By Alex Lange, JKL Political Analyst
Over the course of our turnout series, we laid out how variations in turnout rates, and changes in the share of the electorate by different voting groups can have a real impact on the outcome of elections.
But how do campaigns influence turnout? Are they able to drive turnout up among certain groups in a targeted way? This update in our turnout series takes a brief look at what American election campaigns do to “Get Out The Vote” (GOTV).
To a large extent, turnout rates are driven simply by how much attention is being paid to the election by the general public and other factors like overall enthusiasm, the competitiveness of the election, and how important voters think the election is. Campaigns have very little control over this, which is most evident in the clear difference in turnout rates for presidential elections, compared to mid-term elections when only congressional candidates are elected.
Traditional Methods Meet Modern Analytics
Despite voter turnout being to some extent outside of the control of the campaigns, there is some wiggle room. That is why campaigns usually still invest tremendous resources into so-called GOTV efforts. The GOTV efforts are where traditional campaign methods meet modern analytics. While both sides broadly speaking use similar methods, this year, Hillary Clinton has built an extensive operation that far exceeds what Donald Trump’s campaign has built. Trump appears to be mostly relying on the Republican National Committee for his GOTV efforts.
The basic methods have not changed very much over time. In the last few months, the campaigns have been using armies of volunteers to register new voters in key states and contact voters by knocking on their doors, calling them from phone banks, or by email and social media to determine how likely they are to vote in the election, and whom they will support. What has changed is how the data are collected and then used.
The collected data are combined with other information, like voter registration records, census data, and social media data, to create analytical models of which voters are most important to target in GOTV efforts. The campaigns are looking for voters who are likely to support them, but need to be mobilized to turn out. In other words, modern analytics is not used to revolutionize campaigns from the ground up. Instead they enable campaigns to do the very basic work of turning out the vote, but better.
Campaigns also identify targets who are very likely to vote, but need to be persuaded to vote for them over the other candidate. Those voters less common and are not considered part of GOTV, but rather are treated separately as persuadable voters. As a rule, the base voters campaigns can rely on, plus the GOTV targets, plus the persuadable voters, has to add up to more than the opponent will get, in order to win.
The campaigns can determine what method of contacting the voters is likely to be the most effective for GOTV efforts. Some voters are better contacted by knocking on their door, while others may actually dislike that method of contact, so they are emailed or called instead. Similarly, campaigns determine which method the voters are most likely to use to vote and note whether they need additional support for that. They can vote by mail, vote early, or vote on Election Day.
Using research from behavioral psychology, sometimes obtained by conducting internal studies using their own data, campaigns can also maximize the effectiveness of each contact. A commonly cited example is that when reminded to vote, voters are now almost always asked whether they know where their polling station is and how to get there. This encourages voters to make a mental plan for election day, which makes them more likely to actually go and do it.
Based on polling, and analytics, the campaigns choose a GOTV strategy to pursue. For example, the Clinton campaign has invested a huge amount of resources into turning out Hispanic voters in Florida, while investing less in turning out other demographics because their analysis shows that this would have less of an impact on the outcome of the election.
Once early and mail voting starts, the GOTV efforts are deployed to encourage voters to vote. Good campaigns will use this stage to again collect data on who has already voted, whom they voted for, and how they voted, which further feeds the models, allowing them to tweak the process and deploy more resources in targeted ways.
But does it make a difference? Conventional wisdom dictates that GOTV efforts can change the outcome of election by up to about 3%. If the election is close, and one side succeeds at turning out their voters more effectively and in the places where it matters the most, it can make the difference between winning or losing that state. But typically, the other side will try to minimize that advantage by deploying their own GOTV efforts.
This year could be different. Donald Trump has very little GOTV infrastructure compared to the Hillary Clinton campaign. That means that this year we may get to learn more than in previous years what the effect of GOTV efforts can be, because we will have live experiment.