Category Archives: Exit Poll

2016 Boyle County Exit Poll: Preliminary Results

The 2016 Boyle County Exit Poll was administered by Dr. Benjamin Knoll of Centre College to Boyle County, Kentucky voters on November 8, 2016. Respondents were randomly selected by interviewers to participate in the survey.

Centre College students enrolled in POL 210 (Introduction to American Politics) and POL 330 (Political Parties, Campaigns, and Elections) participated in administering the exit poll surveys. Mr. Ryan New of Boyle County High School provided generous assistance in the planning and execution of the project and his AP U.S. Government students at Boyle County High School also participated in administering the exit poll surveys. We also gratefully acknowledge Candace Wentz of Centre College and Julie Taylor of Boyle County High School for their technological assistance.

Students were on-site from 6:00 AM through 6:00 PM surveying voters as they left the polling locations on Tuesday, November 8th, 2016. Due to interviewer availability, all voting locations in Boyle County were included except for the Mitchellsburg location (which does not bias the results of the survey in a substantial way given that only about 3% of all Boyle County voters typically vote at this location).

In all, 1,724 Boyle County voters (including 1,191 self-reported from Danville) participated in the exit poll. The Kentucky Secretary of State’s website reports that 13,133 individuals voted in Boyle County on Election Day. Surveys were administered in both electronic and traditional paper format. In all, 88.8% of respondents participated with the electronic devices and 11.2% opted for paper surveys.

This year’s survey had a response rate of 48.4%. (We asked a total of 3,560 people to take the survey and 1,724 of them agreed.) This is similar to response rates in previous years of 47.5% in 2015, 47.4% in 2014, and 50.5% in 2012. This is similar to other national exit poll response rates

It should be noted that this is an exit poll of voters only, which are 60.8% of registered voters in Boyle County. Therefore, these figures should not be interpreted as fully representative of all adults in Boyle County, but rather 2016 Election Day voters in Boyle County, Kentucky.

Figures presented here are statistically weighted by gender, race/ethnicity, and age. This is a standard method to increase the representativeness of public opinion polling samples.

It is important to note that in recent years there has developed a bias toward over-representing Democratic respondents in the Boyle County Exit Poll survey, very likely due to Republican voters being increasingly less likely to agree to take the survey than Democrats. This pattern continued in this year’s survey. We can tell this by comparing survey answers to the official results. For example, according to the Secretary of State’s office, Donald Trump won Boyle County with 62.1% of the vote compared to Hillary Clinton’s 33.1% of the vote. After applying the standard demographic weighting, the survey showed Trump winning Boyle County with 48.8% compared to 44% for Clinton. There was a similar effect for results in the 54th Kentucky legislature race, and a smaller effect for Kentucky Senate voting patterns. However, differences between actual results and survey results for the non-partisan elections to Danville city commission and school board were much smaller, with an average difference of 1.5%. This strongly suggests that the Exit Poll is getting a representative sample of voters in terms of demographic characteristics, but under-sampling Republicans by a rate of about 12% this year. The accuracy of exit polling results relies heavily on the willingness of all groups of voters to participate in roughly equal degrees so that everyone has an equal chance of being included. To maximize the likelihood that this happens, we instruct our students to ask every third person to take the survey, thereby seeking to ensure randomization in sampling. It appears, however, that Boyle County Republicans are simply more likely to say “no” than are Democrats when approached by our students and asked to participate. Given the results of yesterday’s presidential election, it seems that this is a systematic problem throughout the entire country and is not isolated merely to Boyle County voters, as there is strong evidence that Trump voters were systematically less willing to participate in pre-election polling than Clinton voters.

Given this reality, I performed an additional calculation to attempt to estimate the likely survey result assuming that Republicans had been as likely to participate as Democrats. In other words, I took my “best guess” as to what the results would have been if Republicans were more evenly represented in the survey. Full details of this process are explained below.[FN1] In the interest of transparency, I present both the actual results and adjusted estimate of the results for these questions if the difference is more than 3%. I’ll leave it readers to determine which is a more accurate representation of opinions of Boyle County voters. Personally, I am inclined to accept the adjusted figures (presented in parentheses) as more accurate given the factors described above, but they should be interpreted as an estimation only. 

With that said, I present the “first pass” results of the survey. Throughout the week I anticipate posting more in-depth analyses of the survey results. Check back in the next few days for more.



Generally speaking, do you believe Danville/Boyle County is heading in the right direction or heading off on the wrong track?

  • 78.9% heading in the right direction, 15.8% off on the wrong track, 5.3% don’t know/no opinion

What is the single most important problem that our local area (Boyle County/Danville) needs to solve?

