A protestor faces police enforcing a curfew April 28 in Baltimore. A line of police behind riot shields hurled smoke grenades and fired pepper balls at dozens of protesters to enforce a citywide curfew. (Matt Rourke/AP)

Examining county-level voting patterns, I find that black-led protests in which some violence occurs are associated with a statistically significant decline in Democratic vote-share in the 1964, 1968 and 1972 presidential elections. Black-led nonviolent protests, by contrast, exhibit a statistically significant positive relationship with county-level Democratic vote-share in the same period. Further, I find that in the 1968 presidential election exposure to violent protests caused a decline in Democratic vote-share. Examining counterfactual scenarios in the 1968 election, I estimate that fewer violent protests are associated with a substantially increased likelihood that the Democratic presidential nominee, Hubert Humphrey, would have beaten the Republican nominee, Richard Nixon.

Those are the central findings of a paper by Omar Wasow, an assistant professor in the Department of Politics at Princeton University. He married data on 1960s protest activity to data on presidential voting in every county in the continental United States.

Specifically, he compared counties where a protest of at least modest size (involving 10 or more protesters) had taken place within 100 miles at least once in the two years prior to each election. In counties where a nonviolent protest met these criteria, Democratic vote share was nearly 1 point higher than in counties that didn’t meet these criteria.

But in counties where a violent protest met these criteria, Democratic vote share was 1 to 1.5 points lower, depending on the exact statistical model. The paper does additional work to determine whether this relationship is plausibly causal. As noted above, this relationship is large enough to cost Humphrey four states (DE, IL, NJ, and OH) in 1968 under the hypothetical that there were no violent protests after the assassination of Martin Luther King Jr. in April of that year.

Wasow’s paper is not attempting to address contemporary politics. But an obvious question is whether his findings have implications for the potential impact of the protests after the deaths of Michael Brown, Eric Garner, and Freddie Gray.

On the one hand, there are obviously many, many differences between the 1960s protests and those in 2014-2015.  Radley Balko discusses this at length here. To pick just one, the 1960s protests were larger and more numerous.

On the other hand, underlying Wasow’s paper is a theory about how nonviolent and violent protests have a different psychological impact. This theory could have some relevance now.

Wasow argues that nonviolent protests trigger an identity that bridges disparate groups — and, in particular, creates more sympathy among majority groups for the minority group that is protesting and therefore the potential for a cross-racial coalition.

But violent protests have the opposite effect — strengthening each group’s own identity and creating a “circle the wagons” mentality motivated by a concern for safety.

That said — and despite some confident predictions — it remains far from clear whether and how recent protests could matter in 2016. Regardless, Wasow’s paper is very interesting and worthy of note.