Quite off from cinyc's Google poll.
Guess we'll see who's right.
The answer will likely be somewhere in the middle. Mason-Dixon doesn't appear to have pushed undecideds as much as I did by omitting the answer entirely.
But I'd trust a Mason-Dixon poll over a one-question Google Consumer Survey poll any day. The one-question GCS polls are fraught with every possible methodological polling error - no real RV screen, no real LV screen, no real proof all respondents are from the state, etc., etc., etc. You get what you pay for.