Assignment Methods in the Three-Form Planned Missing Data Design

Abstract

The 3-form planned missing data design allows researchers to measure more items from more participants on more occasions using the same budget as a complete data design. It also reduces burden, fatigue, and response reactivity. After randomly assigning participants to complete 1 of 3 forms on the first occasion, for subsequent measurements researchers might assign participants to the same or a different form, or use random assignment again. I discuss potential advantages and drawbacks of each approach, including a simulation to compare bias across methods. Results indicate negligible differences between assignment methods only in the absence of retest effects. Reduction of bias due to retest effects makes it preferable to systematically assign different forms over time.