I am considering running a multiple regression on three independent variables: sex (Male=0, Female=1), experience level (0=novice teacher, 1=experienced teacher), and whether there is a history of trauma (0=no trauma, 1=trauma) to predict the secondary traumatic stress of teachers (continuous-level, scale score).
1. Can you run a multiple regression with multiple, unrelated categorical variables?
2. How would those results be interpreted?
3. Would a three-way ANOVA (2x2x2) be a better design?
Thank you in advance for your help!
Hi Silas
1. Can you run a multiple regression with multiple, unrelated categorical variables?
Yes, I can't see a problem with this.
2. How would those results be interpreted?
The regression coefficients tells you the difference between the means of each binary variable while controlling for the other variables in the model.
3. Would a three-way ANOVA (2x2x2) be a better design?
It depends - if all you are interested in is main effects then the regression approach is fine. If you are interested in potential interactions among your IV's then the ANOVA approach would take out some of the leg work involved in creating and testing the interactions in a regression model. It'll also produce nice plots.
If you do the regression models in mplus or other SEM software you get the benefit of using ML estimation - robust estimation and efficient handling of missing data (and other fancy things such as testing constraints if you want to).
Mark
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running