04/21/2008 11:01 pm ET Updated May 25, 2011

Final Update: Pennsylvania by Race, Education & Gender

On the eve of the Pennsylvania primary, here is one last update on the results by race, education and gender and measured by the Quinnipiac University surveys (and kindly shared with us courtesy of Quinnipiac polling director Doug Schwartz).

I have followed these results over the last several weeks for the same reason that ABC News polling director Gary Langer lists education at the top of his column today on "groups to watch" in Pennsylvania:

It’s hard see a single factor more compelling than socioeconomic status, particularly as defined by education. It’s split the Democratic electorate nearly all year, and as with her past victories, it’s what Hillary Clinton will be counting on tomorrow.

Years of education also split the Democratic electorate in past elections, such as 2000, 1992 and 1984, and survey researchers have known for decades that it has been one of the strongest predictors of racial tolerance. Yet amazingly, as per my post earlier today, at least seven of the Pennsylvania pollsters have released surveys that fail to ask or report any measure of income or education. Consider that omission when thinking about which polls to trust.

But I digress. Back to Langer's point about education:

Across primaries to date Obama’s won college graduates by 52-43 percent, while Clinton’s won less-educated voters by a very similar 52-42. The picture sharpens among whites only (there’s no difference by education among blacks): White college graduates have split 47-47 percent, while those with no college degree have gone 2-1 for Clinton, 60-31 percent.

The proportion of college-to-non-college voters isn’t always critical – Obama cruised among both groups in Wisconsin – but it’s mattered more often than not. Last month, in economically stressed Ohio, less-educated voters were in great supply (just 38 percent of white voters were college graduates, compared with an average of 52 percent across all primaries to date) and that helped Clinton immeasurably: She won less-educated whites by 71-27 percent, while her edge among white college graduates was just 52-45 percent.

The numbers that Langer cites above are from exit polls. In Ohio, the final Quinnipiac poll before the primary showed Clinton leading by a six-point margin (50% to 44%) among college educated whites and by more than 30 points (63% to 31%0 among non-college educated whites. Compare that to the numbers below, which include results from the latest Quinnipiac Pennsylvania survey released this morning.

Unlike Ohio, Obama has run ahead of Clinton among college educated whites by 7-9 points over the last three weeks. Clinton's margin among whites without a college education, however, has been roughly the same as in Ohio.

Hillary Clinton's favorable rating shows little or no change by subgroups over the last few weeks, with the possible exception of African Americans. While the subgroup is relatively small (no more than 140 interviews), Clinton's unfavorable rating has increased significantly (to 42%) from the levels measured in late March and early April (30 to 34%).

Similarly, Obama's favorable rating has been mostly stable in the Quinnipiac surveys over the last few weeks, although his positive rating among non-college educated white men has been slightly higher on the three surveys in April than the two surveys in March.