2004 Exit Poll:
http://www.cnn.com/ELECTION/2004/pages/results/states/IN/P/00/epolls.0.html14% self-identified liberals, 42% self-identified conservatives
2008 Exit Poll:
http://www.cnn.com/ELECTION/2008/results/polls/#val=INP00p120% self-identified liberals, 36% self-identified conservatives
That is a surreal change. It's not a category the mainstream media focuses on, but it is fundamental to voting tendencies, much more reliable than Party ID. The 20/36 ratio is not far from true swing state status, which would be about 21/34. For reference purposes, the nation was 21/34 in 2004, and 22/34 this year.
My eyes almost jumped out when I saw the 20/36. I've studied that category since '96 and almost nothing surprises me. A major move from one cycle to the next might be 2-3 points in each category. A 6 point shift in each is unprecedented, as far as I know.
At least it explains why my Excel model was off in Indiana. I forecast 18/38 this year, which indicated Obama needed at least an 8+ margin nationally to carry Indiana. At 20/36 it's completely understandable that a 6 point margin was enough. It puts Indiana alongside North Carolina and Missouri, the other coin flip states this cycle. North Carolina moved from 17/40 in 2004 to 19/37 this year, and Missouri remained stagnant at 19/36 in both 2004 and 2008.