The association between temperature and interpersonal violence
Warmer temperatures have a known association with violent behaviour, a relationship that is likely to be intensified by climate change. However, very little is known about how this relationship varies by factors like demographics, crime type or area. A better understanding of these factors would allow for targeted response strategies as well as improving projections of violence in a warming world. This thesis aims to investigate temperature and seasonal trends relating to aggressive behaviours (physical and virtual), as well as the disproportionate effect of temperature-related violent crime by demographics, premises and urban form. The thesis combines a range of datasets across weather, crime, online sentiment, socio-economics and demographics, premises, urban heat island (UHI) effect and green cover, using the case study region of New South Wales (NSW) and the capital city of Sydney, Australia, generally over a 13-year period (2006-2018).
Key findings were firstly that overall, incidents of violent crime in NSW and Sydney displayed a significant peak in summer and a non-linear association with temperature. The response curve of temperature-related violence was largely uniform by victim demographics; increasing temperature and assault trends were consistent regardless of victim sex or age, or socio-economic status of the area where an assault occurred. The exception was incidents involving child victims, which increased in moderate temperature then declined in hot temperature. While it could be expected that the effect of temperature on violence is less within the shelter of the indoors, this study found the inverse: both domestic and non-domestic violence had a steeper temperature related increase inside than outside, and this was particularly striking for domestic violence. Sexual assault displayed a peak at around 30°C regardless of premises.
Spatially, ambient air temperature did not display a significant relationship with crime rates; however, the effect of UHI did: suburbs with a moderate UHI effect had significantly more violent crime than those with a low UHI effect, even after controlling for spatial and demographic variables. Urban greening, which has a known association with both crime and temperature modulation, was tested in the Sydney region, finding that crime significantly increased with grass cover, while suburbs that had the greatest amount of tree and shrub cover had significantly less crime than those with the least.
The effects of temperature on physical violence were compared to virtual aggression (aggregated angry Twitter posts), asking if the daily count of angry Twitter posts could help predict the likelihood of violent crime. However, the temperature trends were largely the inverse; online aggression was highest at cooler temperatures, and a decreasing angry tweet count was associated with increasing assaults. Trends in the non-violent crime of fraud were also investigated, finding that incidents displayed no seasonality or association to temperature.
There are many social theories proposed to help explain the association between temperature and violent behaviour, with the thesis finding that in some instances these theories were supported and in others they were not supported. For example, routine activity theory applied to temperature proposes that increased socialisation leads to increased violent crime, often from strangers. However, the thesis found that domestic violence inside increased with temperature, indicating that violence is occurring behind closed doors and from known offenders. The General Aggression Model, however, was able to support the many interconnected and multifaceted roles of social, cognitive, personality, developmental and biological factors in the likelihood of aggression, which helped explain why the response curve between temperature and aggression was widely experienced but varied in some regards.
While this thesis did not directly propose means to address temperature-related violence, the findings suggest that some of the common heat and/or violence response strategies are mutually supportive, while others are not. For example, some types of greenspace are associated with a reduction in both crime and urban temperatures. However, other responses like staying indoors in extreme heat may have unintended consequences on incidents of violence.
The thesis is the first to provide a comprehensive assessment of the effects of temperature on violent behaviour in the previously unstudied region of NSW and Sydney. Additionally, the thesis demonstrates that the temperature–violence response curve varies by behaviours, location, demographics and urban form. The findings are valuable for developing targeted heat response strategies as well as better predicting the effects of temperature on violence in a warming world.