My single mediation analyses (Hays Model 4) had sig. indirect effect, but my double mediation analysis with the same variables ( Hays Model 6) resulted in no sig. indirect effect (CI= - 0.02~0.0004). Why would this occur?
Adding the second mediator to the model can lead to changes in the other paths in the model, for example, because of redundancies (correlations among the variables that predict the outcome Y). This is the same as in conventional multiple regression analysis: Adding predictors to the model can change the other paths (regression slope coefficients) when independent variables are correlated. If you could post the results (the regression/path coefficients in each model as well as the correlation matrix of all 4 variables), it would be easier to say what's going on.
Another (or an additional) explanation for the differences could be missing data (i.e., the sample size with all 4 variables may be different from the sample size with only 3 variables). Are the sample sizes the same for both models?
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA