The research takes four steps to test the mediating effects. First, the independent variables
have to directly influence the dependent variable. Second, the independent variables have
to directly influence the proposed mediators. Third, the proposed mediators have to
directly influence the dependent variables. The fourth, and final, step involves a
comparison of the b values when both the independent variables and the mediating
variable are added to the regression equation. Partial mediation occurs when the mediator
reduces the size of the effect of the independent variables on the dependent variables. The
direction of the effects must also be as hypothesized. For full mediation, the significant
relationships between independent variables and dependent variables must become non-significant after the mediator is added to the analysis. The mediator used in this
research was employee outcomes, which was hypothesized to mediate the relationship
between HPWS and organizational performance.