Why Did the Pollsters Blow It on Kentucky?

 

https://www.youtube.com/watch?v=3ffqny2antM

Ladies and Gentlemen of Ricochet, I meant to spend the day taking you on a dreary tour of European Nazi-land, but I got so caught up and excited about following the results of the off-term elections that I didn’t write that post. I figure the Nazis will still be here tomorrow, though, and that you’ll forgive me for taking a detour, because these election results really are pretty interesting, no?

I’m not there, so you’ll have to answer all my questions about what happened. Here’s my first question. The Washington Post has a good round-up (although it’s not as funny or as good as ours, see below). I’m especially intrigued by Matt Bevin’s win in Kentucky. Congratulations, Governor Bevin!

Now, it sounds to me as if even Bevin didn’t really expect to win this. This report from Politico made it sound that way, anyway:

Bevin realizes no party leader has ever gone down in a primary; he knows the prevailing wisdom in Washington is that he has no chance, and that his only purpose is to damage McConnell heading into a tough general election against Democrat Alison Lundergan Grimes. No incumbent has a bigger target on his back this year than McConnell — a host of tea party groups and the entire Democratic Party both want nothing more than to defeat him — and Bevin’s performance could go a long way in determining whether the minority leader survives to serve a sixth term and possibly become majority leader.

Bevin credibly invokes a Horatio Alger-like life story on the trail. Grew up on a New Hampshire farm in a family of eight with a single toilet and a wood-burning stove to heat the three-bedroom home. Worked his way through college on an ROTC scholarship and rose to the rank of U.S. Army captain. Self-made millionaire businessman and investor. Father of nine, including four adopted from Ethiopia.

Yet none of those feats could have prepared the 47-year-old, tea party-backed Republican for the self-inflicted endurance test he is currently experiencing taking on McConnell.

Down 20-plus percentage points in polls — an improvement, Bevin notes in his characteristically upbeat manner, from the 40-point deficit he faced last summer — his immediate task is to convince people not that he will win the May 20 primary, just that it’s not totally inconceivable he could.

“Statistically, even now, it’s crazy long odds, but the tide is turning,” he said. “I’ve always been a risk taker, but I’m a calculated risk taker.”

“I’m going to be the Republican nominee,” McConnell countered flatly last week, when asked about Bevin and criticism that the 71-year-old Republican leader isn’t conservative enough for Kentucky GOP primary voters.

Asked if he wanted to weigh in on Bevin’s candidacy, McConnell said only: “I don’t.”

In the meantime, Bevin, who is prone to military metaphors when he talks about his campaign, is content to play the happy warrior — a first-time candidate with low expectations out to defy the naysayers and prove he can withstand the full brunt of the McConnell machine

But as WaPo reports, “Republican Matt Bevin won a big upset in the Kentucky governor’s race. The guy who Mitch McConnell crushed by 25 points in a 2014 primary will now become just the second Republican to govern the Bluegrass State in four decades.”

And here’s a part that intrigues me: “There should be more soul searching among pollsters, who once again got it wrong. Every survey of Kentucky showed Conway ahead, which meaningfully altered perceptions of the race.”

Every survey? Every one? So what’s up with this? Does it seem to you that the pollsters have been going awry more often than you’d expect these days? Why?

Red State’s asking some good questions about this:

1. How bad is public polling, or is Kentucky unusually hard to poll? Basically every public poll over the past several weeks had Bevin losing – even his campaign could do no better than muster an internal poll showing him tied (so did Vox Populi, a Republican-leaning pollster that had him at 44-44 and which got bragging rights in 2014 as one of the few pollsters showing anything like a competitive Virginia Senate race). But recall that polls consistently understated Sen. Mitch McConnell (R-KY) 54% margin of victory last year by a wide margin. Maybe we are headed to more catastrophic polling failures in 2016, or maybe Kentucky is just a tough state to poll.

If it is, why?

2. How unpopular is the GOP Establishment? Bevin, of course, was hated and scorned by many in the establishment after his primary challenge to McConnell last year, and feuded at times with the RGA over spending in this race. Then again, the RGA poured a lot more money into this race over the summer than Bevin did himself, and ended up with an ad blitz at the end, and both McConnell and Bevin showed the maturity and teamwork to set aside their bad blood and do a joint campaign appearance yesterday. The Louisiana Governor’s race on November 21 will provide another test –Sen. David Vitter (R-LA) 85%, unlike Bevin, is a member in good standing of McConnell’s caucus, and he seems to be in deep trouble there. If Vitter loses after Bevin won, that may underline the anti-incumbent mood, especially since it would mean both states tossing out the party in power.

