Electronic Theses and Dissertations

Identifier

873

Date

2013

Document Type

Dissertation

Degree Name

Doctor of Philosophy

Major

Ed Psychology and Research

Concentration

Educational Research

Committee Chair

Terry Ishitani

Committee Member

Jade Xu

Committee Member

Ernest Rakow

Committee Member

Corinna Ethingoton

Abstract

Modern approaches to estimation of mediation have evolved to create inconsistencies in the science. Researchers claim mediation effects with differential analytic techniques and discrepant considerations for the passage of time. This study compares and contrasts the results and substantive conclusions reached by four approaches to mediation estimation: (a) cross-sectional data with product of coefficients estimation mediation; (b) cross-sectional data with bootstrapping estimation of mediation; (c) longitudinal data with product of coefficients estimation; and (d) longitudinal data with bootstrapping estimates. An empirical example of multiple mediators in the relationship between neighborhood quality and child television exposure is provided. Findings were consistent between the product of coefficients and bootstrapping approaches. Variations between the cross-sectional and longitudinal approaches, however, were substantial. A significant mediation effect was identified cross-sectionally that was not significant, nor even the strongest of indirect effects, longitudinally. This is the first study to provide a real-world illustration of the consequences of divergent approaches to mediation estimation. In light of study findings, it is recommended that the use of the term "mediation" be discontinued as it relates to cross-sectional analyses.

Comments

Data is provided by the student.

Library Comment

Dissertation or thesis originally submitted to the local University of Memphis Electronic Theses & dissertation (ETD) Repository.

Download

Share

COinS