Hi,

For background: i have 1 predictor (self-doubt), two mediators which were predictors in multiple regression (depletion in development, and growth in development) and a criterion well-being).

I ran a multiple regression analysis that provided me with B and β between predictors and the criterion. I ran a mediation analysis as these predictors were also highly correlated with one another (though VIF was low-moderate and tolerance was high). Three of four variables predicted the criterion. Using a theoretical rationale two of the variables were identified as mediators.

From this, when I added all significant predictors the B in hayes process model 6 between the predictor variable and mediators with the criterion was lower than suggested by the multiple regression. However, when i conducted the mediation analysis again, including these variables, they were the same as the multiple regression. How could this be interpreted?