I have a dissertation research question which involves what I am thinking is appropriately analyzed using 4-way ANOVA. The dependent variable is a math anxiety score (uses 5-point Likert scale). The independent variables, for this portion of the analysis, include categories: math/statistics and accounting/finance with criteria of failure (need to or did repeat course) and success (course grade of ‘C or better’ with or without first needing to repeat) for each.
I am thinking this is a four-way ANOVA using two levels for each factor: Math/Stats-Repeat: Yes or no ACC/FIN-Repeat: Yes or no Math/Stats-Success: Yes or no ACC/FIN-Success: Yes or no
Null hypothesis: There are no significant main effects due to failure or success in barrier courses (4 factors with 2 levels each: Repeat math or statistics (yes or no); Repeat accounting or finance (yes or no); Success math or statistics (yes or no); Success accounting or finance (yes or no)). Are there any significant interaction effects among barrier course experience factors? Null hypothesis: There are no significant interaction effects due to failure or success in barrier courses.
I submitted this note to my mentor about my thinking for this…Note on expected findings: Students who express they need to repeat math (for example) but have not yet repeated it might have an unusually high mathematics anxiety score which could show up as a significant interaction effect between repeat MAT/STAT and passing MAT/STAT.
Am I overcomplicating this situation, or should this really be handled in this manner? I am concerned collapsing will miss potential interactions. Also, will Minitab be a good tool for this analysis? I anticipate several hundred participants and unbalanced data.
