Here’s a question: If you were in charge of designing an advertising campaign for a conservative candidate in a swing state, what would your message be? And who would be your target demographic? Could you be sure your ad campaign would move the most persuadable voters and be the most effective use of your finite resources?
If your responses to these three questions fell somewhere along the lines of, “I’d make sure to tell our base that our guy is pro-life because that’s what we’ve always done and that’s what works,” you might well be completely wrong.
“The Left has a full decade head start,” says Adam Schaeffer, the founder and chief science officer of Evolving Strategies, a firm that specializes in executing randomized, controlled trials to accurately predict how a message will affect voters. “Psych research has proven that people are pretty bad at introspection and analysis. When you ask them, ‘How is this message going to impact your vote?’ They’ll tell you something. But they’re thinking about it, they’re trying to come up with an answer, they’re cognating on it — [and] they give you an answer that might have nothing to do with the real answer.”
Conversely, persuasion science is in vogue on the left, from progressive super PACs and advocacy organizations to Democratic political campaigns. Schaeffer tells me his firm is continually having to educate conservative donors, advocacy groups, and potential clients on the basics of designing a message through persuasion research and controlled trials, whereas it’s a standard part of how progressives actually run campaigns and map out strategy.
Schaeffer’s background is in academic political science, where there’s been a real move toward running randomized experiments to find out what is effectively affecting people and what isn’t. Social-science researchers have begun to see the benefits of setting up their studies using hard-science practices.
Only by running controlled tests can researchers distinguish between what people think or say may change their vote and what will actually change their vote.
“Pretty much every month, you hear about a new observational study of diet and how it links to cancer, or say, hormone-replacement therapy that was thought to be helpful for women because of observational data and correlations between health outcomes and that treatment,” Schaeffer explains. “But when they did a randomized, controlled trial, the data overturned [those prior assumptions] and the researchers found that the treatment was actually a net-negative for women.”
“You find this a lot in science. There’s a lot of reliance on correlation rather than causation. And people just assume that something that’s correlated is caused by it – but life’s complicated, right?” Schaeffer continues. “Diet and health are complicated, and politics is really complicated. What actually causes someone to vote for one candidate over the other is a very complicated question, and you can’t just go on correlations and what people think or say.”
But if message design is treated the same way as pharmaceutical research — where a group of test subjects is randomly assigned to receive a “placebo” control (say, a video with no political content) and another group of randomly assigned people see a treatment ad (one with a political message) — there’s no bias, and researchers can be confident that difference in outcome is a result of the advertising input alone, because everything else was the same.
This process can yield surprising results. Schaeffer tells me that in 2014, Evolving Strategies conducted a major effort in support of Joni Ernst’s senate campaign in Iowa that went after her opponent as an extremist on abortion. They conducted the message testing, modeling, and targeting for a large-scale persuasion campaign and found that those most receptive to pro-life messaging tended to be women who leaned left — soft Democrats.
“That was really counterintuitive to people. The typical way that Republicans message on these core issues is to the base: You give pro-life messaging to people who are modeled likely pro-life, pro-gun messaging to [voters] likely modeled pro-gun.”
On the other hand, if you can ascertain that, say, soft Democrats are open to voting for your candidate if only you manage to expose them to your message, your campaign will have a lot more room for growth. Pro-lifers are already likely to vote Republican. Sure, you could pour limited resources into a campaign to drive your numbers in that group up from 80 percent to 85 or 90 percent. Or you could target receptive Democratic women, who support your candidate perhaps only 30 or 40 percent of the time, but are a fertile source of converts. The latter strategy has more upside, since it carries with it the possibility of hurting your opponent, and less downside, since your targets were likely to vote for your opponent in the first place.
“We try to be agnostic about testing messages — about what’s going to work and what’s not,” Schaeffer tells me. “But a big part of it is: Are we introducing something that is new, something that’s potentially disruptive?”
If you’re telling voters something they already know about a candidate, you’re unlikely to change their vote.
Conservatives need to embrace persuasion and messaging research so they don’t find themselves flying blind again.
This phenomenon became extremely apparent in the 2016 Republican primary, when many people were shocked by the relative ineffectiveness of the ads run against Donald Trump. Schaeffer says that most of those ads reinforced impressions that voters had already priced in to Trump’s candidacy — the idea that he wasn’t a doctrinaire conservative, say, or the fact that he had said nasty things about women. Independent groups and opposing campaigns flooded the airwaves with ads that attacked Trump on a wide range of issues but failed to move the needle — because the messaging wasn’t introducing new, disruptive information to the voters.
For example, one Ted Cruz–allied group hit Trump with an ad that emphasized his “New York values” and previous liberal positions. Evolving Strategies tested the ad and found that it was an overall net-negative for Cruz: It was causing more voters to support Trump. Later, Schaeffer’s firm tested four Our Principles PAC anti-Trump ads questioning Trump’s conservative bona fides. Schaeffer found that the ads were completely ineffective in halting Trump’s momentum: Some voters identified Trump’s heterodox ideology as a positive, while most of the rest had decided that it didn’t matter.Schaeffer theorizes that an effective anti-Trump ad campaign would have focused on bringing voters new information, perhaps focusing on some of the, ahem, weirder things Trump has said, such as when he mentioned that he might have tried to date his daughter Ivanka if she wasn’t, you know, his daughter. Maybe some other line of attack would have been effective. Or maybe no single message could have downed Trump.
We can’t really know for sure because no conservative group decided to find out through scientific testing: They were all flying blind.
Going forward, conservatives need to embrace persuasion and messaging research so they don’t find themselves flying blind again. The last few cycles, the Right has fallen behind in the technological arms race. Even if conservatives catch up in some of those areas — the Cruz campaign made at least some progress this cycle — they have a lot of work to do if they want to put one of their own in the White House in 2020.
— Mark Antonio Wright is an assistant editor at National Review.