1
$\begingroup$

Part A: Suppose I have 2 groups (pre-treatment, post-treatment) and for each subject I record 3 variables. So my rows look like:

      pre             post
var1 var2 var3 | var1 var2 var3

I'd like to examine whether the treatment affects var1, var2 and var3. Can I simply do 3 paired t-tests ? Like: paired-t-test(var1-pre, var1-post), paired-t-test(var2-pre, var2-post), etc. Or is there some more appropriate test ?

Part B: Suppose that I also have a categorical variable like:

      pre             post
cat var1 var2 var3 | cat var1 var2 var3

(ofcourse since the subject is the same in each row 'cat' has the same value in two columns). What would be the most appropriate test in this case ?

Relevant question: Multiple paired samples t-tests?

$\endgroup$
1
$\begingroup$

I don't see why a paired t-test would be unsuitable, unless your data is not normally distributed. Your initial track of 3 paired t-tests seems right. For categorical paired tests, you can use the Mcnemar test. Should work in your case.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.