A very popular article by Baron and Kenny (1986), later extended by Kenny, Kashy, and
Bolger (1998), recommended to social psychologists a test of mediation based on a set of steps involving correlations and regression weights. The serial published tests of mediation has come to be known as the Baron-Kenny approach. By the Baron-Kenny approach, a simple complete mediation is to be indicated which is a test of the direct path between an independent variable (X) and a dependent variable (Y) with a mediator variable (M) controlled is not significant. A simple mediation model has three correlations of their variables each. According to sequential regression analysis on a simple mediation model, a mediator M come after an independent variable X exist in the model, has a contribution of the mediator. Otherwise, sample size is a critical component to test as well as statistically significances. We argue the importance of investigating condition and interrelation of the three correlations, sequential contribution of the mediator, and sample size in the simple complete mediation cases by using hypotetical data generated by Microsoft Excel. We indicate some general consequences of simple complete mediation cases that are: (i) average of correlation XY is lower than average of correlation XM that lower than average of correlation MY; (ii) average contribution of mediator, indicated by R2 change, at interval of 23% up to 27%; (iii) distribution of effects X on Y when M controlled is influenced by sample size, the higher sample size, the lower distribution is; and (iv) average of mediation effects is at interval: 0.4 and above for levels of small sampel size (10 up to 40), between 0.2 and 0.4 for levels of medium sample size (50 up to 300), and under 0.2 for levels of large sample size (500 or above).
Copyrights © 2010