2012 Boyle County Exit Poll data file available for download

For interested parties, the data file of the 2012 Boyle County Exit Poll is now publicly available. It can be downloaded here:

http://web.centre.edu/benjamin.knoll/2012bcep_public.zip

Happy number crunching! 

Cross-tabulation percentages in 2012 Danville City Commission voting

I previously posted the correlational relationships between voting for pairs of Danville City Commission members. Here is the same information, but presenting percentages instead of correlation figures. This should hopefully be a little easier to intuitively interpret.

The graph should be interpreted like this: Find the candidate in the rows going down, then find another candidate in the columns going across. The percentage figure where those two candidate intersect can be interpreted: “x% of Danville voters who choose the candidate from the row also voted for the candidate from the candidate in the corresponding column.” For example, the row Smiley corresponds column Montgomery at 28%, which means that 28% of those who voted for Smiley also voted for Montgomery (but not vice-versa, as 44% of those who voted for Montgomery also voted for Smiley).

	Smiley	Montgomery	Louis	Atkins	Caudill	Stevens	Hamner
Smiley	100%	28%	37%	58%	50%	66%	30%
Montgomery	44%	100%	52%	39%	40%	46%	37%
Louis	49%	45%	100%	39%	34%	57%	38%
Atkins	39%	17%	20%	100%	65%	74%	52%
Caudill	38%	19%	19%	73%	100%	76%	54%
Stevens	42%	19%	27%	69%	64%	100%	55%
Hamner	26%	27%	24%	66%	60%	73%	100%

We see the strongest relationship between Hamner and Stevens, as 73% who voted for Hamner also supported Stevens. The weakest relationship is between Atkins and Montgomery, as only 17% of Atkins voters also selected Montgomery.

Exit poll results on Kentucky’s 54th District House Race: Mike Harmon (R) vs. Barry Harmon (D)

This week, Boyle County’s incumbent state legislator Mike Harmon (R) won a tight reelection race against challenger Barry Harmon (D), 54.8% to 45.2%. The 54th district includes Boyle and Washington counties. Mike Harmon (R) won Boyle County with 51.5% of the vote and Washington County with 62.8% of the vote. Boyle County accounted for 70.5% of the total vote for the 54th district seat.

Below are the exit poll results for Boyle County’s share of the 54th district election, broken down by relevant demographic and political groups. The cross-tabulations read across, meaning for example that the first line reports that among those who self-identified as “liberals,” 17% voted for Mike Harmon while 83% voted for Barry Harmon.

	Mike Harmon (R)	Barry Harmon (D)
Liberal	17%	83%
Moderate	41%	59%
Conservative	76%	24%
Democrat	14%	86%
Ind-lean-Democrat	20%	80%
Independent	48%	52%
Ind-lean-Republican	74%	26%
Republican	85%	15%
Below $20K/year	41%	59%
$20-$50K/year	50%	50%
Over $50K/year	49%	51%
Church never	29%	71%
Church sometimes/once per week	46%	54%
More than once per week	55%	45%
High School	51%	49%
College graduate	50%	50%
Postgraduate	39%	61%
White	51%	49%
African-American	15%	85%
Female	46%	54%
Male	51%	49%
18-29	42%	58%
29-44	50%	50%
45-64	50%	50%
65+	51%	49%
Pres: voted for Obama	12%	88%
Pres: voted for Romney	76%	24%
Congress: voted for Williams (D)	13%	87%
Congress: voted for Guthrie (R)	74%	26%
Approve Boyle Judge McKinney (D)	44%	56%
Disapprove McKinney (D)	58%	42%

Ideological and partisan patterns in Danville’s city commission election

More from the 2012 Boyle County Exit Poll:

Among Danville voters:

% calling themselves “conservative” – % calling themselves “liberal”:

13%	Smiley
13%	Montgomery
19%	Louis
15%	Hunstad approval
13%	City Commission approval

In other words, you could say that Paul Smiley had a 13% advantage among self-reported conservatives as compared to self-reported liberals. Use a similar interpretation for all of the following graphs: 

% calling themselves “liberal” – % calling themselves “conservative”:

20%	Hamner
20%	Atkins
14%	Caudill
10%	Stevens

 % calling themselves Republican (and independent-lean-Republican) – % calling themselves Democrats (and independent-lean-Democrat):

18%	Smiley
16%	Montgomery
9%	Louis
9%	Hunstad approval

 % calling themselves Democrat (and independent-lean-Democrat) – % calling themselves Republican (and independent-lean-Republican):

13%	Hamner
17%	Atkins
3%	Caudill
7%	Stevens

It looks like there was a clear partisan/ideological difference in voting patterns for the various city commission candidates and in approval/disapproval of Mayor Hunstad. Smiley, Montgomery, and Louis had anywhere from a 9%-19% advantage among conservatives and Republicans. Atkins, Caudill, Stevens, and Hamner had between a 3%-20% advantage among liberals and Democrats (especially Hamner and Atkins), although Stevens and Caudill seemed to have a smaller ideological/partisan advantage amongst liberals and Democrats.

