A good deal has been written in the last two weeks about the results of the special election for Congressional District 13. Democrat Alex Sink, Florida’s CFO from 2006 to 2010, lost to Republican lobbyist David Jolly in an expensive special election for the seat of Republican Congressman Bill Young; who passed away a few months ago. The district had voted for Barack Obama in 2008 and 2012, and for Alex Sink during her 2010 gubernatorial campaign. The district is 35% Democratic and 37% Republican. The seat was viewed as prime pickup opportunity for Democrats while Republicans were desperate to hold it. David Jolly was Young’s preferred successor. However, the aid-turned-lobbyist had to go through a Republican primary first; where he won with under 50% in a three way race. Sink, meanwhile, faced no primary challenge. While Sink out-raised Jolly overall, third party groups evened-up the money battle. The district was flooded with calls, canvassing, radio and TV ads by the political parties, the candidates, and third party groups. The race was always going to be close, with polling indicate a 1 to 2 point race. Polls showing 5 point gaps or more were generally regarded as outliers by many commentators. Heading into election day, it was my personal view that Sink had an advantage based on the rate of absentee ballot returns. However, when the results came in, Sink lost the race by 2%. Despite winning absentee and early votes, Sink lost the election day ballots and thus lost the election.
The map below shows the raw votes cast in each precinct for the election. In addition it shows how Sink and Jolly did compared to Obama and Romney two years earlier. While Sink won some areas that President Obama lost in the Clearwater area in the north, she lost several regions, especially in Pinellas Park, that the President won.
Once the results were final, the internet and media was flooded with explanations and hypotheses on what happened. Many pieces of analysis and criticism were not based on any abject facts, but rather on perceived notions. Claims of Sink not being liberal enough, to claims the healthcare debate hurt her (she was bombarded with the issue), to gaffs by candidates, or the state of the campaign, were thrown out there with little or no facts, figures, or backup. Meanwhile, I dug through the data available and waited for Pinellas to release its final turnout figures by party. It took 2 weeks but the data finally became available. With the data available to the public, we can now figure out what really happened.
Absentee ballots were the largest source of votes cast in the special election. Early voting provided a handful of votes, but the real fight was over absentee. Heading into election day, Democrats had a reason to be cautiously optimistic. Absentee ballot returns rates were narrowly favoring the Republicans, but by a margin less than their 2012 figures. By election day the Republicans had a 4.7% gap with ballots returned over the Democrats. This was above the 4% bare minimum they would need to have a chance, but below the 5.76% advantage they held in 2012 when they still LOST the district to President Obama by around 1.5% of the vote. A side by side of how the ballots were cast by the parties and how they broke down for the Presidential election in CD13 is below. Click the image to see a larger version.
As the table shows, Obama narrowly lost absentee despite the gap in ballots cast against his party. In addition, a 7% Democratic disadvantage in turnout on election day still saw a near tie when it came to election day ballots for Obama or Romney. These figures reflect the fact that the larger independent population of the district tend to skew towards Democrats. Sink or Obama could afford to have their base not keep up with the GOP turnout because the independents would swing to the left. With Democrats performing better turnout-wise with absentee ballots than in 2012, this gave Sink a larger margin of error with election day voting. As long as the E-Day turnout gap wasn’t much larger than 2012, Sink had a good chance of winning.
When the polls closed on election night, the absentee early vote results were released first. Alex Sink had won both early vote and absentee ballots. She held a lead around 48% to Jolly’s 46% (a libertarian candidate took the rest). The President had lost absentee votes two years earlier. For myself, this gave me more hope for Sink winning.
However, as election day results came in, the results quickly narrowed and soon Jolly was leading. By the time all the ballots were counted, Jolly had won. The culprit was election day, Sink lost it by 12%. What happened on election day could not be verified until Pinellas released all its turnout figures nearly two weeks after the election.
Election Day – Things Fall Apart
If the ballots for early and absentee voting were the only ones deciding the election, the precinct results would have looked like this.
However, the precinct map for election day alone looked like this.
So what happened on election day? For two weeks it was impossible to say for sure. My general belief was that it was a turnout gap much larger than the 7% gap from 2012. The only other explanation would have been independents voting for Jolly. However, for Sink to have won absentee ballots, she must have won the independent vote. It was unlikely that Sink would win independents on absentee, but not election day. In addition, the polls that showed a close race repeatedly showed Sink winning voters with no party affiliation.
