The 2016 US presidential election was marred by reports of political violence not seen for many years. Was this a real increase or just a perception blown up by the media? The science is in, and election rallies did spark extra violence in host cities, but only those of one candidate. It's not hard to guess which one.

"News media sources reported there were violent incidents at some campaign rallies, but it was difficult to gauge whether there really was a systematic problem, and if so, how many additional assaults were associated with each rally," Dr Christopher Morrison of the University of Pennsylvania said in a statement. It's also not necessarily easy to distinguish between assaults that happen inside stadiums where rallies are being held, and those occurring nearby, for example as protesters try to get in.

Morrison decided to try to address these questions by examining statewide city crime rates across entire cities on days when rallies were held, and comparing these with those in the same cities on equivalent days either side of the rallies. Cities with less than 200,000 people were excluded from the sample.

This left 31 Trump rallies in 22 cities, along with 38 Clinton rallies in 21 cities. There was no statistically significant difference in the amount of violence in cities Clinton visited on the days she held the rallies compared to 7, 14, 21, and 28 days before and afterward. Just to be sure, Morrison also controlled for weather conditions.

Trump rallies, on the other hand, were associated with an average of 2.3 more assaults being reported to the police in the host city, 21.7 compared to 19.4.

"To prevent similar violence in the future, it is important to understand the underlying causes of this behavior, perhaps including the role that political rhetoric might play in normalizing or promoting violence," Morrison added.

Of course, this information, if not simply dismissed as “fake news” despite publication in the peer-reviewed journal Epidemiology, is open to partisan interpretation. Trump supporters will allege it was all the fault of trouble-makers looking to disrupt his rallies, while his opponents will suggest that exhortations to “knock the crap out of” people speaking up in opposition, and offering to pay legal fees, might have had more to do with it.

It's not clear if that's a question science can settle, but Morrison is also interested in one that should be more testable. This is whether all the extra assaults occurred in the rallies' vicinity, or if the anger the rallies produced increased assaults elsewhere as a result of emotional contagion. "Violent language may have affected the mood and behavior of rally attendees, as well as those exposed to the rally through news reports and social media,” argued senior author Dr Douglas Wiebe.