VoxEU Column Politics and economics

Racial prejudice, dog-whistle politics, and police behaviour

The US saw a resurgence of racial and ethnic turmoil and tensions during Donald Trump’s presidency. This column examines how Trump's 2016 presidential campaign affected police behaviour towards Black Americans, using data on nearly 12 million vehicle stops by the police in 142 counties where Trump held a rally during his campaign. It finds that the probability that a driver stopped by police would be Black increased after a Trump event, and estimates that Trump's rallies may have led to nearly 30,000 additional stops of Black people by the police in the months following his rallies. 

The US saw a resurgence of racial and ethnic turmoil and tensions during Donald Trump’s presidency, which culminated in the violent protests and forced entry to the Capitol by Trump supporters on 6 January 2021. This latest event followed a series of others, including the neo-Nazi ‘Unite the Right’ rally in Charlottesville in 2017, the Pittsburgh synagogue shooting in 2018, and the El Paso shooting in 2019. Indeed, it was simply the continuation of a worrying phenomenon that only grew over Trump’s presidency and, as we argue in this column, was amplified by his presidential 2015/2016 presidential campaign. 

The behaviour of the police, supposed to provide a rampart against extremist violence, has also come into question. At the Capitol on 6 January, the contrast with the violence and brutality that police regularly uses against African Americans, as recently exemplified by the murder of George Floyd by white Minneapolis police officer Derek Chauvin, was only too striking. As noted recently in a recent New York Times column by historian Timothy Snyder, Trump’s inflammatory racial rhetoric might have comforted the views of “supporters convinced that the enemy was at home” (a euphemism for the police).1 

Some have pointed directly to Trump's inflammatory leadership as the reason for this turmoil. Recent research shows how his campaign and election changed the expression and acceptability of xenophobic views and discrimination against the people targeted by his racially inflammatory rhetoric – namely, Hispanics and Muslim migrants (Bursztyn et al. 2020, Newman et al. 2020). 

In new research (Grosjean et al. 2020), we focus on prejudice against Blacks – in the population as a whole but also particularly among the police. While Trump’s campaign did not explicitly disparage African Americans, numerous political science and law studies show how speech can carry a hidden message that is only understood by a targeted subgroup and that activates threatening stereotypes – a phenomenon known as the ‘dog-whistle effect’ (Lohrey 2006, Fear, 2007, Goodin 2008, Haney-Lopez 2014). As (now) President Joe Biden put it in during his presidential election debate with Donald Trump, on 30 September 2020,2 “[t]his is a president who has used everything as a dog whistle to try to generate racist hatred, racist division.” Put plainly, this would imply that by invoking threatening outgroups of drug dealers, criminals and “rapists”,3 Trump was in fact stirring deep-seated racial divisions against Blacks.  

Motivated by this, we study how Trump's 2016 presidential campaign affected the expression of racial prejudice and discrimination against Black Americans in one of its most fundamental and violent dimensions – namely, police behaviour. To do so, we use data on nearly 12 million vehicle stops by the police in 142 counties where Trump had held a campaign rally in 2015 or 2016, as shown in Figure 1.

Figure 1 Counties with campaign events and police stops


Notes: The map shows the 142 counties for which we have data on police stops and Trump’s 2015-2016 campaign events.

We find that the probability that a driver stopped by the police is Black increases after a Trump rally. As shown in Figure 2, the effect is immediate, constant in magnitude for the first 30 days after a rally, and lasts for at least 50 days. Our estimates suggest that Trump's rallies led to nearly 30,000 additional stops of Black people by the police in the months following the events. 

Figure 2 Impact of Trump rallies on police stops of Black commuters


Moreover, we observe no effect for offences that would automatically trigger a police stop or police intervention such as accidents, collisions, criminal offenses including assault, fleeing the scene of a crime, fleeing the police, speeding, driving an unregistered car or a car with a defective license plate, or brake violations. Instead, the effect is entirely explained by discretionary stops by police, where the offence is either only revealed after a stop (e.g. driving without a license) or subject to value judgement and discretion by the officer (e.g. following another vehicle too closely or generally ‘careless’ driving). 

We do not detect any change in the driving or criminal behaviour of drivers. We also do not observe any increase in the rate of stops of other minorities, or any effect of rallies held by Hilary Clinton (the Democratic candidate for the presidency in 2016) or by the other main contender to the Republican nomination in 2016, Ted Cruz. Taken together, our results suggest that the change in police behaviour after a Trump rally is unjustified and racially motivated, specifically against Blacks. 