How might Paul Smiley vote on the new Danville City Commission?

The results from the 2012 Boyle County Exit Poll allow us to examine whether there is any evidence of how Danville’s new city commissioners are likely to vote over the next two years. Below is a table of correlations between self-reported voting for each of the seven commission candidates among Danville voters, as well as self-reported approval for Mayor Hunstad:

	Hunstad approval	Smiley	Montgomery	Louis	Hamner	Atkins	Caudill	Stevens
Hunstad approval	1	0.059	.315(*)	.384(*)	-.182(*)	-.265(*)	-.254(*)	-.214(*)
Smiley	0.059	1	0.032	.106(*)	-.305(*)	-0.062	-0.068	0.01
Montgomery	.315(*)	0.032	1	.275(*)	-.141(*)	-.268(*)	-.172(*)	-.233(*)
Louis	.384(*)	.106(*)	.275(*)	1	-.138(*)	-.304(*)	-.270(*)	-.118(*)
Hamner	-.182(*)	-.305(*)	-.141(*)	-.138(*)	1	.096(*)	.120(*)	.159(*)
Atkins	-.265(*)	-0.062	-.268(*)	-.304(*)	.096(*)	1	.267(*)	.225(*)
Caudill	-.254(*)	-0.068	-.172(*)	-.270(*)	.120(*)	.267(*)	1	.253(*)
Stevens	-.214(*)	0.01	-.233(*)	-.118(*)	.159(*)	.225(*)	.253(*)	1

In plain English, positive numbers mean that, in general, the voters selected both of the two candidates on the same ballot, or voted for the candidate and indicated approval for Mayor Hunstad on the survey. Negative numbers indicate that voters did not tend to vote for those two candidates on the same ballot. The bigger the number, the stronger the relationship in either direction. Also, the asterisk (*) indicates that there is strong evidence that a relationship exists and is not just due to random sampling error (for statistics nerds: it’s statistically significant at p<0.05).

From these correlations, we can discern some clear patterns. Not surprisingly, those who voted for Atkins and Caudill tended to vote strongly for Stevens and strongly against Louis and Montgomery. There is also a relationship between voting for Atkins/Caudill and Hamner, although the relationship is weaker than that for Stevens. Conversely, those who voted for Montgomery also voted strongly for Louis.

Also, as we would expect, approval of Mayor Hunstad is highly correlated with voting for Montgomery and Louis and with voting against Caudill, Atkins, and Stevens. This is not surprising given that Hunstad, Montgomery, and Louis  voted together (and against Atkins/Caudill) on some key controversial issues over the past two years.

To me, the interesting thing to note is the lack of a clear pattern for voting for Paul Smiley. Voting for Smiley was not highly related to approval for Mayor Hunstad or voting for several of the other candidates. There is a weak relationship between voting for Smiley and Louis (but not Montgomery) and between voting for Smiley but not for Hamner.

This all suggests the following narrative: most Danville voters either voted for Atkins/Caudill/Stevens or Montgomery/Louis. They then had to choose among Hamner and Smiley for the last 1 or 2 votes, respectively. It seems that the Atkins/Caudill/Stevens voters split their last vote between Hamner and Smiley, while more Montgomery/Louis voters choose Smiley over Hamner. 

Based on voter perceptions, then, as evidenced through these voting patterns, it seems that Danville voters expect that over the next two years there will be a majority voting bloc of J.H. Atkins, Kevin Caudill, and Paige Stevens, with Mayor Hunstad on the other side, and Paul Smiley a “swing” voter in the middle. Whether Smiley opts to side with the Mayor or with the majority might not ultimately matter, though, as the three-member majority of Atkins/Caudill/Stevens will be able to decide any vote as long as they vote together.

