Minimizing Error Variance Introduced By Missing Data Routines in Survey Analysis

Abstract

Seven ways of dealing with the problem of missing data are examined in this paper in light of their expected consequences on the strength of association among treated variables. These seven approaches include the (1) delete cases, (2) delete items, (3) sample mean, (4) group mean, (5) pairwise r, (6) regression analysis, and (7) random assignment solutions.