email@example.com,Internet writes: >Interesting. I would argue the other side of the coin (chip?). I'm >not sure >what "independent" means here, but it seems reasonable to believe that >the >batches constitute "matched pairs" in that they are processed during >the same >hour. That is, if flaws are due to pre-baking production problems, >those >baked in the same hour should be essentially the same due to the random >assignment to the ovens, but hour-to-hour flaws may vary due to >pre-baking >problems. Thus, I favor the 1-t test, if a t-test is to be used, or >the Sign >Test for a nonparametric solution.
I was about to offer the same idea. I think students should get credit either way as long as make a reasonable argument for the choice they make.