At the same time, I urge caution in making too much of this conclusion. Previous analysis indicates that even among the previous commission with a clear 3-2 split, they voted unanimously 90% of the time. I expect the same will be true over the next two years.

Voter familiarity and local political issues

One interesting finding of the 2012 Boyle County Exit Poll is the relatively limited familiarity that many voters have with local political issues compared to national political issues. Compare the percentage of Boyle County/Danville voters who declined to answer the following questions on the survey, indicating “don’t know” or “no opinion”:

NATIONAL

• President Obama: 4.6% don’t know/no opinion

STATE:

• Senator Mitch McConnell: 14.9% don’t know/no opinion
• Senator Rand Paul: 18.1% don’t know/no opinion
• Congressman Brett Guthrie: 34.5% don’t know/no opinion

LOCAL:

• Boyle Judge-Executive Harold McKinney: 27.7% don’t know/no opinion
• Mayor Bernie Hunstad: 22.3% don’t know/no opinion
• Danville City Commission: 26% don’t know/no opinion
• Water system expansion project: 24.5% don’t know/no opinion
• Hiring process/choice of Police Chief Tony Gray: 20% don’t know/no opinion
• Hiring process/choice of City Manager Ron Scott: 29.4% don’t know/no opinion
• Management of the city budget: 26.6% don’t know/no opinion
• Preparing for the VP debate: 16.7% don’t know/no opinion
• Purchase of the BISCO building: 29% don’t know/no opinion

It’s interesting to note that even among voters (who themselves tend to be more educated members of the population), somewhere around 25-30% indicate insufficient familiarity with local political issues to indicate an opinion one way or another. Compare this to only 5% who indicated “no opinion” on attitudes toward President Obama. 

Also, Brett Guthrie’s 35% “don’t know/no opinion” approval can probably best be understood as a result of Boyle County’s very recent redistrict into his 2nd congressional district. Boyle County voters are still just barely getting to know their new congressman.

Interpreting the results of the 2012 Danville City Commission election

From the Kentucky Board of Elections website, this is how the Danville City Commission election turned out on November 6, 2012:

Paul SMILEY		13.79%	2,839
Ryan A. MONTGOMERY		8.95%	1,842
Norma Gail LOUIS		10.86%	2,235
Janet HAMNER		13.13%	2,703
James ‘J.H.’ ATKINS		18.16%	3,738
Kevin CAUDILL		15.92%	3,278
Paige STEVENS		19.20%	3,954
			20,589

Incumbents Kevin Caudill and J.H. Atkins were reelected while incumbents Ryan Montgomery and Norma Gail Louis were not. Caudill and Atkins will be joined by Paige Stevens and Paul Smiley (and Mayor Hunstad) in next year’s city commission. 

First observation: this is in line with the results of our 2011 Exit Poll, which showed Caudill and Atkins with net positive approval ratings from Danville voters, while Montgomery and Louis had net negative approval ratings.

Second observation: our 2012 Exit Poll projections from earlier this afternoon turned out to be pretty darn accurate, correctly predicting all four winners even when only a third of the surveys were compiled.

We can look at these statistics to see the “what” of this election, while exit polls can help us understand the “why.” So what can we learn about why the Danville City Commission election turned out the way it did?

One popular explanation is that an election result where incumbents are involved can be interpreted as a “referendum.” In other words, incumbents are reelected when voters are pleased with their performance and not reelected when voters are displeased. We can assess this possibility by looking at the  issues surveyed in our exit poll. We asked respondents a general approve/disapprove of the Commission’s handling of six specific issues:

• Water system expansion project: 54.7% approve, 20.8% disapprove, 24.5% don’t know/no opinion
• Hiring process/choice of Police Chief Tony Gray: 70.3% approve, 9.6% disapprove, 20% don’t know/no opinion
• Hiring process/choice of City Manager Ron Scott: 32.6% approve, 38.1% disapprove, 29.4% don’t know/no opinion
• Management of the city budget: 29.7% approve, 43.7% disapprove, 26.6% don’t know/no opinion
• Preparing for the VP debate: 77% approve, 6.4% disapprove, 16.7% don’t know/no opinion
• Purchase of the BISCO building: 27.1% approve, 43.9% disapprove, 29% don’t know/no opinion

Did opinions toward these issues affect overall opinions toward the Commission as a whole, and thus support a referendum election? Yes. And this would matter because it’s possible that disapproval of certain actions that the Commission majority has taken could be associated with a decreased likelihood of voting for the incumbent members of the majority (i.e. Montgomery and Louis). On the whole, only 30.9% of Danville respondents indicated approval of the City Commission while 43.1% indicated disapproval. So which issues are associated with commission disapproval?

