Calculating 95% Confidence Intervals for Dependent Test Effect Sizes

[Jason Levitt]

Hi All! This forum has been amazing so far in answering my Stata questions, I am learning so much.

So, I am aware that the esize command is an easy way to calculate effect sizes and confidence intervals for independent t-tests. I was wondering if there was a way to calculate 95% confidence intervals for effect sizes based on dependent t-tests (which I am using other software to calculate)? I have all the available data (i.e. means, SD, group sizes), I am just unsure of an easy way for stata to calculate these. Any help would be greatly appreciated!

[Phil Bromiley]

You'll increase your chances of a helpful answer by following the FAQ on asking questions - provide Stata code in code delimiters, Stata output, and sample data using dataex.

Most of us only very rarely would use other software to calculate things, and almost never something as simple as a t-test.

esize has a confidence interval option that should do what you want. Actually, it appears to give confidence intervals automatically, You might be able to use margins after esize but I'm not sure.

[Jason Levitt]

You'll increase your chances of a helpful answer by following the FAQ on asking questions - provide Stata code in code delimiters, Stata output, and sample data using dataex.

Most of us only very rarely would use other software to calculate things, and almost never something as simple as a t-test.

esize has a confidence interval option that should do what you want. Actually, it appears to give confidence intervals automatically, You might be able to use margins after esize but I'm not sure.

Thanks for the reply and advice Phil! I normally use Stata for everything...except there doesn't seem to be an easy way to calculate effect sizes for dependent groups on Stata. I can do it by hand of course...but then I run into the problem of having to figure out another formula for 95% confidence intervals (a journal I am looking at wants them included). Stata is awesome at spitting out effect sizes with CI for independent tests, but not dependent.