by By Nathan Collins
from Pacific Standard
If we’ve learned anything about voter turnout recently, it’s that people can be manipulated into casting a ballot. In a particularly startling example, a set of2011 experiments found that subtle differences in language can increase turnout by 10 to 15 percentage points—a result that garnered quite a good dose of attention in the media (as well as hope for an elixir to perennially low turnout figures).
Well, a bit of depressing news: A much larger field study published today in the Proceedings of the National Academy of Sciences—the same journal that published the 2011 experiments—has failed to replicate the original findings, meaning the results were not quite so promising as many had hoped, if not an all-out fluke.
“One of the most exciting areas of research in psychology is the discovery of brief psychological interventions that can cause substantial behavioral change,” and the 2011 voter turnout study is a particularly important example, writes a team led by Yale University political scientist Alan Gerber. The original idea was that getting people to think of themselves as voters could get people to vote more—but, Gerber and his colleagues write, “we find little evidence that priming individuals to think of themselves as voters produces higher participation rates than when describing them as engaged in the act of voting.”
This time, the researchers found no discernible difference between the noun and verb conditions.
Here’s how the original study worked: Across three experiments, the researchers asked a total of 336 participants questions phrased either in terms of a noun or a verb—for example, “How important is it to you to be a voter in the upcoming election?” versus “How important is it to you to vote in the upcoming election?” The first phrasing, the researchers found, increased turnout between 10 and 15 percentage points, depending on the election and the group of people being studied. If that’s to be believed, thinking in terms ofbeing a voter could dramatically increase voter turnout.
Well, yes. Gerber and his team put out a similar survey in three states during the 2014 primaries. They used four scripts—the two from the original experiment, a “placebo” that served as an experimental control, and a standard get-out-the-vote message—and reached a total of 11,099 people, a sample far larger than the original. They followed up with state and local records to determine whether participants had actually voted.
This time, the researchers found no discernible difference between the noun and verb conditions. The “voter” version increased actual voter turnout by 0.9 percent over the placebo control, while the “vote” version increased it by 1.3 percent. Among those who got the standard get-out-the-vote script, turnout increased increased by 2.1 percent compared to the control group.
The original experiments were conducted during presidential and gubernatorial elections, so it’s possible Gerber and his team didn’t replicate the original findings simply because voters get less excited about primaries. If so, the original results are “highly sensitive” to electoral context and other factors—in which case, well, at least we all learned something.