Recent incidents of police violence, such as the murder of George Floyd in May 2020, have sparked large-scale protests in the US and beyond. In new research, Desmond Ang and Jonathan Tebes examine the effects of police violence on political engagement. Using data from Los Angeles County, he finds that voter participation jumps sharply immediately after and in areas close to a police-involved killing and remain higher for several years, especially among Black and Hispanic citizens, who are 8 percent and 5 percent more likely to register to vote. Such increases in civic participation are also three to four times larger following police killings of unarmed individuals.
Police violence and political engagement
In recent years, acts of police violence have garnered significant public attention. The murder of George Floyd compelled an estimated 26 million Americans to march against police brutality and systemic racism. Just this summer, the police killing of a Black teenager, Nahel Merzouk, in Paris sparked widespread protests around France. These events have a long historical precedent. Since the 1960s, the four largest episodes of urban unrest in America were all triggered by police use of force.
At this point, we’ve seen enough examples to know that these tragic events can trigger explosions of social turmoil and unrest, but how do they affect formal political engagement – whether individuals register or vote, and for what causes? This is a complicated question. On the one hand, police brutality is clearly a factor that generates significant concern among many communities. On the other hand, the police are perhaps the most visible arm of government and incidents that erode trust in law enforcement may thus erode trust in the larger political system as well.
In new research, we provide the first causal evidence of the impact of police violence on voter participation. To do so, we combine highly detailed voter registration data from Los Angeles County with novel incident-level data on the timing, location, and context of nearly three hundred police killings spanning almost a decade.
Given that police killings are far more likely to occur in disadvantaged and minority neighborhoods, a simple comparison of voter turnout in neighborhoods with high and low rates of police violence is likely to be confounded by a number of correlated factors. To identify causal effects, we instead compare changes in civic engagement before and after police killing in the exact Census blocks (neighborhoods make up about 11 census blocks) where a killing occurred to changes among surrounding blocks in the same neighborhood sharing nearly identical demographic and economic characteristics. This approach leverages the fact that, in contrast to the intense media coverage of George Floyd’s murder, most police killings receive little or no media attention. In our sample, over 80 percent of police killings were never mentioned in local newspapers, suggesting that knowledge and impact of incidents is often limited to residents living nearby.
Tracking civic engagement in Census blocks
Using this approach, Figure 1 compares trends in voting and registration during the eight elections around each police killing for the exact blocks where incidents occurred against blocks in their surrounding neighborhoods. The coefficients on the left side of the red vertical line indicate that, prior to the police killings, trends closely mirrored each other across both sets of areas. However, looking to the right of the red line, we see that voter participation jumps sharply in incident blocks immediately after a killing and remains at elevated levels for several years. These results tell us that police killings, on average, increased local voter registrations and turnout in our sample by about 5 percent. It is important, however, to note that these effects do not rule out the possibility that some people are further disenfranchised by police violence. Indeed, a host of research has shown that individuals with criminal justice contact are some of the least likely to register and vote. Our findings tell us simply that, overall, any demobilizing effect is outweighed by increased mobilization among other individuals.
Figure 1 – Effects on Civic Engagement
These effects mask significant differences across demographic groups. Figure 2 below reveals that increased civic engagement is driven by Black and Hispanic citizens, who are 8 percent and 5 percent more likely to register as the result of exposure to local police killings, respectively. We find no impact on the political behavior of nearby whites and Asians. Larger effects are also observed for younger individuals and registered Democrats. These findings accord with a host of survey evidence documenting deep racial and partisan divisions in views of law enforcement, with minorities and liberals particularly concerned about use of force and racial bias in policing.
Figure 2 – Effects on different groups
While criminal charges are very rarely filed against involved officers (none in our sample), public backlash to police killings is often greatest for incidents involving unarmed victims. To examine this, we collected data on whether victims were armed with a weapon from district attorney incident reports. Consistent with national statistics, less than half of victims possessed a gun and about 20 percent were entirely unarmed. When we unpack effects based on weapon status, we find that increases in civic participation are three to four times larger following police killings of unarmed individuals.
Figure 3 – Effects by deceased weapon
Implications of our research
Our results suggest that community members may be turning out to hold law enforcement responsible for actions they view as unjust. That is exactly what we find when we investigate changes in policy preferences. Examining voting on referendums aimed at reducing the scope and scale of criminal justice penalties, we find a large increase in support for criminal justice reform in incident blocks relative to other blocks in their neighborhood, particularly after police killings of unarmed individuals (Figure 4).
Figure 4 – Effects on support for criminal justice reform
Together, our findings call to mind the same sentiments expressed in the Kerner Commission Report over 50 years ago: that police brutality contributed to a widespread belief among many communities of color in a “double standard of justice and protection.” Our empirical results suggest that these beliefs may spur Black and Hispanic citizens to strategically engage with formal electoral systems to hold institutions accountable.
Yet, what do they tell us about more recent events? Our data pre-dates the proliferation of social media and the advent of the Black Lives Matter movement, which greatly increased the visibility and importance of police violence. But even in our sample, which contained no nationally viral police killings, we can already see that the spatial reach of these events may be amplified by media coverage (Figure 5). Whereas the impact of police killings that were unmentioned in local newspapers are limited to the Census block of the incident, the effects of media-covered killings is felt up to half a mile away.
Figure 5 – Effects by media coverage
Thus, it is possible that the political importance of police violence has only grown over time. Indeed, data from the 2020 election suggests that Joe Biden’s victory was fueled by record turnout among minority youth, the same group we see most responsive to police killings in our study. In this light, we may expect police brutality to remain a dominant election issue until the concerns raised by the Kerner Commission are finally addressed.