Around a week after the election I sat down with the data and tried to get an estimate of what turnout was like on election day. I started off with the absentee and early vote results. Pinellas already had the partisan breakdown of turnout for those two voting methods. Based on the partisan share of the vote, I calculated the percents that each candidate would need from their own party and from independents to end up with the votes they got. I calculated separately for absentee and early voting. Estimating these vote shares was also aided with cross-tabs from public polls.
In both cases I estimated that each candidate received a vast majority of their party’s vote. Polls indicated the libertarian took more from Jolly than from Sink, so I calculated that Jolly would have slightly less party unity than Sink. With absentees, Sink needed around 59% of the independent vote to get the win she did. With early votes, her margin was achieved with either 70% of independents or even stronger Democratic loyalty. Early voters generally skew much more liberal and democratic than the rest of the electorate.
I applied roughly the same percents for party loyalty and independent share to the election day ballots cast. I estimated that, assuming the two parties were largely loyal to their candidate, and independents sticking with Sink, then Republicans would have needed a 20 point advantage in terms of their share of the electorate casting ballots. I estimated that the partisan breakdown of those casting ballots on election day was over 50% Republican and just over 30% Democratic. My work is below, and in this link,
I posted my estimates the weekend after the election. Otherwise I continued to wait for the final turnout data to be released. The data became available the next week.
When the data was released, it showed election day had a 17% gap in favor of the Republicans. Independents made up a larger share of the vote than I estimated. The electorate on election day was 50% Republican, 32% Democratic, and 18% Independent. Based on the turnout data I revised my calculations. Assuming the parties remained loyal, Sink likely won 59% or so of the independent vote.
Comparing the results and turnout in the special election with the ballots cast for President show where the issue lied. While Sink outperformed the President with absentees and did roughly the same with early votes, she did much worse with election day votes.
While the conservative media tried to frame the loss as a reflection on the healthcare law or the President, the data does not back that claim up. Sink won the independent vote in the district. The problem laid in an issue Democrats constantly struggle on, turnout.
Why Was the Turnout Gap so High
Before the data for election day was even released there were already groups trying to explain why turnout was so low. These groups were reacting to the overall turnout number of 39%. However, these figures did not include party turnout. Most assumed, correctly, that Democratic turnout had not kept up with Republicans. Without the data available, theories began to abound.
The most common theory was the effect of municipal elections on turnout. Several coastal communities, St Pete Beach being the most prominent, were holding principle elections the same day as the special election. In some of these municipalities the turnout was higher than the district overall. In addition, many of these cities were more Republican than Democrat. Some people argued that the higher turnout in these Republican cities, fueled by local election interest, aided Jolly by inflating Republican turnout. However, this theory does not hold up to scrutiny. It was not just a handful of beach cities hosting elections. Eleven municipalities had local races that evening.
In some of these cities turnout was much higher than the district-wide turnout. St Pete Beach, a small Republican city, won the turnout race with 54%. The turnout spike in St Pete Beach and the surrounding area fueled the narrative that municipal elections drove up the vote in key areas. However, when you combine all eleven municipalities together, you get a different story.
First and foremost, turnout was not high in every city hosting a local election. Turnout ranged from 33% to 54%. Second, several of these cities have more registered Democrats than Republicans, and even more were evenly divided. Clearwater, the largest municipality holding elections, is evenly split between the parties. Third, Democratic turnout was outpaced by Republican turnout in ten of the eleven municipalities. When voters went to the polls in those ten areas, the electorate was more Republican than the registration figures. The issue was not that Republican cities were voting, it was that Democrats were not as energized across the board. Fourth, the notion that local elections drove higher turnout does not add up. Looking back at turnout for these municipalities, I found the last time each had a local race, and turnout was much lower back then; with the exception of Kenneth City. These municipalities saw a spike in turnout, indicating that the special election drove people to the polls, not the municipal races. Fifth, and final, the combined areas hosting municipal elections were more Democratic than those not hosting down ballot races. The combined cities voted for President Obama 50.9% to 47.6%, and narrowly voted for Jolly 47.9% to 46.8%. The registration figures were nearly even, and the Democratic under-performance was smaller in the cities than it was outside the cities. This excel file shows all the cities that had local races; including their registration and ballots cast. It shows the Democratic margins (mostly negative) for registration and cast votes. Then it shows if the margin improved or got worse based on turnout. It shows the turnout for this year and for the last time each city had a municipal race. It then shows the totals for all the cities combined, the totals for the non-city areas, and the totals for the district. As the data shows, the city areas were better territory for Democrats.
