I think what Monte is talking about here makes a lot of sense in a controlled, limited playtest. I can see how contamination could skew the results.
However, I also think you can counter this in an open playtest by going big. Get the largest sample sizes you can muster, and the larger...