HW 08: Moderation and Stratification

Identifing different trends within subgroups


To start to motivate multiple regression, we need to explore how a third variable can muck up the relationship between a predictor and an outcome variable. This assignment will have you testing prior analyses for potential moderators.


  1. For each type of analysis (ANOVA, \(\chi^2\), Regression), decide on a third variable to test as a moderating variable.
    • The potential moderator can be different for each analyses. You need to decide what variables are most appropriate for your research question.
    • This third variable must be categorical and the fewer number of levels it has the easier it is to interpret.
  2. Adjust your bivariate visualization to account for this third variable (see ASCN Ch 2.7 for guidance and code).
    • Typical methods for plotting a third variable include paneling or coloring by the third variable.
    • State your thoughts about how the relationship between the explanatory and response variable has changed with levels of the moderator variable.
  3. Adjust your analysis hypothesis to include this third variable.
    • Is the relationship between X and Y the same for all levels of Z?
  4. For each analysis, run the original model, and stratified model.
  5. Determine if the third variable is a moderator or not and justify your answer.

See ASCN Ch 8.6 for a templated example.

Submission instructions

  • Download this [QMD] file and use it to write your answers in.
    • Right click and ‘save as’, put this in your scripts folder.
  • Render to pdf and upload your final PDF to Canvas by the due date.