So, if not municipal elections, what caused Democrats to underperform Republicans?
This is where blogs and news articles spent a good deal of time guessing. One issues brought up was candidate appeal. Some claimed Alex Sink didn’t appeal to the base, or didn’t seem charismatic enough. However, this negates the fact that Democratic turnout lagged Republican turnout in a special election for Florida House District 36 the previous fall. In that race, the Democrat, Amanda Murphy, won a close battle. The district was 37% Democratic and 34% Republican. However, of the votes cast for that race, 46% were Republican and 37% were Democratic. This included a 18% gap in favor of the GOP on election day and a 5% gap on absentee. Murphy was well liked by Democrats and proved to be a very strong candidate, but Republicans still won the turnout battle. The Murphy win very likely came from stronger cross-over support from Republicans thanks to her support from the GOP State House member who was leaving the seat, Mike Fasano. In addition, the Republican challenger, Bill Gunter, was painted as too far-right for the district. The party-level turnout for the HD36 race is below. Despite the GOP turnout advantage, Murphy won by 300+ votes.
Murphy was a strong candidate with charisma and fit the district well. However, turnout for Democrats still lagged. The cause for this was not Murphy’s appeal as a candidate. Therefore, we cannot just assume that turnout lagged in CD13 because of Sink’s appeal or lack-there-of.
Special Elections are Special
The issue of turnout is a complex issue for Democrats across the county. In both the special elections, Democrats used a well-publicized ground game to push voters to the polls. I have no doubt that no ground game would have resulted in greater turnout disparities. However, major parts of the Democratic base are made up of two key groups, young voters and minorities. Both groups are subject to vote less and special elections are especially the case. This is why Nate Silver, Dave Wasserman, and other experts, will always tell you that special elections are special. Rarely do special elections mark indicators of coming waves. Democrats lost the NYC special election a year before President Obama was re-elected, Republicans lost an upstate New York race a year before their 2010 landslide. Special elections are only indicators if they occur in ground where one party has little hope for success. The string of Democratic special election wins in conservative districts in Illinois, Louisiana, and Mississippi foretold the 2008 cycle. In addition, contrary to claims, running avid liberals do not guarantee higher democratic turnout. Many liberals have lost elections in years with depressed turnout and many conservative democrats have won elections on the coattails of Obama or on their own.
So what happened to Alex Sink? In short, she lost because of turnout on election day. Why was turnout so rough? The Democratic base is made up of those who would not be considered super voters. Questions about Sink and her ability to connect to voters, and particularity her inability to articulate her ties to the Pinellas area, may have turned off some voters. However, I believe any gaffs or bad quotes aided more in limiting cross-over support from moderate Republicans than it did in depressing Democratic turnout. Democratic turnout was reasonably high for a special election, it just was not enough. Remember, the district is largely white and older. Older white voters skew toward the GOP, and they vote. The beach communities didn’t cast ballots because they had local races, they voted because they are affluent; a group which constantly votes in higher numbers. Look at the turnout map below.
Now, look at the average income in the district
Those conservative costal areas had something more in common than hosting munical elections. Those areas are higher income. Thus, they were more likely to vote, and more likely to vote Republican at that. The demographics of the district, this just being one, made turnout an easier task for Republicans. The GOP base is older and wealthier, thus more likely to vote. The Democratic base on young, minorities, and working class, are less likely to vote. Add in that this district is largely white, older, and wealthier, and it’s no surprise the turnout gap was so high.
Claims that Obamacare doomed Sink do not hold up, claims that she wasn’t liberal enough do not hold, claims that she wasn’t appealing enough don’t hold. Special elections are tricky, you win some and lose some. Democrats won the HD36 special while being outspent by hundreds of thousands of dollars. They lost CD13 a few months later. I don’t consider this a reflection on anything other than special elections continuing to be special, subject to turnout. The electorate casting ballots does not reflect the electorate as a whole. Democrats will continue to find ways to increase turnout among their base. Republicans will continue to hope that does not happen. As the special election showed, attacks on Obamacare are not working. Sink won the independents resoundingly. If Democrats can increase Democratic turnout by a few more points, then Republicans will find themselves in real trouble. If Democrats cannot increase their base’s turnout, then they will find themselves short of a House majority anytime in the near future.