The “shame” treatment raised turnout by +6.3 percentage points; the “pride” treatment raised turnout by +4.0 percentage points.
We already know from Gerber et al. (2008) that social pressure can boost voter turnout. As part of Political Behavior‘s special issue on social pressure and turnout (read some background), Gerber et al. join forces again with a new question: Does negative pressure (shame) work better than positive pressure (pride)?
To find out, they sampled 369,211 Michigan voters in 2007. All these voters had voted in either the 2005 or 2006 elections, but not in both. Each voter was sent a copy of his voting record, but only for one of these two elections. Voters were randomly assigned to receive a record that included the election they voted in or the election they abstained from.
These two mailers were identical in every other way. They contained the exact same wording. The only difference was whether the voter history showed “did not vote” or “voted” as the only entry in the voting history.
Both treatments raised turnout, of course. In 2008, Gerber et al. found that sending voters their voting history boosts turnout by around +4.9 percentage points (that’s the “intent-to-treat” effect). Likewise, the average result in the present study (averaging across both treatments, shame and pride) was a +5.2 percentage point boost in turnout.
However, pointing out an earlier abstention had a stronger effect than pointing out an earlier vote. The “shame” treatment raised turnout by +6.3 percentage points; the “pride” treatment raised turnout by +4.0 percentage points.
Bear in mind, of course, that 2007 was a low-salience election in Michigan. If this study were replicated in a high turnout year like 2008 or 2012, the boost would be a bit smaller. That’s true of all turnout experiments, as discussed here.
Comment and criticism
This issue of Political Behavior also includes an attempt by Costas Panagopoulos to assess whether shame motivates voters better than pride. It’s worth comparing them for a moment.
Conceptually, Panagopoulos’s experimental treatment seems to be a more persuasive operationalization of “shame” and “pride.” Panagopoulos told some voters that those who turned out would be honored by having their names listed in the local newspaper; he told others that abstainers would be shamed by having their names in the local paper. This treatment seems more likely to induce feelings of shame or pride. By contrast, the present study (Gerber et al.’s) may or may not be tapping into shame and pride. As they concede in their discussion, it’s possible that those who received the “pride” treatment saw their positive turnout record as justification for staying home. Rather than feeling pride that they were good voters, they may have felt that they had already fulfilled their civic duty by voting in the past, so they could justify staying home.
Empirically, however, Gerber et al. use better experimental methods than Panagopolous. As I noted in my review of Panagopoulos, he sent his “shame” and “pride” treatments to different populations, so the effects are not directly comparable. By contrast, Gerber et al. sent their shame and pride treatments to the same population, so the effects are directly comparable.
Each study has its flaws, then. Taken together, though, they seem to strongly suggest that shame does, in fact, motivate voting better than honor does.