Did I hit a speed bump in my brain tonight?
I've seen nothing on this forum regarding a new WV poll by rather reliable Mason-Dixon. Yes, it was done for a private client. So, it's not (yet) on their web site (I got it off of PollingReport) and this may also bring into question its reliability somewhat. Still, it's Mason-Dixon... largely touted as one of the better firms.
What's more, based on its results, I feel like I missed some political development. In *most* polls Bush has led there... yes, in some it's been within the margin of error. But, it has been pretty consistent with the Bush lead. I had little question that Bush was in the driver's seat there. The Mason-Dixon results...
Kerry 47%, Bush 41%, MOE 4% (12% undecided). A small but definite lead for Kerry. I'd sure like to know who the private client is, to see if this is biased somehow. Nonetheless, I'm surprised by these results.
Excellent firm those M/D boys are, agreed.
A couple points:
Often if a poll is done for a private client, there are actually several different results produced by the polling firm, in fact it is quite standard to do so.
Democrats and Republicans have different likelyhoods of voting, and generally speaking, the higher the turnout, the better the Democrats do.
Based on this, M/D likely ran their usual (very good) voter screen and projected results, but then they would also run addition results as "what ifs"
What if the 70% most likely of voters actually voted...
the 75% most likely... etc...
Depending on which turnout was used, and which result "leaked" the result could vary quite radically.
If this is a "normal" turnout model it means something, if it a "100% turnout including folks who have been dead since the civil war " turnout model it means less.
Untill we see the "guts" it is hard to make a hard call on what the poll actually said.
Finally, even the very best can't get around statistical errors. Assuming a 600 sample size, 1 poll in 20 will report a lead that is wrong by 8%.
It is one poll, by a very good firm. I more likely to believe it, than not, especially given that I think WV is a natural Dem state, but it is, hey, one poll.