Wednesday, November 4, 2015

Kentucky Governors, polls, voter turnout, and hockey jerseys

Matt Bevin was supposed to be Ken Cuccinelli, more or less. You remember how Cuccinelli, the former Virginia AG who ran for Virginia's governorship against seasoned Democratic politico Terry McAuliffe, was basically abandoned by the GOP because of his tea party-esque views, and ultimately lost in a-bit-closer-than-expected 2013 election? Truth be told, that Virginia election sent a scare through both the Democrats and the Republican establishment, as it seemed in the bag for McAuliffe a few weeks out. Polling data supported this idea, with McAuliffe leading basically every poll from September until the November election. The final RCP average had him almost seven points ahead. But the election was a nail-biter, with McAuliffe behind through most of the evening until late results from counties outside of DC pulled him back in front and gave him a margin of victory of two and a half points (and only 48% of the total vote, as a Libertarian candidate pulled in 6.5% and Cuccinelli 45.5%).

There was some fear going into last night's election that things might break similarly in Kentucky, as the tea party guy—Bevin—has consistently trailed in almost all polls, though by only a few points, and had received minimal support from his national party, though there was a very late injection of dollars into his campaign, to be fair. Regardless, his opponent—Democrat Jack Conway, former AG of the State, and seasoned politician—has had the full support of the national party from the get-go. And since World War II, the citizens of Kentucky have only elected Republicans twice before, in 2003 and in 1967. This was an election that the Democrats clearly expected to win, even if it was a little tighter than the polls were indicating.

Indeed, this lengthy analysis at FiveThirtyEight bears out the expectation of a Dem win, but with this small caveat:
This year, Bevin (a tea party outsider who ran against McConnell in the Republican Senate primary last year) has made a number of missteps on the campaign trail and has alienated many members of the Republican establishment. He’s been tagged as too extreme, including on Kentucky’s relatively popular expansion of Medicaid. Meanwhile, Conway, the state’s attorney general, is the definition of smooth. So far, these factors have worked together to give Conway a lead, but one that is small enough that Bevin could win.
Again, this is basically the same take experts had on the Virginia race in 2013: McAuliffe should win, but maybe, just maybe Cuccinelli could win, or at least make things hairy.

So what happened yesterday? Bevin won easily, garnering 52.5% of the vote, as compared to Conway's 43.8% and independent Drew Curtis's 3.7%. Note how very different this is from the Virginia results. In the case of the latter, some argued that the Libertarian candidate—Sarvis—cost Cuccinelli a victory, while others argued that Sarvis' participation is what made the race close (i.e., that Sarvis cost McAuliffe votes). In Kentucky, Bevin won outright, third party candidate or no third party candidate. Regardless though, Bevin cleaned Conway's clock with a winning margin of 8.7%. That's way outside the polls, way outside the analysis at FiveThirtyEight and elsewhere.

Which begs the question, why? How could the polls be so wrong?

The answer is probably a confluence of events. First, there is the shitstorm that is the Kim Davis Affair, then there is the overwhelmingly negative nature of the race (almost all ads by both candidates were attack ads), the issue of Obamacare (which, like Obama, is unpopular in Kentucky), and finally there is the issue of voter turnout and how it was likely impacted by all of the above.

As is the case in almost all off-year elections (election years with no national races for the Senate, the House, or the Presidency), turnout out was low. Kentucky has 3.2 million registered voters, yet only about 975,000 went to the polls yesterday for a turnout rate of 30.7%, slightly higher than in 2011 (the last gubernatorial election), but significantly lower than in 2003 and in 2007. And obviously, it was Conway who suffered in this regard,

Truth be told, Conway's campaign hinged primarily on the results in two of Kentucky's 120 counties, Jefferson County (Louisville) and Fayette County (Lexington). These two counties alone accounted for over 150,000 of Conway's near 427,000 votes (35%). And turnout in both looks to have been a little higher than the statewide average (35% and 33%, respectively). But it needed to be much higher, obviously. Looking just at Jefferson County, there are 538,023 registered voters there (as of 2014 statistics). Some 58% are registered as Democrats (which matches last night's results). The low turnout numbers cost Conway anywhere from fifty to one hundred thousand votes. In 2012, Obama received over 186,000 votes in Jefferson County. In this election, Conway received just over 112,000. That's a huge difference. And it's mirrored in Fayette County and a handful of other smaller ones where Conway scored victories.

But this was always the game from the beginning, Conway always needed huge numbers from these counties, yet there was this oft-repeated idea—buttressed by apparently deeply flawed polling data—that Conway could and would win, even as everything at a more micro level was stacked against him. Look at this piece from The Atlantic's Russell Berman from the later part of October. It's built around this assumption that things were breaking just right for Conway in Kentucky, but it's completely devoid of any real analysis to support this idea. Indeed, the FiveThirtyEight piece, while loaded with historical data, suffers in the same way: no hard, on-the-ground analysis.

I submit that the abundance of polling data available for political races is contributing to this kind of lazy, usually self-serving, analysis. Any serious look into the factors in play for the Kentucky race would have indicated just how difficult this race was going to be for Conway to win, that in fact an easy victory for Bevin was the only logical conclusion one might draw. Sure, I'm Monday morning quarterbacking here, but this stuff was available all along for all those opining on this race beforehand. And obviously, no one was interested in really looking, they were interested only in predicting and justifying.

Oh, and this is one of my favorite hockey jerseys, and it happens to be from Kentucky (like the person modeling it):

No comments:

Post a Comment