Each time a new Ohio Presidential poll comes out, I like to take a look at the cross tabs, specifically the breakdown of D,R,I. The breakdowns in the last four statewide elections were as follows (always D,R,I - from cnn.com):
- 2004 (Presidential) - 35,40,25
- 2006 (Average of Gov. and Sen. races) - 40,37,23
- 2008 (Presidential) - 39,31,30
- 2010 (Average of Gov. and Sen. races) - 36,36,28
Average those four together gets 37.4%,36.2%,26.4%. I expect that 2012 will be somewhere between 2006's D+3 and 2010's even.
With that in mind, I filter each new poll (that provides D,R,I breakdown) to get a sense for how much the sample skews things.
The 9-13 NBC/WSJ Poll, for example, has Obama (O)+7, but with a D+10 skew. That's right, they feel that Ohio will have a bigger partisan skew towards D than 2012.
When normalized based on the final breakdowns above, the race is:
- Tied using the average
- Romney (R)+1.8 using 2010
- O+2 using 2006
Similarly, recent Ohio Poll, PPP, and Columbus Dispatch Polls show a very close race.
It's important when looking at polls to understand that who the pollster surveys will greatly influence the results (not that that is anything new)