3. Is school choice a winning issue? Americans for Prosperity ran this ad heavily in Democratic Louisville, which has a significant black population, and Josh Kraushaar noted on Twitter that “Internal polling showed it was #2 issue behind jobs” – Jack Conway underperformed past Democratic campaigns in Louisville:

(Kraushaar also notes that Bevin’s running mate, Jenean Hampton, will be the first African-American ever elected to statewide office in Kentucky).

5. Is there a religious-liberty backlash on the way? You may recall that the largest political story of 2015 in Kentucky was the fight over county clerk Kim Davis’ refusal to issue same-sex marriage licenses. Despite the fact that Davis was an elected Democrat, Beshear and Conway ran away from her like the plague, while Bevin embraced the idea that state officials can have some conscience protections, and Davis wound up smelling which way the wind was blowing and switching parties. Bevin otherwise ran a hard-edged social-issues campaign. Most of America isn’t Kentucky, where a pro-traditional-marriage amendment had passed with 75% of the vote in 2004, but clearly, the controversy didn’t hurt Bevin.

6. How much of a factor is Obama fatigue? Bevin worked hard to nationalize the race, and ads pounded the Democrats by tying them to a president who has never been popular in Kentucky. Vitter is doing more of the same in Louisiana, and if he survives there, it may show that down-ballot Democrats, at least in red states, simply aren’t going to escape Obama’s shadow as long as he’s in office. I’d be feeling pretty glum tonight if I was a candidate planning to oppose Sen. Rand Paul (R-KY) in his re-election race next fall.

7. How much do good candidates and political experience matter? Does Bevin herald the virtue of political amateurs? Maybe not so much, since his campaign was still a rocky one and perhaps just as importantly, it was his second go-round as a statewide candidate, so he had clearly learned some lessons from getting clobbered by McConnell and surviving an expensive and hotly-contested gubernatorial primary. And he was facing an opponent with a long track record of losing competitive races, in a state that is increasingly dark red. But naysayers who called Bevin an unelectable loose cannon were proven wrong.

So what lessons, if any, should we learn from this? WaPo warns: “But, but, but: Turnout was VERY low across the country. Republicans could over interpret the results at their own peril. The presidential election, exactly one year way, will have dramatically larger and more diverse turnout, which works against the GOP.”

Maybe. But maybe not?

Perhaps we should be asking the pollsters to tell us more about their methodology. And would any of our Kentucky denizens care to shed some light on this upset for the rest of us?

 

There are 33 comments.

Become a member to join the conversation. Or sign in if you're already a member.
  1. iWe Coolidge
    iWe
    @iWe
    • #1
  2. iWe Coolidge
    iWe
    @iWe

    You may want to edit: #5 repeats

    • #2
  3. iWe Coolidge
    iWe
    @iWe

    One is reminded that many right wing voters stayed home instead of voting for Romney. Higher turnout could go either way.

    • #3
  4. billy Inactive
    billy
    @billy

    Clearly, after this race and many other races in 2014, polling data isn’t as reliable as it once was.

    • #4
  5. Commodore BTC Inactive
    Commodore BTC
    @CommodoreBTC

    McConnell did not clobber Bevin. An incumbent senate majority leader barely able to muster 60% in his primary is an historically poor performance.

    Lots of theories, but the polling error usually seems to be toward the left (Israel, UK, US 2014/2015).

    As polling becomes increasingly unreliable, it will be replaced by decentralized prediction markets:

    • #5
  6. iWe Coolidge
    iWe
    @iWe

    billy:Clearly, after this race and many other races in 2014, polling data isn’t as reliable as it once was.

    This pleases me a great deal. Not just because things are more exciting when they are unpredictable, but because I instinctively like to believe that people are not automatons, and free will can triumph in the end.

    • #6
  7. The Reticulator Member
    The Reticulator
    @TheReticulator

    I don’t have an answer for your question, but I hope it is OK to use this space to (again) place a public service announcement:

    “Remember, it’s important to lie to pollsters.  It’s your civic duty.”