  • 28.1% Prescription drug abuse
  • 21.4% Jobs
  • 11.0% Crime
  • 10.4% Education
  • 9.3% Economic development
  • 4.1% DK/no opinion

Over the past year do you believe that Danville/Boyle County’s economy has generally…?

  • 32.5% gotten better, 42.9% stayed the same, 12.1% gotten worse, 12.5% DK/no opinion

Do you approve or disapprove of the proposal to repurpose the Tennessee Gas Pipeline (which runs through Boyle County) from natural to liquid gas and reversing the direction of flow to north-to-south?

  • 34.8% approve, 47.1% disapprove, 18.1% DK/no opinion (Adjusted estimate: 38.2% approve, 43.7% disapprove)

How effective is the partnership between the City of Danville and the EDP organization at promoting jobs and economic growth in the area?

  • 9% very effective, 50.8% somewhat effective, 13.3% somewhat not effective, 7.5% not effective, 19.4% DK/no opinion

How enthusiastic are you about the candidate you voted for President?

  • AMONG TRUMP VOTERS: 47.6% very enthusiastic, 27.3% somewhat enthusiastic, 16.9% not very enthusiastic, 8.2% not at all enthusiastic
  • AMONG CLINTON VOTERS: 48.4% very enthusiastic, 34.4% somewhat enthusiastic, 10.8% not very enthusiastic, 6.5% not at all enthusiastic

Do you approve or disapprove of the way that the following political leaders or community groups are handling their job?

  • President Obama: 46.7% approve, 50.6% disapprove, 2.7% DK/no opinion (Adjusted estimate: 35.4% approve, 61.2% disapprove)
  • Senator Rand Paul: 43.4% approve, 47.8% disapprove, 8.8% DK/no opinion (Adjusted estimate: 52.1% approve, 39.1% disapprove)
  • Senator McConnell: 24.3% approve, 67.2% disapprove, 8.5% DK/no opinion (Adjusted estimate: 29.5% approve, 62% disapprove)
  • Congressman Brett Guthrie: 39.3% approve, 28% disapprove, 32.7% DK/no opinion (Adjusted estimate: 43.3% approve, 24% disapprove)
  • Governor Matt Bevin: 41.1% approve, 46.5% disapprove, 12.4% DK/no opinion (Adjusted estimate: 50.4% approve, 37.2% disapprove)
  • Judge Harold McKinney: 61.1% approve, 15.5% disapprove, 23.5% DK/no opinion
  • Danville/Boyle EDP organization: 45.2% approve, 18% disapprove, 36.8% DK/no opinion



Do you approve or disapprove of the way that the following political leaders or community groups are handling their job?

  • Danville Mayor Mike Perros: 61.2% approve, 17.4% disapprove, 21.5% DK/no opinion
  • Danville City Commission: 63% approve, 12.6% disapprove, 24.4% DK/no opinion
  • Danville City Manager Ron Scott: 54.2% approve, 16.3% disapprove, 29.4% DK/no opinion

Do you approve or disapprove of Danville City Commission’s handling of the following issues:

  • Fostering economic growth: 63.6% approve, 21.6% disapprove, 14.8% DK/no opinion
  • Fostering a good quality of life for residents: 70.6% approve, 14.9% disapprove, 14.5% DK/no opinion
  • Management of the city budget: 55.5% approve, 25.1% disapprove, 19.4% DK/no opinion
  • Management of the water plant project: 61.7% approve, 20% disapprove, 18.2% DK/no opinion
  • Management of the Weisiger Park renovations: 53.6% approve, 27.6% disapprove, 18.9% DK/no opinion



Do you consider yourself: 25.5% liberal, 35.8% moderate, 35.6% conservative, 3.2% DK/no opinion

Do you think of yourself as a (an): 35.6% Democrat, 10.8% Independent lean Democrat, 6.5% Independent, 12.6% Independent lean Republican, 31% Republican, 3.5% DK/no opinion (Adjusted estimate: 34.3% Democrat or Independent-lean-Democrat, 55.7% Republican or Independent-lean-Republican)

Additionally, 53.6% of survey respondents were female, 46.4% male; 67.3% report an income over $50K/year while 13.7% report an income under $20K/year; 8.8% report never attending church and 29.5% report attending once a week or more; 88.1% report white ethnicity with 9.5% reporting African-American ethnicity; 35.5% report a high school education or less, 35.7% report college level of education, and 27.9% report a post-graduate level of education. 22% reported being under 35 years old, 35.2% reported being between 35-55 years old, 35.8% reported being between 55-75 years old, with 7% report being older than 75. Evangelical Protestants make up 44.4% of the sample, with 16.8% Mainline Protestants, 9.5% Catholic, 7% “nothing in particular,” 7.1% atheist/agnostic, 5.7% “other” religion.


