Explore Harvard's Nieman network Nieman Fellowships Nieman Lab Nieman Reports Nieman Storyboard

What really gets citizens to turn out to vote in elections?

ASK THIS | November 06, 2006

When it comes to getting out the vote, personal tactics work much better than automated ones – while microtargeting may not be all it’s cracked up to be. Yale Professor Donald Green thinks the two parties are evenly matched this year when it comes to outreach.


By Donald Green
donald.green@yale.edu

Q. We hear a lot about voter mobilization during the final stretch of every campaign.  How much difference do GOTV (Get Out the Vote) campaigns actually make in affecting election outcomes?

The impact of a GOTV campaign is a function of the scale and quality of its outreach effort.  Randomized experiments on the effectiveness of GOTV tactics such as door-to-door canvassing, phone calls, mail, and email suggest that the more personal a tactic is, the more effective it is. Face-to-face contact seems to be the most effective tactic, raising turnout by 7-12 percentage-points among those who are contacted.  E-mail and robotic calls, on the other hand, are impersonal tactics that seem to have no detectable effect on turnout, even among those who listen to the call or open email from a campaign.

Very often, campaigns confront a trade-off between quantity and quality. Robotic calls, for example, are inexpensive and therefore give a campaign the opportunity to reach very large numbers of voters.  The catch is that robotic calls, mass emails, and other impersonal techniques typically have negligible effects on voter turnout.  When personal contact is possible -- in areas where population density is high and where dwellings are accessible to canvassers -- it appears to be the most effective and cost-effective means of turnout for voters.  When geography and logistics make canvassing impossible, phone calls from volunteers appear to be the next-best option, increasing turnout by 3-5 percentage-points among those contacted by the campaign.  Commercial phone banks appear to be less effective than volunteer efforts, perhaps because paid callers are less committed and persuasive, and their calls typically strike voters as more mechanical.

Thus, a GOTV campaign that can successfully reach large numbers of supporters using personal appeals has the potential to swing an election in which the candidates are separated by a 2 or 3 point margin.

Q. Campaigns' use of "microtargeting" has received increasing press attention in recent weeks.  What is microtargeting?  How valuable is this campaign tactic?

Microtargeting refers to a campaign tactic that involves identifying potential supporters and tailoring messages and outreach strategies to them.  Whereas "macrotargeting" involves broad efforts to persuade and mobilize social groups or geographic areas, microtargeting represents an attempt to craft campaign tactics that will persuade and mobilize specific individuals or households.

In practice, microtargeting starts with a voter file, which contains background information such as a voter's location, age, gender, and in some cases party registration, ethnicity, and voter turnout patterns in past elections.  These data are then augmented in one of two ways.  First, they may be combined with survey data; certain voters may be asked about their vote preferences in the coming election.  Using these data, a statistical model may be developed to predict the vote preferences of the respondents based on the attributes in the voter file. This model is then used to forecast the vote preferences or political opinions of everyone in the voter file.  Another method of augmenting the voter file is to add consumer purchase data, such as magazine subscriptions, car ownership, and the like.  These variables, too, may be used to forecast candidate preference.

Once campaigns have a sense of who their supporters are and what issues swing voters care about, their outreach efforts are crafted to mobilize supporters and persuade swing voters.  Microtargeting is said to enhance the persuasiveness of mail, phone calls, and canvassing visits because the campaign can appeal to voters in terms of the issues they care about.

That's the theory, anyway.  The problem with microtargeting is that it is not cheap. Campaigns pay top dollar for these augmented voter files.  So the question is whether the lists actually pay for themselves by increasing the efficiency with which the campaign generates votes.  That remains an untested conjecture.  There has yet to be an experiment that compares the effectiveness of two outreach campaigns: one based on macrotargeting and the other on microtargeting.  Until the data come in, a bit of skepticism may be in order. Although journalistic accounts of microtargeting have captured the public's imagination with amusing stories about how drivers of certain car models have distinctive political views, the fact is that consumer preferences do not predict a large degree of variance in voting preference beyond what is predicted by party registration, voting in party primaries, and other more directly political predictors. Maybe microtargeting will one day be proven to be valuable, and skeptics will be forced to eat their words.  In the meantime, there is no reliable scientific evidence -- that is, evidence based on randomized experiments -- showing that microtargeting is worth the cost.

