Optimizing Power for Planned Missing Data With Multiform Designs

• In Event: Measurement and Modeling Issues in Structural Equation Modeling

Abstract

This study explores the influence of four factors—subject assignment, percentage of missingness, missingness pattern types, and item/indicator numbers—on power for planned missing data with multiform design. Using a simulation study, we found that missing items within construct optimizes power, especially when items per construct increases. No appreciable difference between balanced and unbalanced missingness designs in terms of influencing power. When entire subsets are missing, increasing numbers of items for other constructs made little difference; however, with missingess spread across constructs, increasing numbers of items improves power notably. While percentage of missingness affected power predictably, it did not do so differentially across other conditions.

Authors

• Huili Liu, University of Maryland - College Park

• Gregory R. Hancock, University of Maryland