Methods for Moderation Analysis with Missing Data

Master's Thesis

Abstract

In the study, I consider a simple moderated multiple regression (MMR) model, where the effect of predictor X on the outcome Y is moderated by a moderator U. My primary interest is to find ways of estimating and testing the moderation effect with the existence of missing data. I mainly focus on cases when X and/or Y are missing completely at random (MCAR), missing at random (MAR) or missing depending on auxiliary variables (missing not at random; denoted AV-MNAR). Four methods are proposed and compared: (1) Listwise deletion; (2) Normal-distribution-based maximum likelihood estimation (NML); (3) Normal-distribution-based multiple imputation (NMI); and (4) Bayesian estimation (BE). Results from simulation studies show that the proposed methods had different relative performance depending on various factors. The factors are missing data mechanisms, population moderation effect sizes, sample sizes, missing data proportions, and distributions of predictor X. Influence of adding auxiliary variables is also discussed in terms of estimation accuracy for NML and NMI.

Attributes

Attribute NameValues
URN
  • etd-04122014-140759

Author Qian Zhang
Advisor Scott Maxwell
Contributor Lijuan (Peggy) Wang, Committee Member
Contributor Scott Maxwell, Committee Member
Contributor Ke-hai Yuan, Committee Member
Degree Level Master's Thesis
Degree Discipline Psychology
Degree Name MA
Defense Date
  • 2014-02-10

Submission Date 2014-04-12
Country
  • United States of America

Subject
  • moderated multiple regression

  • missing data

Publisher
  • University of Notre Dame

Language
  • English

Record Visibility and Access Public
Content License
  • All rights reserved

Departments and Units

Files

Please Note: You may encounter a delay before a download begins. Large or infrequently accessed files can take several minutes to retrieve from our archival storage system.