Mean versus Mediat Ct Values - Which to use for statistical analsyis (Oct/23/2009 )
I had just a quick question. So you run your qPCR in triplicate; does one typically use the mean or median Ct value from the triplicate experiment? It seems to me that the median Ct value would be less influenced by a potential outlier, thereby skirting the issue of "outlier analysis", whereas the mean would be very influenced by a potential outlier.
Finally, as far as use of error for statistical analysis, does the error from the Ct values run in triplicate need to be propagated through the equations (for me, I'm doing ChIP analysis where my final data is in the form of % input = 100 / 2^(-dCt_x)). I'll have about 5 samples per biological replicate, so if error has to be propagated from the Ct values to the % input value, how to you calculate the error in the biological replicate? Typically statistical tests (e.g, t-tests) don't have an input for the error inherent in each individual data point, but only an error associated with the entire biological replicate group.
weill I dont know much of statistics but as you rightly say mean is a better measure to calculate than median. Also depends how you are preparing the replicates. If it is different preparations then it accounts for all kibnd of errors but if it is from the same preparation, it just tells you about the pipetting error or the instrument variation!!
Also if you have set an acceptance criteria for the variation of the Mean Ct values, the error need not be corrected. Typically a 2% RSD is a good data and in which case you need not worry about correcting the error!!!