Hello,
I am doing a quantitative study where I need to run a regression analysis via PROCESS Model 8 to test the evidence for moderated mediation. Before doing the analysis, I checked the assumption of regression to see whether they indeed hold. I have a strange-looking scatterplot. The variance errors are not funnel-shaped, but they are not randomly scattered either. They follow single lines. I have a dummy-coded independent variable with three levels. (0, 1 and 2) and a dummy-coded moderator with two levels (0 and 1). I have a mediator and a dependent variable with a continuous scale. I computed an interaction variable by multiplying the independent and moderator variables to do the assumption checks. I tried transforming the variables (IV, moderator, interaction and DV) using ln, log10, and square functions, but the graph shape did not change. I am unsure whether I need to check for homoscedasticity and whether there is evidence for homoscedasticity in the scatterplots. What should I do to solve this problem?
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Writing