Files
Abstract
This simulation study compares the effects of six procedures which impute values for missing items. An ESS list survey to estimate hog variables is used in this study. Specifically, the effects of the procedures on two survey variables--first quarter expected farrowings and second quarter expected farrowings--are analyzed under a range of conditions. These conditions include the percentage of missing items and the designation of which values are missing. Analysis of the mean square errors, the effects on the correlations. and the cost show that two versions of a ratio give the test results for very large sample sizes.