Testing multilevel mediation using hierarchical linear models: Problems and solutions

Zhen Zhang, Michael J. Zyphur, Kristopher J. Preacher

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Scopus citations

Abstract

Testing multilevel mediation using hierarchical linear modeling (HLM) has gained tremendous popularity in recently years. However, biases could arise when no centering or grand-mean centering is used in these models. This study first summarizes three types of HLM-based multilevel mediation models, and then explains that in two types of these models, biases are produced when using current procedures of testing multilevel mediation. A Monte Carlo study was conducted to illustrate that HLM applied to grand-mean-centered data can under- or overestimate true mediational effects. Recommendations are provided with regard to the differentiation of within-group versus between-group mediation in multilevel settings.

Original languageEnglish (US)
Title of host publicationAcademy of Management 2008 Annual Meeting: The Questions We Ask, AOM 2008
StatePublished - 2008
Event68th Annual Meeting of the Academy of Management, AOM 2008 - Anaheim, CA, United States
Duration: Aug 8 2008Aug 13 2008

Other

Other68th Annual Meeting of the Academy of Management, AOM 2008
CountryUnited States
CityAnaheim, CA
Period8/8/088/13/08

    Fingerprint

Keywords

  • Hierarchical linear models
  • Mediation
  • Multilevel models

ASJC Scopus subject areas

  • Management Information Systems
  • Management of Technology and Innovation

Cite this

Zhang, Z., Zyphur, M. J., & Preacher, K. J. (2008). Testing multilevel mediation using hierarchical linear models: Problems and solutions. In Academy of Management 2008 Annual Meeting: The Questions We Ask, AOM 2008