I'm running a split-plot, 18 factor, mixed level DOE. There will be ~20 complex Y's as outputs. Each test will create a lot of output data for each Y variable. To simplify analysis, I will calculate scalar means and variances for each Y variable for each test. It seems straightforward to use scalar Y means data to perform analysis, such as Fit Model. Intuitively, just using the means data doesn't take advantage of the known variances of each test's means.
How do I perform analysis of my sampled means and include the effects of the variances, associated with each mean point?
Just use the mean for your response in one column and add the standard deviation for the Y as well. When you analyze pick your target for the mean (minimize, maximize, or target) and also your standard deviation (presumably minimize).