Among those who disapprove of the Danville City Commission’s job performance:

• 64.3% approve of the work on the water system expansion
• 86.3% approve of the appointment of Tony Gray as police chief
• 90.6% approve of the city’s preparations for the VP debate
• 24% approve of the appointment of Ron Scott as city manager
• 17.8% approve of the management of the city budget
• 17.4% approve of the city’s handling of the BISCO building purchase

Thus, it seems that Danville voters judged the City Commission’s performance primarily on how it has handled the controversial issues (the city manager replacement and the BISCO building purchase). Since both Montgomery and Louis supported Mayor Hunstad on the city manager issue, and Montgomery has been associated strongly with the BISCO building purchase controversy in the week leading up to the election, it strongly suggests that the results of the election can best be interpreted as a referendum on these controversial issues. 

We can further examine this explanation by comparing views on these controversial issues to voting for the various city commission members:

Among those who disapproved of the appointment of Ron Scott as city manager:

• 19.1% voted for Montgomery and 19.3% voted for Louis
• 72.2% voted for Atkins and 69.2% voted for Caudill
• 76.2% voted for Stevens, 54.7% voted for Hamner, and 37.7% voted for Smiley

Among those who disapproved of the handling of the BISCO building purchase:

• 18.5% voted for Montgomery and 25.5% voted for Louis
• 69.6% voted for Atkins and 64.8% voted for Caudill
• 75.3% voted for Stevens, 54.4% voted for Hamner, and 42.3% voted for Smiley

There is a very similar pattern for disapproval of handling of the city budget.

On the flip side, opinions toward the city’s handling of the water plant expansion as well as the appointment of Tony Gray as police chief have no strong relationship with voter choice for city commission members. (For example, among those who approve of the water plant expansion, 28.1% voted for Montgomery, and among those who disapprove of the water plant expansion, 28.0% voted for Montgomery. A similar pattern holds for the other Commission candidates.)

In sum, the poll results strongly suggest that the results of the Danville City Commission election can best be interpreted as a referendum on the controversial decisions that the majority of the commissioners made throughout 2011 and 2012 (specifically the on the city manager issue and the BISCO building purchase), despite the overall general public approval of more popular decisions such as the appointment of Tony Gray as police chief and the handling of the water plant expansion.

2012 Boyle County Exit Poll: response rate and actual vs. sample comparisons

RESPONSE RATE:

We had a total of 1,461 individuals who agreed to take our exit poll survey. We asked 1,435 individuals to take the survey but declined. Thus, we asked a total of 2,896 people to take the survey, 1,461 who agreed, leaving us with a response rate of 50.4%, almost exactly the same as the response rate in 2011.

ACTUAL VS. SAMPLE COMPARISONS

	Actual Results	Exit Poll Results	Difference	Weighted exit poll results	Difference
Mitt Romney	62.3%	54.2%	-8.1%	57.2%	-5.1%
Barack Obama	36.2%	41.5%	5.3%	39.1%	2.9%
Brett Guthrie	57.7%	50.2%	-7.5%	53.2%	-4.5%
David Williams	37.6%	32.3%	-5.3%	30.9%	-6.7%
Mike Harmon	51.5%	45.6%	-5.9%	49.0%	-2.5%
Barry Harmon	48.5%	49.3%	0.8%	47.9%	-0.6%
Const. Amendment YES	83.4%	80.6%	-2.8%	73.5%	-9.9%
Const. Amendment NO	16.6%	19.4%	2.8%	17.2%	0.6%
	AVERAGE		-2.6%		-3.2%

The sample this year was off at a slightly higher rate than it was in 2011. Looking at the refusal rates, this is likely due to the disproportionate amount of 65+ age voters who declined to take the survey, compared to 2011. (Fewer senior citizens led to a slightly more Democratic-leaning response set in the partisan questions.)

Overall, though, the average difference between the actual and sample figures is 2.6% (unweighted) or 3.2% (weighted for gender and age), an impressive amount either way.

