# How to calculate sample size needed for comparing the “change from baseline” scores between two groups?

I'd like to know how to calculate the sample size needed for comparing the "change from baseline" scores between two groups?

-
More information will be required to answer this question. Specifically, how variable are the scores within each of the groups and/or what kinds of numbers are these? Normally distributed continuous numbers? % correct? –  rpierce Jan 21 '13 at 14:13

If the variable you are measuring is normally distributed, then you have two sets of measurements (one set of measurements at baseline, one set at some time point thereafter) that are related since they were obtained on the same objects/units (i.e. people). Thus you can use a t-test for dependent samples. There are many tutorials that will give you the formula for calculating your sample size.

-
The question does not provide much information but clearly states that the comparison of interest is between two groups. I don't see how a dependent sample T-test is useful in this situation. –  Gala Apr 21 '13 at 20:04
You are right in that the two groups are probably unrelated and so we can use a T-test for independent samples. If the first group represents baselines measurments and the second post-baselines measurments on the same people then we can use a dependent sample t-test. –  dandar Jun 13 '13 at 5:10