Question 326:

Asked on August 7, 2008

Tags: t-test

1

Answer:

No answer provided yet.

For a t-test, you need two continuous variables, which can be ordinal, or scale as you put it. So you'd test say the time to take a test for group A versus the time to complete for group B.

If you have a nominal dependent variable you can't really use the t-test. It also depends on whether you are looking at difference between 1 group or multiple groups. For multiple groups you can use the ANOVA, but you can only use this on your dependent variable that is continuous or scale(ordinal), not nominal. 

To asses the other dependent variable that is nominal, you'd need to use a categorical test, this depends on how many values your nominal variable takes. If just two you could use logistic-regression(binary). I'd need to see a sample of your data and what your research question is to help you more.