Political scientists have grown considerably more comfortable conducting scientifically rigorous experiments in recent years, and the results have proven surprisingly useful for real-life campaigns: Canvassing works better than phone banking; snitching to neighbors about people not voting in the past increases turnout; TV ads are effective, but the effects are short-lived.
Here's another one to add to the list: In some cases, campaign communications can be designed to actually reduce turnout from people you don't want voting.
Duke grad student Ying Shi conducted an experiment, later published in the journal Political Communication, around the May 2012 primary elections in North Carolina, which featured a vote on Amendment 1, a proposed state-level constitutional amendment banning same-sex marriage.
Shi sent 9,211 individuals — roughly half Democrats, half Republicans — one of two messages: a pro–marriage equality message (including arguments like, "Denying marriage to same-sex couples is a form of discrimination") or an anti–marriage equality message ("Nontraditional forms of marriage result in children growing up in an environment that often lacks a biological parent"). She then compared their turnout patterns in the primary with those of a control group.
Telling people they're wrong: pretty effective
As you might expect, Democrats getting a pro-equality message and Republicans getting an anti-equality message were likelier to vote, and Democrats getting an anti-equality message and Republicans getting a pro-equality message were less likely to vote. That's what you'd expect: When people's initial beliefs are confirmed, they're more motivated to turn out, but when they're contradicted, that reduces their confidence in their position and decreases their inclination to vote.
But the size of the effects varied, and the effects on Democrats, and the effects of an anti-equality message on Republicans, mostly weren't statistically significant. The only group for which the results were consistently significant were Republicans getting the pro-equality message.
Republicans in big households, with three or more people, were particularly prone to being demobilized by pro-equality messaging. While Republicans in large households in the control group had a turnout rate of 49 percent, getting the pro-equality message reduced that to 42.9 percent. By contrast, individuals in households of two or fewer didn't demobilize much, making the overall effect on Republicans smaller: 1.8 percentage points, rather than 6.1 points among just people in large households.
"Co-residence with two or more individuals provides opportunities for interpersonal discussions that may influence the magnitude and direction of treatment," Shi explains. "Individuals seeking to avoid interpersonal disagreement and to minimize threats to their social relationships may withdraw from political participation." Translation: Getting the message might convince Republicans to stay home for fear of generating conflict with people they live with, who disagree with them.
It's easier to get people to stay home than to turn out
Why is the effect of cross-cutting messaging bigger with Republicans than Democrats? One possibility, Shi notes, is just that the study is "underpowered," and that if it'd had a bigger sample you would've seen significant effects for both Democrats and Republicans. That story is supported by the fact that the results for Democrats become significant when you increase the sample by adding in nontargeted individuals living at the same addresses as those targeted.
Then again, maybe the text trying to persuade Democrats was just less persuasive than that trying to convince Republicans. Maybe it's just this specific issue, and Republicans living with pro–marriage equality people in their household felt pressure to stay home for which there was no equivalent among Democrats.
But the one thing that Shi finds consistently is that demobilization effects are stronger than mobilization effects: Telling people they're wrong is more effective at getting people to stay home than telling people they're right is at getting them to vote.
This is just one study; experiments in this area are still pretty limited in number, and the existing literature on the effects of exposure to disagreement is pretty mixed. In factual matters, there's some evidence that debunking people — for example, by telling Republicans that Iraq had no weapons of mass destruction, or Democrats that the Bush administration did not ban all stem cell research — actually backfires and leads them to hold their current wrong beliefs more stridently. The messaging Shi sent out also identified itself as for research, which might affect results in a way that could limit the tactic's usefulness to actual campaigns.
That said, if the effect she's identifying is real and broadly applicable, you could imagine it proving rather useful, especially in issue campaigns. Suppose California were considering a ballot initiative to abolish the death penalty. Supporters of the initiative could identify voters they think are likely to oppose it and want to keep the death penalty, and then send them postcards outlining the basic case for abolishing it. If Shi's findings hold up, that should depress turnout among death penalty supporters and increase the odds of abolition.