calculating the standard deviation of samples measured repeatedly

Home > Statistics Every Writer Should Know > The Stats Board > Discusssion

calculating the standard deviation of samples measured repeatedly
Message posted by Harleen on March 7, 2000 at 12:00 AM (ET)

I am trying to determine whether there is a difference between a control group and a treatment group. Each group has a sample size of 4 and each sample was assayed (measured) three times each for a test value and a normalizer value. I have averaged the test values and divided them by the average of the normalized values for each sample, thereby giving me 4 normalized values in each group. From these 4 normalized values, I determined the mean and standard deviation to determine statistical significance between my control and treatment groups. The question is, should I take into account the intial experimental variability , ie the triplicates? or is this variabilty accounted for in my means? What is the best calculation of the dispersion of my data, ie standard deviation and how do I test it for statistical significance?

READERS RESPOND:
(In chronological order. Most recent at the bottom.)

Re: calculating the standard deviation of samples measured repeatedly
Message posted by Jack Tomsky on March 10, 2000 at 12:00 AM (ET)

You should always take into account the variance within runs (i.e., replications) as well as the variancce between runs. Depending on your instument there could be some drift over time, which would have to be modeled in. If this is not the case, then compare the overall means with a two-sample t-test. The variance of each overall mean is

s2(xbar) = (1/N)*[sb2 + sw2/k],

where sb2 is the variance between runs, sw2 is the variance within runs, and k is the (common) number of replicates for each run.

Jack

Your $5 contribution helps cover part the $500 annual cost of keeping this site online.