    • #7
  8. BrentB67 Inactive
    BrentB67
    @BrentB67

    So what lessons, if any, should we learn from this? WaPo warns: “But, but, but: Turnout was VERY low across the country. Republicans could over interpret the results at their own peril. The presidential election, exactly one year way, will have dramatically larger and more diverse turnout, which works against the GOP.”

    I hope some other folks comment on this point. I’ve read some very astute comments lately about how low HRC’s favorability is among some of the demographics that strongly supported and turned out for Obama.

    If 2016 turnout is anemic compared to 2012 and looks more like 2010, 14, and this election that seems to tip the outcome toward the right.

    Hoping to read some insightful analysis form others: James of England?

    • #8
  9. Claire Berlinski, Ed. Editor
    Claire Berlinski, Ed.
    @Claire

    iWe:You may want to edit: #5 repeats

    Done, thanks!

    • #9
  10. Frank Soto Contributor
    Frank Soto
    @FrankSoto

    iWe:

    billy:Clearly, after this race and many other races in 2014, polling data isn’t as reliable as it once was.

    This pleases me a great deal. Not just because things are more exciting when they are unpredictable, but because I instinctively like to believe that people are not automatons, and free will can triumph in the end.

    This isn’t true though.  Best evidence suggest that polling is, if anything, better than it has ever been, though getting more difficult to do.

    The reality is though pollsters do get races wrong sometimes (hence margins of error)  they are getting them right more often than in the past.

    When you break down into statewide races, in off year elections, turnout is low enough that correctly sampling and modeling the actual electorate becomes difficult.  Presidential polling has become incredibly accurate.  State races in off year elections are less so.

    • #10
  11. iWe Coolidge
    iWe
    @iWe

    Frank Soto: Best evidence suggest that polling is, if anything, better than it has ever been, though getting more difficult to do.

    The UK election was pretty amazingly “off” – bigger pollster error than any other in the history of the country as far as I am aware.

    • #11
  12. Marion Evans Inactive
    Marion Evans
    @MarionEvans

    Pollsters are businesses and they try to do it as cheaply as possible, which means using small samples. Because the electorate is now more fragmented than in the past, sampling size should be increased. Sample sizes (usually around 1000 people) which were ok for decades are now probably too small to be representative.

    • #12
  13. Frank Soto Contributor
    Frank Soto
    @FrankSoto

    iWe:

    Frank Soto: Best evidence suggest that polling is, if anything, better than it has ever been, though getting more difficult to do.

    The UK election was pretty amazingly “off” – bigger pollster error than any other in the history of the country as far as I am aware.

    Anecdote is not the singular of data.

    Here is fivethirtyeight on the Kentucky results.

    It’s not yet clear whether pollsters simply projected that more Democratic voters would show up than actually did or whether undecided voters broke overwhelmingly for the Republican candidates. The former suggests an electorate modeling problem that could be a big problem during the presidential primaries, when turnout is low. On the other hand, trouble modeling the electorate would be less of an issue in the 2016 general election, when turnout is at its highest.

    However, if undecided voters broke toward the Republicans and the “fundamentals” — Kentucky is a very Republican-leaning state in federal elections — that could be a sign that a candidate hitting 50 percent in general election polls is a bigger deal than we previously thought. Remember, polls showed Democrat Mark Warner winning the2014 Virginia Senate race easily with 50 percent of the vote; Warner earned almost exactly 50 percent, but he barely won.

    Whole article is here.  It goes into other details, such as though two polls dramatically overstated the democrat’s chances, one overstated the republican’s chances.  If there had been more than just 3 polls in the final days. it’s wholly possible that the average would have been fairly accurate.

    Polling is expensive though these days.  I suspect that this is a primary reason we are not getting enough polling of local races to get accurate pictures.

    • #13
  14. iWe Coolidge
    iWe
    @iWe

    Marion Evans:Pollsters are businesses and they try to do it as cheaply as possible, which means using small samples. Because the electorate is now more fragmented than in the past, sampling size should be increased. Sample sizes (usually around 1000 people) which were ok for decades are now probably too small to be representative.

    Not sure that a bigger sample size solves the problem.

    Many of us lie to pollsters.

    More of us hang up on all such calls, or are otherwise off the calling grid.

