Dear All, can two set of data (say, 80 and 90, with "mean" = 85) define a "standard deviation"? I was told that at least 3 sets of data are required to have a "standard deviation". Is that correct?

Well, if you have only 2 set of data, the standard deviation will always be 0 and there is no point in finding the standard deviation of two sets of data, for both will be equally away from the mean.

Anyway, point is, you can find the standard deviation, but it would be meaningless.

So, 3 sets of data is indeed the minimum to get a 'useful' standard deviation.