I have 2 groups, each with 3 subjects and each subject with 5 replicate values. I want to compare the means of the measured values between the two groups but I'm unsure how best to do this.

First I considered taking the mean of the 15 values from each group (3 subjects X 5 values each) and running an unpaired t-test, but this treats each of the 15 values as independent from one another when in reality they came in sets of 5.

So then I thought it made sense to take the mean of each subject's values and then take the mean of the three subjects' means to form the group mean. Then perform a t-test using that group mean. So basically taking the mean of a mean.

Is this a sensible way to analyze my data?

Thank you very much.