Sunday, June 10, 2012

The Relevance of Obama's Wisconsin Exit Poll Numbers to the General Election

Last Tuesday, Republican Scott Walker won a (to some) surprisingly large victory over Democrat Tom Barrett in Wisconsin's recall election, 53.1% to 46.5%. In the wake of this defeat, many liberal commentators turned to the silver lining of Obama's position in an exit poll question about the general election, where he led Romney by 7%. Analysts like Jay Cost and Sean Trende were quick to point out problems with the polling methodology. Another obvious problem with considering this to be a positive result is that Obama won Wisconsin by 14.1% of the two party vote in 2008. For reasons I may explain in a different post, I wouldn't actually compare the 7% to the 14%, but it got me interested in what the Wisconsin results would look like if Romney won a close race nationally.

The chart below shows John McCain's two party vote margin for all fifty states and DC in the 2008 general election. The independently allocated Congressional districts of Maine and Nebraska are depicted in addition to the whole state results.
This chart is a useful representation of the election results, because a shift in margin can be imagined as the zero line moving up or down while the vertical extent of the bars remains static.

The next chart shows just the states that McCain won, in decreasing order of two party vote margin, with the cumulative electoral vote haul graphed left to right. Electoral vote quantities represent 2012 values adjusted after the 2010 census (McCain actually won 173 electoral votes).
We'll use McCain's results as a baseline. Obama won the two party vote in 2008, 53.69% to 46.31%, a margin of almost 7.4%. If we assume that vote share (and margin) change uniformly throughout the states, Romney will have to improve to slightly better than even to win the 270th vote (and the 271st thru 275th). Colorado, which Obama won by 9.1% in 2008, holds the deciding votes.

The third chart shows Romney's performance if he is able to increase the Republican vote to about 50.9%, just enough to capture Colorado and therefore electoral majority. Again, only states he wins are shown, and the cumulative electoral vote total is given.
This shift adds North Carolina, Indiana, Nebraska's 2nd district, Florida, Ohio, and Virginia to the Republican column before Colorado flips. So what about Wisconsin? The final chart adds back in the states Romney would still expect to lose.
Our conclusion: using a vote model based on the 2008 results with a uniform shift, Romney can lose Wisconsin by 5% and still win the election. Of course, the electorate doesn't really work this way. Particular trends in the South and the Northeast (and everywhere else) will continually reshape the electoral landscape, even in the face of overarching national trends. But to the extent that these trends cancel each other out, or that they have already taken their toll on the electoral map, this gives us an idea of where Wisconsin sits relative to the other states in contention if public opinion has shifted towards Republicans since 2008. Suddenly, a June poll with questionable methodology showing a 7% lead for Obama doesn't sound like such good news.