Few saw it coming.
Leading into Tuesday’s elections, political prognosticators were predicting a close race for governor of Kentucky. If anything, public polling suggested a slight edge for Democrat Jack Conway.
But GOP candidate Matt Bevin defeated Conway by 9 percentage points, leading to elation for Republicans gathered in Louisville on Tuesday and deflation among Democrats watching results in Frankfort.
Upsets happen. But statewide polls — most notably the Bluegrass Poll, conducted by Survey USA in partnership with four state media outlets — were way off this election cycle, just as they were in Kentucky in 2014.
Less than a week before Election day, the Bluegrass Poll showed Conway up 5 points on Bevin. The result was a 14-point swing the other way.
Theories abound on why polling has apparently become so difficult in Kentucky. Dale Emmons, a Democratic political operative here, said it’s likely polls conducted in the state assumed Democrats would vote in greater numbers than they actually did.
“The real dilemma that all pollsters and political campaigns are having across the country — not just unique to Kentucky — is picking the right turnout model so that you can measure public sentiment that reflects that turnout,” Emmons said.
Then there is the innate weirdness of Kentucky politics.
The state has been experiencing a sharp decline in Democratic voter registration numbers, as well as a decline in Democrats voting for other Democrats. For years, the Democratic Party in the Bluegrass State has been able to rely on conservative rural voters. But that base is becoming less dependable, which also made sampling tricky.
[soundcloud url=”https://api.soundcloud.com/tracks/231615961″ params=”color=ff5500&auto_play=false&hide_related=false&show_comments=true&show_user=true&show_reposts=false” width=”100%” height=”166″ iframe=”true” /]
Harry Enten, a senior political writer and analyst at the data-centric FiveThirtyEight, said the number of undecided voters sampled in the polls also complicated things.
“It was that the Republicans did considerably better than the polling suggested,” Enten said. “So a lot of these people who said they were undecided may have gone and ended up voting Republican in unison.”
Emmons agreed that’s why polls were a little harder to put stock in this year.
“You know, you got to recognize a poll for what it is, which is a snapshot in time of where the electorate is at that moment,” he said. “With a double-digit undecided factor, there is a lot of uncertainty.”
Enten, an expert at analyzing polls, said we shouldn’t look at trouble with Kentucky polling and extrapolate that to other states or nationwide polling. But he said the lesson is not to take political polls “as gospel.”
Enten also said there was just not enough polling conducted in Kentucky to paint a complete picture on where the race stood.
This week wasn’t the first time polling was off-base in Kentucky. A year ago, polls showed Democrat Alison Lundergan Grimes trailing Republican Mitch McConnell by 5 percentage points just days before the general election.
McConnell’s margin of victory was three times that.
With that recent history, the state’s most prominent poll may soon change. This election cycle, the Bluegrass Poll showed a narrow edge for Conway; that was wrong. The Bluegrass Poll was conducted by Survey USA and backed by The Courier-Journal, WHAS-11, the Lexington Herald-Leader and WKYT in Lexington. In a note to readers, C-J Editor Neil Budde said the group would consider switching from Survey USA.
In his note, Budde noted modern polling has become more difficult; he cited “changes in the electorate” as the reason.