Which test for case controls

[Joe Tuckles]

Hello, I have a case-control study matched 1:1 that I just want to do a simple test to see which has a higher percentage of smokers, cases or controls and which has a higher mean BMI cases or controls. I do not want to test whether the smoking and BMI variables predict being a case or not (i.,e. predict the illness that defines being a case) I just want to see the differences between groups?

[Carlo Lazzaro]

Joe:
see -help cc-.

[Joe Tuckles]

Thanks so I will be doing odds ratios even though I don't want to test whether the variables predict being a case? It wouldn't be t-test to do what I'm looking at?

[Carlo Lazzaro]

Joe:
I'm not clear with what you're after.
Since you're dealing with proportions, -ttest- is out of debate here.
I would probably use -logistic- for what you're seemingly after (I would adjust BMI for further predictors, though):

logistic BMI i.group