Mediation

TABLE OF CONTENTS:

15–A: Introduction and Learning Outcomes

Video (0:36)

15–B: Installation and Loading Packages

Note that when working through the .Rmd files, R packages will often need to be installed, and libraries loaded. Do NOT quit R while installing or loading packages and libraries. This can corrupt crucial files. Instead, you can STOP the process (by clicking the red stop sign in the upper right corner of the console) and then quit the program, or wait until the process has been completed.

Video (10:17)

15–C: Baron and Kenny (1986)

n a mediation analysis, we are examining whether the effect of our independent/predictor variable on our outcome variable operates through a third variable, the mediator. Mediation analysis investigates the potential mechanism of the relationship between the predictor and outcome.

Before running mediation analysis, you think your causal variable CAUSES a change in the mediator which then CAUSES a change in the outcome. Note the causal relationship here. As a result, the mediator must be a variable that can be manipulated by the researcher.

In this chapter and example, we will be using Baron and Kenny's (1986) mediation model. This is one of the original types of mediation analysis.

While Baron and Kenny's method has been considered fundamental to mediation analyses, it has also had it's critiques -- largely due to the fact that it uses cross-sectional data (Cole & Maxwell, 2003; Maxwell & Cole, 2007). There are newer methods, including structural equation modeling (Hayes, 2009), that have become popular for conducting mediation analyses, but other methods like the Sobel test and bootstrapping are popular as well.

New methods and resources for conducting mediation analyses are still being developed as well! Check out this cool tool for doing mediation analyses in R (as well other statistical analysis programs like SPSS and SAS) here: https://www.processmacro.org/index.html

While there are still some critiques of Baron and Kenny's 1986 model and alternative methods have been proposed since, this method is still a great way to understand the fundamentals of what is happening in mediation -- so that is why we will be using this method in this chapter and lesson!

15–D: Baron and Kenny (1986) Mediation Steps

There are four steps in the Barron and Kenny model. The video below walks through this idea and we can additionally use the images included here to help us understand the relationships we will be investigating in this analysis.

STEP 1: Is the independent/predictor variable significantly related to the outcome variable? In other words, is there a relationship that can be mediated?

15–E: Research Question and Hypotheses

Video (1:56)

15–F: What is Dummy Coding?

Video (4:12)

15–G: Mediation Analyses

In the video below, you are walked through the mediation analysis. After conducting the analysis, we will have the coefficients for the paths (A, B, C and C') in our schematic model above (see section 14–D). We will fill these values into the model to more clearly interpret our results. Note the coefficients used here are only relevant to this example and each analyses will yield different results. Use this as practice for when you will interpret your own data.

Video (8:31)

15–H: Mediation Interpretation

There are three very clear results that you can have for mediation:

1) Steps 1, 2, 3, 4 are all not significant

• This would very clearly indicate that you do not have mediation within your results. If you do not have a relationship between the original predictor and criterion, there is no relationship to be mediated!

After walking through the analysis in the video above, we can fill in our own coefficients into the model and interpret the result. Recall that coefficients can be positive or negative. The directionality is important in interpreting our results, much in the same way it is important in linear regression analysis.

STEP 1: There is a signifiant relationship between note modality and exam score (b=-9.267, t=-5.40, p<0.001).

15–I: Mediation Final Write Up

Video (5:41)