Friday, March 03, 2006

Electoral Math Revisited

In today's State House News Weekly Roundup (link will expire), reporter Craig Sandler does the math, and with Christy Mihos in the picture, things do not look good for Lieutenant Governor Kerry Healey and the Massachusetts Republican party.

If Kerry Healey wins 85 percent of the Republican vote in November, and 15 percent of the Democratic vote, she'll have 16.6 percent of the total vote. To win 50.1 percent of the total vote Nov. 7, she'll have to earn 68.9 percent of the independent vote. This relies on the very reasonable assumption that the electorate will retain its 49 percent-independent, 38 percent-Democrat, 13 percent-Republican composition.

This sounds like a remarkably formidable feat - yet no Republican has failed to accomplish it in the past two decades, and no Democrat has managed to muster the paltry 35 percent among independents needed to win.
Sandler's math is right, but his assumptions are not necessarily as reasonable as he thinks. Readers of my blog know that according to exit polls, Republicans turn out at greater rates than Democrats or independents. In 2002, the composition of the electorate was 42 percent-independent, 38 percent-Democrat and 19 percent-Republican. You may think that this is a small difference, but it actually makes his second paragraph false. Again, according to exit polls in 1998, Scott Harshbarger got 37% among independents, higher than the 35% that Sandler thinks is enough for a Democratic victory. This discrepancy is due to the higher than expected levels of Republican turnout. In addition, Sandler overestimates the number of Democrats who have historically been willing to vote for the Republican. Again, in 1998, only three-out-of-four Democrats voted for the Democrat, not the 85% that he predicts. Now, of course, it's true that I'm basing all these results on exit polls, but the turnout pattern is similar in both 1998 and 2002, so either the exit polls were wrong in both years (and the 2004 presidential election), or we can assume that GOP turnout is better than what might be predicted just using the numbers of registered voters.

That said, the fact is that if Mihos becomes a credible candidate, the math does not look good for Republicans. By my estimates, the largest subgroup of voters in 2002 was Democratic voters who voted for Shannon O'Brien (28% of the electorate). The second largest bloc was unenrolled voters who voted for Mitt Romney (25% of the electorate). If that bloc splits with Mihos peeling off only a fifth of them (5% of the electorate) then, all else being equal, the Democrat will win.