in reply to Re^3: standard deviation accuracy question in thread standard deviation accuracy question
the sample is the data set i have - it is drawn from the a potentially infinite data set known as the population
as we cannot collect an infinite (or in practice unfeasably large) amount of data we are almost never working with the population but with a sample drawn from it
the standard deviation given by excels STDEVP function asumes that the data set is an entire population and is used in those few instances when we are in fact working with the population which is why it isn't the default
as the sample size N increases these two functions should converge (as the ratio N / N -1 approaches 1)
the reason that i need the SAMPLE std dev is that i am using this to compute the standard error which is the standard deviation of all the differences between the means of all the different possible samples i could draw from the population and the population mean itself - it gives a measure of how reliable any given sample mean is as an approximation to the population mean i.e. it is a measure of the amount of noise in your data
it is (i believe) a consequence of the central limit theorem that the standard error can be computed from the standard deviation of the sample as stddev / sqrt(sample_size)
|