An application of mediation and moderation in health sciences
by
Sangeeta Agrawal
Niedfelt Nursing Research Center, College of Nursing, University of Nebraska Medical Center, Omaha, Nebraska, USA

Hierarchical regression analyses is used to identify mediating and moderating variables in Health Sciences. A moderating variable (as defined by Baron and Kenney, 1986) affects the direction or strength of relationship between independent and dependent variable. A mediating variable explains the relationship between Independent and Dependent variable by defusing the mechanism by which the former affects the latter. The first part of the paper explains steps involved in identifying significant mediating and moderating variables using hierarchical regression.

In the second part of the paper the mediating moderating role of self-efficacy variables will be examined in an effort to explain temporal changes in drinking behavior. 825 alcohol abusers were interviewed about their confidence, seriousness, importance and intention to quit drinking, at an assessment to assess their self-efficacy. These variables correlated significantly with drinking variables measured at 1-year pretreatment and 1-year post-treatment the Timeline follow back assessment method. A mediation - moderation model will be used-to identify role of self-efficacy in explaining temporal changes in drinking behavior.

Date received: October 25, 2002