[FN1] This is how I calculated the adjustment estimate based on partisan sampling differences: I took the two races with the most and least degree of partisan voting patterns: U.S. president and Danville School Board. The partisan correlation was 0.662 and the school board correlation was 0.09 (at most). This represents an upper and lower range to estimate partisan bias in other questions. I then calculated the correlation between partisanship and each other question. I subtracted 0.09 (the lower boundary correlation) from the correlation and then divided this by the upper boundary (0.662) and then multiplied that by the degree of partisan bias in the upper-boundary turnout figures (12.1% for U.S. presidency). I then either added or subtracted that amount based on the direction of the partisan correlation in the response question. For those questions in which the difference is greater than 3%, the new figure is presented as the “adjusted estimate” figure in parenthesis.


Update on 2016 Boyle County Exit Poll results

For interested parties: the data from this year’s exit polling project has been collected and compiled. I hope to be able finish the analysis and post the preliminary results sometime tomorrow (Wednesday). Thank you all for your patience!

Previewing the 2016 Boyle County Exit Poll



The exit polling boxes are ready to go!

As has been the case since 2011, next week on Election Day my Centre College students will be fielding exit poll surveys at voting locations throughout Boyle County. This year they will be joined by students from Ryan New’s AP U.S. Government course at Boyle County High School.

This year’s survey project will be largely the same as previous years, but with one key innovation: we are introducing electronic devices to field the exit poll surveys. Students will be collecting survey responses using iPad devices at each voting location. This will increase efficiency and decrease the time needed to input the survey data on the evening of election night. (We will still have paper surveys available as a backup or for those who would prefer that option.)

As before, I intend to publish the results of the survey here on this blog. I will post the initial results either late in the evening on Election Day or sometime the next day (Wednesday), depending on how long it takes to get everything compiled and analyzed.

As always, we look forward to the participation and involvement of Boyle County voters who have been great to work with in previous years.

2015 Boyle County Exit Poll: comparison of gubernatorial voting responses

Our exit poll showed that of those who answered the question on the survey, 48.8% reported voting for Bevin, 47.1% reported voting for Conway, and 4.2% reported voting for Curtis. According to the Secretary of State’s website, 54.7% of Boyle County voters voted for Bevin, 40.1% for Conway, and 4.6% for Curtis. That’s a difference of 5.9%, 7%, and 0.4%, respectively, and in a direction that over-states support for Conway and understates support for Bevin.

There are a few possible (and not mutually exclusive) explanations for this:

  1. It is possible that the 1.9% of voters who did not complete the question on the survey were mostly Bevin voters.
  2. It is possible that some Bevin voters indicated on the survey that they voted for Conway, despite voting for Bevin in actuality.
  3. It is possible that Bevin voters were less likely to agree to take the survey than Conway voters.

My strong hunch is that #3 is the most likely explanation, especially given that the 2014 Exit Poll had a 3-6% bias in favor of Democratic candidates (but not non-partisan candidates), although it is not possible to definitely prove this.

We attempt to correct for these types of effects by using a sample weighting procedure, which is a standard procedure used to correct for differences in how different demographic groups respond to the initial invitation to take the survey.

It should also be noted that a 6-7% difference is still not terribly far off from the standard accepted margin of error in the polling industry of 3% for most national surveys and professional polling firms. And I’ll also note that our survey was ultimately more accurate than the plethora of professional and partisan telephone surveys taken statewide in the lead up to the election.

This is an interesting puzzle to consider going forward: why are Democrats slightly more willing to take the Exit Poll survey than Republicans in Boyle County, Kentucky elections? I welcome ideas from interested parties.

Boyle County, KY voting patterns in the 2015 gubernatorial race

The 2015 Boyle County Exit Poll asked voters about their opinions on six issues that were discussed in the Kentucky gubernatorial campaign. Here are voting patterns for the major two-party candidates among those with each policy preference:

  Bevin Conway
Agree with Kim Davis’s decision to refuse marriage licenses (32.6% of total) 84.2% 13.2%
Agree on raising the minimum wage to $10.10/hour (64.3% of total) 32.5% 63.9%
Agree on random drug tests for recipients of public benefits (72.3% of total) 62.0% 33.5%
Agree on EPA regulation of the coal industry (56.2% of total) 25.9% 70.2%
Agree on Gov. Beshear’s Medicaid expansion decision (61.9% of total) 31.3% 65.4%
Agree on Gov. Beshear’s implementation of KYNECT insurance exchange (55.2% of total) 23.1% 73.5%

For example, among those who agree with Governor Beshear’s expansion of Medicaid, 31.3% of them voted for Matt Bevin and 65.4% of them voted for Jack Conway. It is interesting that on several of these issues, a sizable number of people who prefer Conway’s position on the issue voted for Matt Bevin (and logically vice versa).