Q. What is the difference in the likely success of the Democratic vs Republican GOTV efforts?

The 2004 election will be remembered as an election in which both parties emphasized voter mobilization rather than voter persuasion.  The campaigns sensed early on that that there were relatively few swing voters; they focused their energies on energizing their respective activist bases, who in turn supplied armies of canvassers and volunteer callers.

The setting of the 2006 election is quite different.  In many states, such as my home state of Connecticut, moderates are squaring off against one another, and the campaigns focus as much on persuasion as mobilization.  The volume of mobilization activity is much smaller than in 2004, reflecting the smaller scale of midterm campaign spending.

The question is whether, in a year that seems to favor Democrats, the GOP can muster the devoted army of conservative canvassers that it deployed so effectively in 2004.

In states where Republicans are attempting to distance themselves from the Bush agenda, local GOP activists may become dispirited.  Moreover, it may be difficult to entice GOP activists from other noncompetitive districts to perform GOTV work in competitive regions -- an innovation that seemed to be successful in the 2002 Republican campaigns.

My sense is that both parties are evenly matched this time around in terms of the quantity and quality of their GOTV efforts but that, overall, both sides have focused a larger percentage of their resources on voter persuasion (e.g., TV ads) than mobilization.  For that reason, I expect a smaller surge in turnout from 2002 to 2006 than we witnessed from 2000 to 2004.

Q. Why do voter turnout rates in the United States tend to lag behind turnout rates in other countries?  Why are the rates in the U.S. so much lower now than they were in the late 19th century?

One of the more interesting puzzles in social science is the question of why voter turnout in the U.S. nowadays lags behind other countries and behind America's own historical turnout rates.  In recent years, the puzzle has acquired a new twist: voting rates in the U.S. increasingly resemble those in European countries, as the latter experience a gradual decline in turnout.  This twist diminishes the force of various institutional explanations, such as the fact that the U.S. holds frequent elections, lacks a proportional representation system, and requires voters in most states to register in advance of Election Day.  Deepening the puzzle is the fact that turnout is lower now than in the 19th century despite the fact that voters nowadays are, on average, much better educated and much older than their 19th century counterparts.

Social scientists are far from answering this puzzle, but recent experiments suggest some interesting hypotheses.  Here is a bit of background.  Voting in the 19th century was an all-day affair.  People would mill about for hours, socializing with friends, imbibing free booze supplied by the political parties, and watching their neighbors cast what was until the 1880s a public vote.  The advent of secret balloting did not bring about an immediate drop in turnout.  In fact, the effects of the secret ballot were initially fairly modest.  But the same social movement that cleaned up elections by instituting the secret ballot also instituted the rule that said that party workers had to remain a good distance from the place where balloting occurred.  That innovation seems to have undermined parties' incentives to supply booze and food; voting gradually became a sober affair in which voters cast ballots quickly and quietly.  As the fun went out of voting, turnout rates gradually declined.  In the North, for example, turnout rates in the 1880s were roughly 30 percentage-points higher than in the 1920s.  One might suppose this decline was due to the end of petty bribery, but in fact the decline is evident not only in cities (where machine politics was common) but also in outlying and rural areas as well.

In an effort to understand the role that the social environment may have played in the 1880s, recent experiments have investigated the extent to which a festive, carnival-like atmosphere increases voter turnout. Randomized experiments have gauged whether Election Day festivals increase turnout.  Potential festival sites are identified, some are randomly assigned to the treatment group, parties are thrown, free food and (non-alcoholic) drinks are served, and votes tallied.  Approximately two dozen such sites have been studied, and the results suggest that creating a festive atmosphere generates a statistically significant increase in turnout.  Of course, 21st century voters have no idea what to make of announcements of an Election Day festival, so time will tell how this innovation alters turnout patterns over the long term.  Nevertheless, the results are an exciting potential development in the understanding of what makes -- and made -- people vote.



The NiemanWatchdog.org website is no longer being updated. Watchdog stories have a new home in Nieman Reports.