    The remaining pool may well be not representative – and that would hold whether it is 1,000 or 10,000 people.

    • #14
  15. Tom Riehl Inactive
    Tom Riehl
    @TrinityWaters

    All good commentary so far, but I haven’t seen the one reason I believe polls are so far off, lately.  Cell phone dominance is the devil for pollsters.  I, like billions of others, have forsaken a landline entirely.  My security is even wireless.  Exit polling is the remaining solid venue for catching the zeitgeist.

    • #15
  16. billy Inactive
    billy
    @billy

    iWe:

    Marion Evans:Pollsters are businesses and they try to do it as cheaply as possible, which means using small samples. Because the electorate is now more fragmented than in the past, sampling size should be increased. Sample sizes (usually around 1000 people) which were ok for decades are now probably too small to be representative.

    Not sure that a bigger sample size solves the problem.

    Many of us lie to pollsters.

    More of us hang up on all such calls, or are otherwise off the calling grid.

    The remaining pool may well be not representative – and that would hold whether it is 1,000 or 10,000 people.

    If I don’t recognize the number, I usually don’t answer the call. Most people I know do the same. The pollsters need to evolve past the era when phones were attached to the wall.

    • #16
  17. Casey Member
    Casey
    @Casey

    The problem isn’t with pollsters, it’s with how people understand polls.

    I get a kick out of reporters asking football players how they feel about being 6-point underdogs.  I wish one of these guys would explain to the reporter that the 6-point spread is about balancing bets and has nothing at all to do with football.  If the favored team wins by 6 that’s simply coincidence.

    Polls tell you the things polls tell you about but they don’t tell you much about the actual game on the field.  Sometimes they square up and sometimes they don’t.

    • #17
  18. The Reticulator Member
    The Reticulator
    @TheReticulator

    Frank Soto: Anecdote is not the singular of data.

    Yes, it is.

    • #18
  19. iWe Coolidge
    iWe
    @iWe
    • #19
  20. iWe Coolidge
    iWe
    @iWe

    iWe:

    billy: If I don’t recognize the number, I usually don’t answer the call.

    I answer the call. But I end all calls that are not important enough to take up my time. That spare time is saved for Ricochet, and then only as a background activity.

    • #20
  21. Percival Thatcher
    Percival
    @Percival

    Claire Berlinski, Ed.: Is there a religious-liberty backlash on the way?

    Nah.  Religious sentiment is down in the US.  You haven’t heard?  There’s a poll…

    • #21
  22. The Reticulator Member
    The Reticulator
    @TheReticulator

    iWe:

    billy:Clearly, after this race and many other races in 2014, polling data isn’t as reliable as it once was.

    This pleases me a great deal. Not just because things are more exciting when they are unpredictable, but because I instinctively like to believe that people are not automatons, and free will can triumph in the end.

    I also like it when politicians can be human beings rather than automatons whose every utterance is tuned to the polls.

    • #22
  23. Majestyk Contributor
    Majestyk
    @Majestyk

    A similar phenomenon is going on here in Louisiana, where Sen. David Vitter (who has a decent history of easily winning statewide elections) is facing off against John Bel Edwards for the Governorship in the upcoming runoff election.

    In the jungle primary Edwards and all Democrats managed to capture just 42% of the votes and all Republican Candidates functionally garnered the remainder.  Nonetheless, several polls have emerged which ballyhoo Edwards as being north of 50%.

    Color me skeptical.  Under the heading of “people lie to pollsters” I think they’re engaging in catharsis because Vitter has been plastering the airwaves with negative ads for some time and only just got over the threshold with 23% of the vote in a very low turnout (39%) environment.  One can only assume that if turnout follows the patterns that have existed historically we could see turnout as low as 30% given that this is an off-year election and there is no black or black-aligned candidate on the ballot.

    It’s my contention that Edwards is nonetheless toast, barring serious crossover from white Republican support, which doesn’t really strike me as plausible.

    People are just venting steam at Vitter to the pollsters.

    • #23
  24. Percival Thatcher
    Percival
    @Percival

    Tom Riehl:All good commentary so far, but I haven’t seen the one reason I believe polls are so far off, lately. Cell phone dominance is the devil for pollsters. I, like billions of others, have forsaken a landline entirely. My security is even wireless. Exit polling is the remaining solid venue for catching the zeitgeist.

