ANOVA with failed cases
I have a question about ANOVA I was hoping I could get some advice on. I am doing a problem where I have 12 inputs and 2 outputs. Sometimes, a combination of two inputs is physically impossible, and I told it to output "case failed!" for me. I'm wondering now how that's going to affect my data and results. Running the cases takes a long time, and I would hate to do it again to try to avoid these bad regions.
I'm considering replacing the "case failed" with average output values to minimize the effect on the analysis, or maybe the worst observed output values, but is there a standard, statistically sound way to do this? It's a project I'd like to do by the books, if I can (Happy)
Thanks for your advice!