Premises of data gathering:
#1. I claim that criticals are fairly unambiguous. (There's some handwaving there.) If someone in the bathroom knows from the cheering that something big just happened, that's a crit.
#2. I claim that crits can be divided into four categories: good for the Steelers' score, bad for the Steelers' score, good for the Cardinals' score, bad for their score.
I'll try to record what I think the crits are, and which of those categories they fall into. Since I claim that crits are blatant and memorable, I think I can record the crits even if I don't have a notepad with me continuously.
Based on this methodology, I'll make these predictions:
#1. There will be 4 +/- 2 crits during the game.
#2. I predict that the "non-crit" score is 20 Steelers, 17 Cardinals, and that each crit is worth about 6 points. I claim that if you take that non-crit prediction of 20 and add or subtract 6 points per crit in the obvious way, the result will be within 4 points of the Steelers' final score.
#3. Ditto for the Cardinals.
#4. At least one of Kurt Warner, Larry Fitzgerald, or Troy Polamalu will score a critical success.
#5. Critical successes will outnumber critical failures.
I probably won't be correct with all five of these predictions, but if I score 3 out of 5, I'll regard this as a weak success, and if I score 4 out of 5, I'll regard this as a strong success.
Edit: I should add another claim: there will be at most one incident for which there's significant uncertainty over whether it's a crit or not. I think that I need to claim this separately from the 3 out of 5 of my previous claims; part of my claim here is that crits are qualitatively different from normal rolls.