Although the effect of large presidential rallies is larger, we nonetheless find a significant effect of smaller rallies, earlier in the race, when Trump was still only a marginal candidate for the Republican nomination. Four years into his presidency, one can only speculate about the total and cumulative effect of his presidency on police behaviour towards African Americans.4

At first, these results may seem odd given that, as we observed earlier and discuss in our paper based on Trump’s campaign speeches, Trump's inflammatory remarks during his campaign were often directed against Hispanics and Muslims, not Blacks, whereas we only observe an increase in police stops of Black drivers. We interpret this result as the manifestation of the dog-whistle effect, whereby coded language appeals to deep-seated stereotypes of groups that are perceived as threatening. In other words, individuals prejudiced against Blacks heard in Trump’s speeches a confirmation of their racist views. 

We provide two pieces of empirical evidence that supports our interpretation. Firstly, we show that the effect of Trump’s rallies is much larger in areas with more racist attitudes today, or that relied more on slavery historically. By contrast, we do not find any heterogeneity of the effect along other dimensions, including income, education, political affiliation or severity of the recent trade shock with China. 

Secondly, we use data from a 2016 online experiment (Collingwood 2020) to show that respondents who were already prejudiced against Blacks became more prejudiced when exposed to Trump's inflammatory speech on violent Hispanic immigrants. Strikingly, the effect is specific to prejudice against Blacks (as opposed to other racial minorities, such as Asians or Hispanics, for whom no effect is observed) and specifically that Blacks are violent (as opposed to other dimensions of prejudice against blacks, for example that they might be lazy, or lack intelligence). 

Overall, our research shows how the rhetoric used by highly visible individuals can have tangible consequences. Even when Donald Trump was not holding any governmental position and many polls placed him as unlikely his presidential victory, we show that his rallies had an effect on police behaviour towards Black drivers. This is particularly striking given that the explicit xenophobic views of Trump were almost completely targeted towards Hispanics and Muslims. This result contributes to the ongoing debate regarding the value and cost of freedom of speech.

These findings are also of significant policy relevance because of the many indirect effects that racially targeted behaviour by police may have on minorities. For example, unjustified police repression can act as a way of voter suppression, when disabused citizens extend their lack of trust in the police to all public institutions. As recent research shows (Williams 2019), historical discrimination and violence against Blacks is associated, to this day, with lower voter registration by Black voters. 

Editors’ note: An earlier version of this coloumn was published on on 6 August 2020. 


Bursztyn, L, G Egorov, and S Fiorin (2020), “From Extreme to Mainstream: The Erosion of Social Norms.” American Economic Review 110(11): 3522-48.

Collingwood, L (2020), “Replication Data for: The Trump Effect An Experimental Investigation of the Emboldening Effect of Racially Inflammatory Elite Communication”. 

Fear, J (2007), “Under the Radar: Dog-whistle Politics in Australia", The Australia Institute 96.

Goodin, R E (2008), Innovating Democracy: Democratic Theory and Practice after the Deliberative Turn, Oxford University Press (reprint ed.).

Grosjean, P, F Masera, and H Yousaf (2020), “Whistle the Racist Dogs: Political Campaigns and Police Stops”, CEPR Discussion Paper 15691.

Haney-Lopez, I (2014), Dog Whistle Politics: How Coded Racial Appeals Have Reinvented Racism and Wrecked the Middle Class, Oxford University Press.

Lohrey, A (2006), Voting for Jesus: Christianity and Politics in Australia, Black Inc.

Newman, B, J L Merolla, S Shah, D Casarez Lemi, L Collingwood, and S K Ramakrishnan (2020), “The Trump Effect: An Experimental Investigation of the Emboldening E_ect of Racially Inflammatory Elite Communication”, British Journal of Political Science, 1-22.

Williams, J A (2020), “Historical Lynchings and Contemporary Voting Behavior of African Americans”, mimeo, Clemson University.


3 During his presidential announcement speech on 16 June 2015, Trump remarked: “When Mexico sends its people, they're not sending their best [...] They're sending people that have lots of problems, and they're bringing those problems with them. They're bringing drugs. They're bringing crime. They’re rapists”.

4 Although fatal encounters between police officers and black civilians have tended to capture more of the media's attention, over-enforcement of minor infractions and the kind of unjustified stops by the police that we document in this paper provide a daily and generalized expression of discrimination against minorities. According to Bureau of Justice Statistics, 8.6% of US residents aged 16 and over, more than 20 million people, were pulled over by the police during a traffic stop in 2015. Our own estimates suggest that, at baseline, Black drivers overrepresented by a factor of two compared to their proportion in the population. 

2,415 Reads