50 points
Prior to conducting a statistical analysis using moderation or mediation, one must demonstrate understanding of the conceptual relationship between the variables to be tested. The following video describes how to develop conceptual models for moderator and mediator analyses based upon the hypotheses to be tested.
Choosing Your Variables
Moderator Analysis
Using ordinary least squares regression (the type of regression you are learning), the criterion variable (Y) must be a continuous variable (interval or ratio level of measurement). The predictor (X) should be (but is not required to be) continuous. This requirement will make your results easier to understand as you learn moderator analysis. The moderator (W) may be either continuous or nominal. If you use a continuous moderator, expect to probe the interaction using either the pick-a-point approach or Johnson-Neyman approach, as describe by Hayes (2022). If the moderator (W) is nominal and dichotomous (two groups), then no additional treatment is necessary. If the moderator (W) is nominal and multicategorical (three or more groups, such as a Race variable with groups White, African American, and Hispanic), then you will need to identify the variable as multicategorical so that PROCESS handles the variable correctly. Select Multicategorical in the PROCESS window (Figure 1), select the variable that is multicategorical–in this case, the moderator (W), then select OK (Figure 2).
Figure 1
Selecting Multicategorical in PROCESS
Figure 2
Selecting a Multicategorical Moderator
Although moderation may be performed with a nominal variable as either the predictor or the moderator, one should not attempt moderation where both variables are nominal. If this is the case, two-way ANOVA is more appropriate.
Mediation Analysis
Using ordinary least squares regression (the type of regression you are learning), the criterion variable (Y) must be a continuous variable (interval or ratio level of measurement). The predictor (X) should be (but is not required to be) continuous. This requirement will make your results easier to understand as you learn mediation analysis. The mediator (M) must be continuous. A common mistake students make is to attempt a mediation with a categorical variable as the mediator. Often, this does not make theoretical sense, such as one’s satisfaction with life having an effect on one’s gender. Statistically, this would also require logistic regression, which is outside of the scope of this course.
Submitting the assignment
When submitting this assignment, you must introduce what the reader is viewing. The write-up should include a statement of the problem statement, research questions, and hypothesis being described by the model. It should then present a figure of the conceptual model. You must do this for two separate research questions. Be sure to use the variables identified in Week 2 Quiz: Pick Topic assignment. Please note you may consider including an additional continuous variable as the mediator for any tests of mediation. I demonstrated this in the sample document and in the video. See the sample document for a model submission. As a reminder, you must do your own work. Copy and pasting elements from the sample document may result in receiving a 0 on the assignment. All submissions must be a single Microsoft Word document.
