Inclusive Strategy in Missing Data Analyses: Inclusion of Interaction or Higher Order Terms of Auxiliary Variables

• In Event: Roundtable Session 117
  In Roundtable Session: Research on Accommodating Missing Data

Abstract

Auxiliary variables ( ) are strongly recommended to be included in missing data procedures so as to obtain better modeling performance (Collins, Schafer, & Kam, 2001; Graham, 2003, 2009). Yet strategies to include these are lacking. We investigated the possible advantages and disadvantages of different inclusive strategies and provided practical guidance for including during missing data procedures. Through simulation studies and comparison of models with different inclusive strategies with different missing data procedures, we concluded that the inclusion of interaction, quadratic terms of , or a mixture of both (non-linear inclusive strategy) was beneficial and probably necessary for missing data procedures. The theoretical implications of the non-linear inclusive strategy and practical limitations of the present study are also discussed.

Authors

• Xiaoxu Li, Peking University

• Xiaoyan Sun, The Chinese University of Hong Kong

• Kit-Tai Hau, Chinese University of Hong Kong