The Morning After, Kentucky and Mississippi

Here’s a quick update regarding the gubernatorial polls we conducted from October 13-20, 2019, in Kentucky and Mississippi.

As you might guess, there’s lots of chatter about Kentucky, and about this poll that had Beshear winning the race for governor.

First, October polls are a dicey proposition due to the lower cooperation and contact rates that result from the noise of competitive political campaigns bombarding potential voters with messages. This is true for both online and phone polling. I should note cooperation rates are much better online than by phone.

That said, Mississippi went as we expected. Three weeks from election day, we had Tate Reeves up +3 among likely voters with 9% undecided. With leaners, we had it 47%/46% with 7% undecided.

In the end, undecided voters in a very red state went with the republican, Tate Reeves. The democratic candidate, State Attorney General Jim Hood, only picked up 1 point by election day.  

Mississippi

In Kentucky, we had the democratic candidate, State Attorney General Andy Beshear, cruising to a comfortable victory much like his father, former Kentucky Governor Steve Beshear, did in 2011.

As of this morning, Beshear has a very narrow lead over Republican incumbent Governor Matt Bevin.  

Kentucky

So, what happened in Kentucky?

By region, all of our numbers are in line with this morning’s unofficial results, with a couple of notable exceptions.

In the Louisville region, Bevin did much better than expected. We had 8% saying they were still undecided in the middle of October, and it appears all 8% went with Bevin in the end.

The Northern and Eastern regions of Kentucky account for about 34% of the state’s voters, and our response rates for the poll were lower than expected in these areas.

One of the key challenges for online polls is eliminating fraudulent responses. Response fraud is a form of cybercrime that’s a huge problem for public opinion research. Fraudsters in the U.S. and abroad try to participate in online surveys to get the monetary incentive offered for their time without providing genuine answers.

Our protocols to eliminate these fraudsters is extensive and uses the same technology as online banking websites. The downside of this security is lower contact rates which might have hurt us in this case, particularly in some northern counties Bevin carried such as Gallatin, Carroll, Bracken, and Robertson. As you can see below, Bevin performed better than expected in this region.

Kentucky by region

We were worried about releasing this poll. The Bevin numbers looked like an outlier, our sample size was on the low end from our typical targets, and there’s a history of polling misses in Kentucky, which has scared off a lot of groups from releasing anything in the state.

Mason Dixon had the race as a dead heat, and some party sponsored polls had substantial leads for Beshear, which were closer to our numbers. There were also rumors of internal polling that scared many on the Republican side last month.

After checking our numbers and looking at other data, we went ahead and released the results showing Beshear winning for several critical reasons.

First, there’s a big herding problem with public polls. There’s a tendency for polling firms to play it safe, produce results that mimic one another as we draw closer to election day. We didn’t want to contribute to this problem by withholding the Kentucky results.

Second, Bevin’s low ballot number mirrored his approval rating. Morning Consult had Bevin’s approval rating in Kentucky at 34%, which is in line with our ballot number for him. It’s also the second-lowest of any governor in the country. Bevin has also been attacked for supporting tolls for bridge improvements, which is a big deal in areas north of Louisville and would contribute to a low ballot number in that area. Also, his troubles with teachers has been well documented and teachers in Kentucky have stronger favorables than Bevin.

Third as we mentioned earlier, we looked at the 2011 race when Beshear’s father, Steve, won reelection by a margin of 56%/35% which is similar to what we saw in this poll.

So, for these reasons, we released the Kentucky results. We believed Beshear would win, and we were correct.

While it would have been easier to just withhold or adjust the results, our goal is to do the right thing, and I’m a “go big or go home” type, so we went with it. Not the easiest thing to do, but it was the right thing to do.

One other point. A lot has been said about the impact of Trump’s rallies on getting out the vote. His election eve rally occurred at Rupp Arena, which is a close drive from the areas where Bevin performed better than we saw in this poll. It’s entirely possible the rally could have saved Bevin from a more significant loss by convincing Republicans to stick with Bevin despite the issues he’s had with teachers, the legislature, members of his own party, etc. Turnout was much stronger than expected, which had to help Bevin keep pace with the enthusiasm of Democrats in the state.

After we see the final election results and do a deeper dive into the data, we will have more for you. I hope you’re all having a great Wednesday.

One more item. Go Vols! Beat Kentucky.