    I don’t have a land line.  When I relocated, I maintained my old phone number, including the area code, on my cell phone.  To a pollster, I’m in the red part of “blue” Illinois instead of Iowa.

    • #24
  25. Fake John Galt Coolidge
    Fake John Galt
    @FakeJohnJaneGalt

    They did not blow it in KY.  It just happened that KY did not follow the reality that the pollsters were trying to create in the time they were given.  I have no doubt if given a few more months the pollsters would have delivered the Democratic victory they were seeking.

    • #25
  26. OkieSailor Member
    OkieSailor
    @OkieSailor

    We’re new in Kentucky, moved here last December but my impression of the campaign ads I saw were that Bevin’s ads were positive and gave some indication of how he intends to govern while the other side relied on spurious attacks and deceptive innuendo. I would like to hope that enough voters saw through that to give Bevin the win, but I’m a little sceptical of that. I do think it’s interesting that Bevin won by a much wider margin in the general than the primary (he won by 68 votes in the primary against a well-known GOP candidate).
    I think the major issues for the next few years will be jobs and addressing the unfunded liabilities of this state, mainly teacher and state employee retirement programs which are on a collision course with disaster. Bevin has outlined sensible plans to address these issues while Conway fell back on promising ever more ‘free stuff’ for more folks. Again I can hope that enough voters have seen through this scam to swing a close election, but I’m a little sceptical.
    What do I think made the difference? Barrack Obama. His incompentence, his arrogance, the cascading failure of his policies. I think that energized conservative voters and depressed the ‘liberal’ crowd. So I expect to see msm reports of how very low turnout gave Republicans the victory, but remember, folks it depends more WHOSE turnout is low.

    • #26
  27. Commodore BTC Inactive
    Commodore BTC
    @CommodoreBTC

    Bevin’s final pitch was very strong

    • #27
  28. David Knights Member
    David Knights
    @DavidKnights

    I’ve been a Bluegrass state resident for 30 years.  I even remember the last time KY elected a republican governor. (Yes, back when dinosaurs roamed the earth)

    Here are a few impressions.

    1. I was shocked that Bevin won and more shocked by the size of his win.  Bevin isn’t a particularly good candidate and supplied tons of bad soundbites that his opponent turned into ads.
    2. I want to know the final spending number, including PACs.  The ads on TV were running about 6 anti-Bevin ads to each pro-Bevin ad.
    3. While KY is red in national elections, it is reliably blue in state races, which makes it all the more shocking that Not only Bevin won, but that the GOP took the auditor and ag. commissioner posts.  This tells me it was more than just about Bevin and the Governor’s race. (BTW, the two state offices the Democrats did hold, Atty Gen and SOS, they held onto by bare margins.  These are usually lopsided wins for the Dems.)
    4. My impression is that all GOP candidates got a lot of support based on anti-Obama feelings.  I think the President’s rule-by-decree behavior has created a lot of bad feeling. (BTW, before anyone blames Kentucky rednecks hating a black president, keep in mind that our new Lt. Gov. is the first black elected to statewide office in KY EVER.
    • #28
  29. David Knights Member
    David Knights
    @DavidKnights

    As to why was the polling wrong, who knows?

    I suspect that with voter fragmentation, the rise of cell phones and caller ID, people lying to pollsters more (yes, it is your civic duty to do so.  Every pollster who gets in touch with me is told he is polling a female pacific islander who supports the socialist or communist candidate.) that polling is just so much more difficult than it use to be.  I think most pollsters are still using methods from 40 years ago and they haven’t yet figured out how to poll in the new environment.  Statistically accurate polling may no longer be possible.  I don’t know.

    In KY, given the higher number of landlines per capita than in most states, I’d have thought that the older polling methods might still work in KY more than other places.

    • #29
  30. OkieSailor Member
    OkieSailor
    @OkieSailor

    David Knights: In KY, given the higher number of landlines per capita than in most states, I’d have thought that the older polling methods might still work in KY more than other places.

    So it that turns out to be so then we should give polling in general even less  notice than we have lately, right?
    I can hope that means Mr. Trump is not the inevitable shoo-in he thinks he is ;)

    • #30

Comments are closed because this post is more than six months old. Please write a new post if you would like to continue this conversation.