Swing vote ‘trumped’ turnout in 2016 election

A new study concludes that people switching their votes from Democrat to Republican better explains the GOP’s success in 2016 than did increased turnout.
A map of the 2016 presidential election results by county

A map of the 2016 presidential election results by county. (Ali Zifan via Wikimedia Commons)

Swing voters in battleground states delivered Donald Trump his unexpected victory in the 2016 presidential election, suggests a new study coauthored by Yale political scientist Gregory A. Huber.

The study, published on April 21 in the journal Science Advances, compares the outcomes of the 2012 and 2016 presidential elections in six key states: Florida, Georgia, Michigan, Nevada, Ohio, and Pennsylvania. The analysis merged voter turnout records of 37 million individuals with precinct-level election returns to determine the sources of Trump’s electoral success. It examined the relative roles of conversion — voters switching their support from one party to the other between elections — and changes in the electorate’s composition, which are driven by mobilization and variations in voter turnout.

The researchers found that conversion was the greater factor in four of the six states, including Florida and the pivotal Rust Belt states of Ohio, Michigan, and Pennsylvania. Overall, people switching their votes from Democrat to Republican more consistently explained the GOP’s success in 2016 than did increased turnout by the party’s base, they concluded.

Despite increasing political polarization, a lot of voters aren’t committed partisans and will cast ballots for a Democrat in one election and a Republican in the next,” said Huber, the Forst Family Professor of Political Science in Yale’s Faculty of Arts and Sciences. “Turnout certainly matters — the parties benefit from mobilizing their bases — but our study suggests that swing voters were a bigger factor in 2016.”

Studying the sources of electoral change is challenging. The secret ballot prevents researchers from observing individuals’ vote choices. At the same time, the composition of the electorate constantly changes as people move, become eligible to vote, or fall off the voter rolls. The absence of a centralized election administration in the United States presents another obstacle.

Survey data can offer some insight into voters’ choices, but its reach is limited, Huber said.

It’s fairly easy to get committed Republicans or Democrats to tell you that they support their teams, but it’s much harder to reach the people who aren’t partisan or don’t vote consistently,” he said. “Those kinds of voters are important drivers of electoral change.”

Huber and coauthors Seth J. Hill of the University of California-San Diego, and Daniel J. Hopkins of the University of Pennsylvania relied on public records to avoid the recruitment bias and other shortcomings of survey data. They gathered comprehensive lists of eligible voters from each state, allowing them to identify changes in voter turnout between 2012 and 2016. They matched the voter lists to election returns at the precinct level — the smallest geographical unit for measuring vote counts — to estimate the extent of conversion that occurred between 2012 and 2016 in each precinct.

Trump improved on Mitt Romney’s 2012 performance in each state but Georgia. The researchers found that the balance between conversion and the electorate’s composition varied by state, but their analysis clearly indicated that conversion more consistently explained the pro-GOP electoral change between the two elections. Trump outperformed Romney in precincts where the electorate’s composition, or turnout, remained stable between 2012 and 2016 as well as in precincts where shifts in party registrations had favored the Democrats, according to the study.

The researchers found that conversion was especially relevant in Michigan, Ohio, and Pennsylvania — the states with the largest swings in party margin between the two elections. For example, in the average Michigan precinct, Trump netted 101 votes over Romney’s 2012 total. Changes in electorate composition increased the Democratic vote total by an estimated 102 votes. To net those 101 votes, Trump gained an estimated 203 votes from voters who had cast ballots for Barack Obama in 2012, the study found. In all, the composition effect in Michigan was estimated to be only half the conversion effect.

In a sense, the difference between composition and conversion comes down to simple math,” said Huber. “Mobilizing one additional voter adds a single vote to your margin but converting a swing voter adds one to your candidate’s tally while subtracting another from your opponent’s,” Huber noted.

In Nevada and Georgia, the estimated compositional effects were 3 and 1.4 times larger than the conversion effects. The Democrats’ enhanced voter mobilization efforts in Georgia, which are credited with enabling Joe Biden’s 2020 victory in the state, were already producing results in 2016, Huber explained.

Georgia demonstrates the importance of voter mobilization,” Huber said. “The Democrats had a massive expansion in registration and an enormous number of new voters entering the political system, which resulted in Trump losing votes relative to Mitt Romney. It set the stage for a surprising win in 2020.”


Share this with Facebook Share this with X Share this with LinkedIn Share this with Email Print this

Media Contact

Bess Connolly : elizabeth.connolly@yale.edu,