Gary Burns (Committee Member), David Lahuis (Advisor), Valerie Shalin (Committee Member), Debra Steele-Johnson (Committee Member)
Doctor of Philosophy (PhD)
This study examined a method for calculating the impact of multicollinearity on multilevel modeling. The major research questions concerned a) how the simulation design factors affect (multilevel variance inflation factor) MVIF, b) how MVIF affects standard errors of regression coefficients, and c) how MVIF affects significance of regression coefficients. Monte Carlo simulations were conducted to address these questions. Predictor relationships were manipulated in order to simulate multicollinearity. Findings indicate that a) increases in relationships among Level 1 predictors and also relationships among Level 2 predictors led to increased MVIF for those specific variables, b) as MVIF increases for a predictor, the standard errors for the regression coefficients also increase., and c) when MVIF values for the regression coefficients were 5 or higher, margins of error were around .20, and therefore any coefficients around .20 or lower will become non-significant.
Department or Program
Department of Psychology
Year Degree Awarded
Copyright 2013, all rights reserved. This open access ETD is published by Wright State University and OhioLINK.