Mediation analysis of Likert scale through Baron and Kenny, booststrapping using Hayes and AMOS

Likert scale needs to be tested in terms of reliability using Cornbach alpha. Then, the overall composite score needs to be calculated to form a construct. After that, mediation analysis can be run using three different techniques, namely Baron and Kenny (1986), bootstrapping through Process Macro by Hayes, and analysis of moment structure (AMOS).

I. Baron and Kenny (1986)
Mediation is tested through three regressions:
1) Independent or predictor variable predicting the dependent variable or outcome variable.
2) Independent or predictor variable predicting the mediator.
3) Independent or predictor variable and mediator predicting the dependent variable or outcome variable.
The Sobel test can be computed from here: https://quantpsy.org/sobel/sobel.htm
II. Hayes Process Macro
Here is the link to the add on: http://processmacro.org/index.html
III. AMOS
Measurement model can be tested using exploratory and confirmatory factor analysis. Structural model can be tested using regression tests. Error terms and indices of model fitness, RMSEA and effects can be used to estimate the path A, B, C and C'.
Full mediation happens when the IV to DV has no significant impact, whereas the M to DV has a significant impact.
Partial mediation happens when the IV to DV has a significant impact; moreover, the M to DV has a significant impact.

#spss #mediation #Likertscale #BaronandKenny