It also preserves relationships among variables and marginal distributions and is attentive to major features of the sample design.
|