While conducting research for my thesis, I have run into many twists and turns in my stats functions that have not been covered in my stats classes. The majority of my data is categorical. My dependent variable is an interval scale variable (with three levels), which I'd like to keep constant in my covariate tests. I'd like to know if my fixed factor variable (which is categorical) can have more than two levels. Additionally, can my covariate (also categorical) have more than two levels? The majority of my data consists of three or more levels.
Hello Alexander,
1. Yes. There's no upper limit to how many levels a factor may have in anova. However, more levels will result in a larger overall N of cases being required.
2. Yes. If you have a categorical variable with more than two levels, you'll need to recast it into L - 1 dummy variables (for L levels), and enter all the dummy variates at one time to represent the information for that covariate/control variable. If your control/covariate is interval strength, just include it as is, even if it has only a few possible values.
Good luck with your work.
Regarding the L-1 dummy variables (aka., indicator variables) that David described, your software very likely has a way to generate those indicators internally so that you do not have to compute new variables yourself. E.g., in Stata, one can use an i. prefix for categorical variables and c. for "continuous" variables.
In SPSS, many commands allow one to enter categorical variables in the "Fixed factor(s)" box (in the GUI dialog) and continuous variables in the "Covariate(s)" box. In the corresponding command syntax, categorical variables follow the keyword BY and continuous variables follow the keyword WITH. E.g.,
GLM Y BY v1 WITH v2...
would treat v1 as a categorical variable and v2 as a continuous variable.
HTH.
- Badges
- Science topic
- Similar topics
- Thesis Research