Earlier I was under the impression that experimental questions were oddball questions (i.e. something you probably haven't seen before). After taking PP it seems as though experimental questions are no different from other questions and often tend to be easier.
For example, I got an experimental question in Quant that was two equations and two unknowns (a simple subsitution problem). I thought that I had screwed up on a question before hand (thus giving me an easy question) but it turns out that it was experimental. I also noticed that experimental questions on PP started after # 10 and took a pattern (i.e. every other question during the middle part of the test).
I think the takehome message here is that if you get an easy question, it doesn't mean that you got the previous one wrong. Secondly, if you see a hard question, there's a 25% chance that it is experimental. Therefore, if you are able to narrow the question down through POE to two possible answers, the chances that you will get the question wrong AND have it count, are less than 50 percent (37.5%). So it's probably best to take those odds and move on without jeopardizing future questions.