1
$\begingroup$

I have two groups of an experiment, and I want to show first that they are similar in terms of demographics. One of the demographic factors is gender. I've looked for the most recommended way to compare the proportion of females in the two groups - I found some mixed answers about proportions comparison and wondered - why can't I just treat the IsFemale variable as a numeric (0/1), and than run independent 2 sample t-test on the groups (which will provide the differences between the %female)? Are there any risks in this method?

$\endgroup$
1

1 Answer 1

0
$\begingroup$

The two-sample t-test will allow you to assess whether there is a significant difference in the means of the two groups.

However, if you are using a t-test, you are assuming the distributions of both groups are close to normal. With your binary variable, this assumption will not hold.

$\endgroup$
2
  • $\begingroup$ Thanks, do you have a recomendation for this type of comparison? $\endgroup$ Commented Nov 24, 2017 at 9:13
  • $\begingroup$ Because your data doesn't have a distribution (and an associated variance), finding a valid statistical test is problematic... $\endgroup$
    – RussellB.
    Commented Nov 24, 2017 at 10:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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