We have exit polls! – maplesmedia-llc.com

Data nerd that I am, I always want to dig into election data to see if there are trends, anomalies, or just interesting things to discover. And this year, thanks to a number of colleges and universities working together, we have actual exit polls to look at!

Who did the work

The 2019 Collaborative Kentucky Exit Poll was carried out by researchers at six schools:

Centre College
University of Pennsylvania
University of Kentucky
Morehead State University
Campbellsville University
University of Cincinnati

The work was jointly directed by Dr. Benjamin Knoll of Centre College and Drs. Michele Margolis and Dan Hopkins at the University of Pennsylvania. Other research collaborators included:

Drs. Jaclyn Johnson and Ryan Lloyd of Centre College
Dr. Michael Hail of Morehead State University
Drs. Josiah Marineau and Shawn Williams of Campbellsville University
Drs. Andrew Lewis and Jack Mewhirter of the University of Cincinnati
Dr. Mike Zillis of the University of Kentucky
David Azizi of the University of Pennsylvania

Where they did the surveys

The exit polling was done from 6 AM to 6 PM at selected locations in these counties:

Boone
Boyle
Campbell
Fayette
Kenton
Pulaski
Taylor
Rowan

Respondents were randomly selected by interviewers to participate in the survey. In all, 3,913 Kentucky voters participated in the exit poll. NOTE: This is an exit poll of voters only, and therefore these figures should not be interpreted as fully representative of all adults in Kentucky, but rather of 2019 Election Day. The margin of error for the full sample is ±2% and approximately ±3% for subsamples.

What the exit polls showed

If you’re already read Robert Kahne’s excellent analysis of the results of the governor’s race, the following will not be too surprising. The interesting things in these exit polls, to me, is the clear split in some demographic groups. These splits and trends may help inform strategy for races in the future.

First of all, here is the data presented in table form. The data should be read across each row; for example, of all males surveyed, 52.4% of them voted for Bevin, while 44.4% of them voted for Beshear.

Also, an important note: All the data is percentages, not total numbers! So, a bar being higher means a higher percentage, not necessarily a greater number. You’re comparing percentages of a group, but remember the groups can be different sizes.

	Bevin	Beshear	Hicks
Male	52.4%	44.4%	3.1%
Female	41.8%	56.7%	1.0%
White	49.5%	48.1%	2.1%
Black	14.3%	85.7%	0.0%
Other race/ethnicity	35.6%	62.3%	2.1%
> $50k	40.9%	56.1%	3.0%
$20k – $50k	49.1%	47.9%	2.5%
< $20k	47.9%	50.0%	1.7%
High school or less	53.3%	42.6%	3.4%
Some college	45.0%	53.9%	1.0%
College degree	38.6%	59.8%	1.5%
< 40	33.7%	62.1%	3.6%
40-64	51.9%	46.1%	1.9%
> 65	52.9%	46.7%	0.4%
Attend religious services weekly or more	58.7%	39.0%	1.9%
Attend religious services sometimes	45.9%	51.8%	1.6%
Attend religious services never	25.9%	71.1%	3.0%
Consider yourself an Evangelical or born-again Christian?	62.1%	35.8%	1.7%
Do NOT consider yourself an Evangelical or born-again Christian	30.3%	67.0%	2.7%

And, for everyone who sees data better in a graph, I broke out all the individual subgroups into their own graphs:

Obviously, you can look at these breakouts and draw your own conclusions. It would be interesting to compile a picture, for example, of the “typical” Bevin voter and the “typical” Beshear voter.

Of course, as we’ve all noted, the race was more about Bevin himself than about Republicans as a whole. Still, the breakouts may be instructive on building a coalition for the future.

The guv race compared to two other races

The exit polls also tracked how people voted in two other races. Again, these are interesting, especially as an indicator of the antipathy toward Bevin among Republicans.