According to the NYT 2012 exit polls, Democrats won African-Americans 93-6, Hispanic-Americans 71-27, and Asian-Americans 73-26. These large leads are what allowed the Democrats to hold the white-house, as Obama lost White-Americans by a 20 point margin (59-39). Based on these statistics, the final map from the 2012 election should have looked like this:
So why doesn't it? The answer has multiple parts, one being turnout among different groups (with those percentages, Obama should have won by a margin 6.2 points higher then it was, at 53.6-44.6), another being that any poll you do will have a margin of error associated with, and another being that the votes were not uniform across the country. For example: in some areas whites voted left of their average, in some areas right. With these averages, we can look at how liberal or conservative each state was in 2012.
Key is this:
20% = -5-0%
40% = 0-5%
50% = 5-10%
60% = 10-15%
70% = 15-20%
80% = 20-25%
90% = 25-30%
So here you go. This last map is easier to look at, with the 90% states being places where the liberal/conservative number was above the 4% MOE, and the 40% states being places where the liberal/conservative number was below 4%, or within the MOE.
As such we can see that all things being equal, big states like California and Illinois really are not that liberal, but rather that their votes reflect the makeup of their electorate, and Florida and its fellow tossup states are only made tossup because their beliefs (or other reasons not looked at in these maps), and without that lean would fall safely into their opposite camp.
Note: this is part of a far bigger project that focuses on projecting how states will vote in the future, but I found these numbers interesting enough to post on their own.