Although biology has moved on from the idea of bad air, debate around crime still focuses on bad people. Slutkin thinks this is in part because contagious violence is less intuitive than disease. ‘Here you don’t actually have an invisible microorganism that you can at least show somebody under the microscope.’ However, the parallels between infectious disease and violence seemed clear to him. ‘I remember an epiphany when I asked someone “what’s the greatest determinant of violence? What’s the greatest predictor?” And the answer was “a preceding violent event”.’ In his mind, it was an obvious sign of contagion. Which made him wonder: perhaps methods used to control infectious diseases could be applied to violence too?
There are several similarities between outbreaks of disease and violence. One is the lag between exposure and symptoms. Just like an infection, violence can have an incubation period; we might not see symptoms straight away. Sometimes a violent event will lead to another one soon after: for example, it might not take long for one gang to retaliate against another. On other occasions it may take much longer for knock-on effects to emerge. In the mid-1990s, epidemiologist Charlotte Watts worked with the World Health Organization (who) to set up a major study of domestic violence against women.[7] Watts had trained as a mathematician before moving into disease research, focusing on hiv. As her work on hiv developed, she started to notice that violence against women was influencing disease transmission because it affected their ability to have safe sex. But this revealed a much bigger problem: nobody really knew how common such violence was. ‘Everybody agreed that we needed population data,’ she said.[8]
The who study was the result of Watts and her colleagues applying public health ideas to the issue of domestic violence. ‘A lot of previous research treated it as a police issue or focused on psychological drivers of violence,’ she said. ‘Public health people ask, “What’s the big picture? What does the evidence say about individual, relationship and community risk factors?”’ Some have suggested that domestic violence is completely context or culture specific, but this isn’t necessarily the case. ‘There are some really common elements that consistently come out,’ Watts said, ‘like exposure to violence in childhood.’
In most of the locations in the who study, at least one in four women had previously been physically abused by a partner. Watts has noted that violence can follow what’s known in medicine as a ‘dose-response effect’. For some diseases, the risk of illness can depend on the dose of pathogen a person is exposed to, with a small dose less likely to cause severe illness. There’s evidence of a similar effect in relationships. If a man or woman has a history involving violence, it increases the chance of domestic violence in their future relationships. And if both members of the relationship have a history of violence, this risk increases even further. This isn’t to say that people with a history involving violence will always have a violent future; like many infections, exposure to violence won’t necessarily lead to symptoms later on. But like infectious diseases, there are a number of factors – in our backgrounds, in our lifestyles, in our social interactions – that can increase the risk of an outbreak.[9]
Another notable feature of disease outbreaks is that cases tend to cluster together in a certain location, with infections appearing over a short period of time. Think about that cholera outbreak in Broad Street, with cases clustered around the pump. We can find similar patterns when looking at violent acts. For centuries, people have reported localised clusters of self-harm and suicide: in schools, in prisons, in communities.[10] However, clustering of suicides doesn’t necessarily mean contagion is happening.[11] As we saw with social contagion, people may behave in the same way for another reason, like some shared feature of their environment. One way to exclude this possibility is to look at the aftermath of high-profile deaths; a member of the public is more likely to hear about the suicide of a well-known person than the other way around. In 1974, David Phillips published a landmark paper examining media coverage of suicides. He found that when British and American newspapers ran a front-page story about a suicide, the number of such deaths in the local area tended to increase immediately afterwards.[12] Subsequent studies have found similar patterns with media reports, suggesting that suicide can be transmitted.[13] In response, who have published guidelines for responsible reporting of suicides. Media outlets should provide information about where to seek help, while avoiding sensational headlines, details about the method involved, and suggestions that the suicide was a solution to a problem.
Unfortunately, outlets often ignore these guidelines. Researchers at Columbia University noted a 10 per cent rise in suicides in the months following the death of comedian Robin Williams.[14] They pointed to a potential contagion effect, given that many media reports about Williams’ death did not follow who guidelines, and the largest increase in suicides occurred in middle-aged men using the same method as Williams. There can be a similar effect with mass shootings; one study estimated that for every ten US mass shootings, there are two additional shootings as a result of social contagion.[15]
Because there is often an immediate rise in suicides and shootings following such media reports, it suggests that the delay between one contagious event and another – known in epidemiology as the ‘generation time’ – is relatively short. Some clusters of suicides have involved multiple deaths over a matter of weeks: in 1989 there was an outbreak of suicides at a Pennsylvania high school, which saw nine attempts in eighteen days. If these events were the result of contagion, the generation time may in some cases have been