2012 Boyle County Exit Poll – Approve/disapprove and local community priorities

Here are the preliminary results from the 2012 Boyle County Exit Poll, administered by Centre College. Upcoming posts will include a look at the degree of accuracy of this exit poll as well as analyzing the outcomes of the local races. This exit poll was administered to Boyle County voters on November 6, 2012. Respondents were randomly selected by interviewers to participate in the survey. Centre College students enrolled in Professor Benjamin Knoll’s GOV 110 (Introduction to Politics), GOV 210 (Introduction to American Politics), and GOV 330 (Parties, Campaigns, and Elections) courses participated in designing and administering the exit poll. Students were on-site from 6:00 AM through 6:00 PM surveying voters as they left the polling locations on Tuesday, November 6^th, 2012. In all, 1,461 Boyle County voters (including 1,038 self-reported from Danville) participated in the exit poll. The following statistics have a margin of error of ±2.4% for questions given to all Boyle County voters.

ALL BOYLE COUNTY VOTERS

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

• 30% – Jobs
• 12.8% – Education
• 12.6% – Economic development
• 10.3% – Prescription drug abuse

 Do you approve of the way that the following political leaders are handling their job?

• President Obama: 42% approve, 53.4% disapprove, 4.6% don’t know/no opinion
• Senator Mitch McConnell: 33% approve, 52.2% disapprove, 14.9% don’t know/no opinion
• Senator Rand Paul: 41.5% approve, 40.4% disapprove, 18.1% don’t know/no opinion
• Congressman Brett Guthrie: 44.1% approve, 21.4% disapprove, 34.5% don’t know/no opinion
• Boyle Judge-Executive Harold McKinney: 58.4% approve, 13.9% disapprove, 27.7% don’t know/no opinion

DANVILLE VOTERS ONLY

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

• 63.5% right direction, 26.3% on the wrong track, 10.2% don’t know/no opinion

Do you approve of the way that the following political leaders are handling their job?

• Mayor Bernie Hunstad: 32.2% approve, 45.6% disapprove, 22.3% don’t know/no opinion
• Danville City Commission: 30.9% approve, 43.1% disapprove, 26% don’t know/no opinion [note: corrected from previous post]

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

• Water system expansion project: 54.7% approve, 20.8% disapprove, 24.5% don’t know/no opinion
• Hiring process/choice of Police Chief Tony Gray: 70.3% approve, 9.6% disapprove, 20% don’t know/no opinion
• Hiring process/choice of City Manager Ron Scott: 32.6% approve, 38.1% disapprove, 29.4% don’t know/no opinion
• Management of the city budget: 29.7% approve, 43.7% disapprove, 26.6% don’t know/no opinion
• Preparing for the VP debate: 77% approve, 6.4% disapprove, 16.7% don’t know/no opinion
• Purchase of the BISCO building: 27.1% approve, 43.9% disapprove, 29% don’t know/no opinion

OTHER DEMOGRAPHIC/PARTISAN CHARACTERISTICS OF ALL BOYLE COUNTY RESPONDENTS

• 24% liberal, 33% moderate, 33% conservative
• 43.5% Democrat or Independent-lean-Democrat, 6.6% Independent, 41.6% Republican or Independent-lean-Republican
• 55% female, 45% male
• 89.7% white, 6.4% African-American, 2.3% “other” race
• 21% age 18-29; 26% age 30-44; 40% age 45-64; 13% age 65+
• 3% Centre College student

Preliminary (but very incomplete) figures: 2012 Boyle County Exit Poll

At the time of this writing, it is 4:30 PM on Election Day. We have about 500 surveys compiled, which I estimate to be somewhere between 30-40% of the total amount of surveys that will be completed today by Danville and Boyle County voters. Because this is INCOMPLETE, these are unofficial and non-binding figures. I am simply reporting what we have compiled so far. The final result will likely change. But this might give an indication of how the races are headed. AGAIN: THESE FIGURES ARE UNOFFICIAL, NOT FINAL, AND INCOMPLETE. Someone might be shown to be in the lead here but ultimately lose the election (or be trailing but win the election). But this is what we have so far:

54TH DISTRICT HOUSE RACE (margin of error about +/- 4%):

Barry Harmon (D) 51.7%, Mike Harmon (R) 48.3%

DANVILLE CITY COMMISSION – % of voters who chose candidate as one of the four (margin of error about +/- 4%):

• Paul Smiley: 51.6%
• Ryan Montgomery: 25.9%
• Norma Gail Louis: 27.2%
• Janet Hamner: 50.8%
• J.H. Atkins: 68.7%
• Kevin Caudill: 56.1%
• Paige Stevens: 64.2%

FULL AND COMPLETE RESULTS TO BE POSTED THIS EVENING.