Of course, issue preferences are not the only basis for voting decisions. Factors like partisanship and demographics also make a difference. Thus, I used a “multivariate regression” analysis to see what difference each factor made in predicting a vote for Matt Bevin, controlling for the effect of every other factor. Here are the results:

Republican partisanship 59.5%
Obama disapproval 47.9%
Kentucky’s economy is getting worse 32.7%
Disapproval of KYNECT health exchange 32.1%
Conservative ideology 31.7%
Agree with Kim Davis on marriage licenses 25.0%
Agree with random drug tests for welfare recipients 23.1%
Disagree on minimum wage increase 21.6%
Female 16.2%

(For statistics nerds, these are the minimum to max predicted probabilities of each factor in predicting a vote for Bevin in a logistic regression model. Presented coefficients are statistically significant at p<0.05.)

This is telling us that, controlling for all other factors, being a Republican was the strongest factor in predicting a vote for Matt Bevin in Boyle County: Republicans were 59.5% more likely than Democrats to do so. Disapproval of President Obama was the second-highest factor: those who disapprove were 47.9% more likely than those who approve to vote for Bevin.

It seems that the campaign issues of the KYNECT health exchange, same-sex marriage licenses, random drug tests for welfare recipients, and minimum wage increases all made a difference as well, although to a lesser extent than partisanship and opinions toward Obama.

It is also interesting to note that opinions on the EPA/coal and Medicaid did not matter when controlling for these other factors, nor did demographics like age, income, church attendance, or education. Also, Tea Party supporters were no more or less likely to vote for Bevin once other factors like partisanship and ideology were controlled for.

In sum, it seems that basic political factors like partisanship and attitudes toward President Obama were the key factors in explaining gubernatorial voting patterns in Boyle County. This suggests that state-level elections in Kentucky are following wider trends in becoming more nationalized. Assuming these results are generalizeable to the state as a whole, it seems that Kentucky voters are linking their voting preferences at the state and national level to a stronger degree than once was the case.

The Drew Curtis effect in Boyle County voting patterns

According to the 2015 Boyle County Exit Poll, 48.6% of those who voted for Drew Curtis for governor in Boyle County would have picked Matt Bevin as a second choice, and 51.4% would have picked Jack Conway as a second choice.

Also, Curtis voters were a more moderate group when it came to specific issue preferences:

  • 78.7% of Curtis voters disagreed with Kim Davis’s refusal to issue marriage licenses, compared to 90.7% of Conway voters and 41.4% of Bevin voters.
  • 56.3% of Curtis voters agree that Kentucky’s minimum wage should be increased to $10.10/hour, compared to 90% of Conway voters and 44.2% of Bevin voters.
  • 55.4% of Curtis voters tend to agree with the “EPA’s regulation of the coal industry” compared to 86.8% of Conway voters and 31.8% of Bevin voters.
  • 52.2% of Curtis voters agreed with Governor Beshear’s expansion of Medicaid under the ACA, compared to 90.7% of Conway voters and 42.4% of Bevin voters.
  • 48.9% of Curtis voters agreed with Governor Beshear’s implementation of the KYNECT insurance exchange, compared to 93.8% of Conway voters and 28.9% of Bevin voters.

Drew Curtis voters were also slightly less likely to have incomes over $50K/year (52.6% compared to 66.2% of Bevin voters and 69.3% of Conway voters).

Otherwise there were few discernible political or demographic correlates of voting for Drew Curtis instead of one of the major two-party candidates.

This suggests that (in Boyle County, at least) Drew Curtis voters were not predominantly Republicans or Democrats, further suggesting that his candidacy did not ultimately help or hurt either Conway or Bevin’s chances of winning. The lack of any clear demographic or political patterns for his supporters also suggests that he pulled in a diverse group of supporters – a true “independent” coalition.

Boyle County approval ratings by partisanship

The following are the % approval for each of the local and federal political leaders and other community organizations featured on the 2015 Boyle County Exit Poll, among Democrats and Independent-lean-Democrats, pure Independents, and Republicans and Independent-lean-Republicans:

Democrats and leaners (46% of sample) Pure Independents (6% of sample) Republicans and leaners (44% of sample)
President Barack Obama 80.1% 20.9% 4.1%
Senator Rand Paul 14.1% 46.8% 70.5%
Senator Mitch McConnell 14.7% 32.3% 51.0%
Congressman Brett Guthrie 39.7% 53.6% 64.6%
Governor Steve Beshear 88.0% 54.5% 30.7%
County Judge-Executive Harold McKinney 88.0% 56.6% 70.4%
Danville Mayor Mike Perros 79.0% 56.1% 72.9%
Danville City Commission 84.0% 56.9% 64.2%
City Manager Ron Scott 79.5% 52.8% 64.6%
Danville/Boyle EDP organization 72.5% 50.9% 55.5%