Short-Term Association Between Ambient Temperature and Homicide in South Africa Open Access

Gates, Abigail (Spring 2019)

Permanent URL:



Criminology research has traditionally examined sociodemographic predictors of crime, such as sex, race, age, and socioeconomic status. However, evidence suggests that short-term fluctuations in crime vary more than long-term trends, which sociodemographic factors often cannot explain. This has redirected researchers to explore how environmental factors, such as meteorological variables, influence criminal behavior. Current evidence on the

relationship between temperature and homicide is relatively limited and uncertain. This study investigates the association between daily ambient temperature and homicide incidence in South Africa, a country with one of the highest homicide rates in the world.


Data were analyzed using a case-crossover design with conditional logistic regression. Cases (n=205,932) were deaths with an immediate or underlying cause of death recorded as either homicide (ICD-10 codes X85-Y09) or due to other violent causes of undetermined intent but likely to be homicides (ICD-10 codes Y22-Y25, Y28-Y29). Case periods were defined as the day on which a death occurred. Control periods were selected using a day-of-week match within the same month and district. Analyses investigated same-day and lagged effects of maximum, mean and minimum temperature as a linear relationship with an additional investigation of possible non-linearities.


A one-degree Celsius increase in same-day maximum temperature was associated with a 1.3% (1.2-1.5%) increase in homicide. Significant positive associations remained when applying other temperature metrics (mean, minimum), lags (1, 0-1), and when stratifying by definite versus total homicides. The shape of the association did not display any clear nonlinearities. There was no evidence of confounding by public holidays or interaction by district.


This study suggests a small but positive association between daily ambient temperature and homicide in South Africa and provides insight as to the shape of this relationship. This temperature-health relationship may be of particular concern in the context of climate change. The ability to include meteorological variables as a predictor of criminal activity and violent behavior could prove valuable in resource allocation for crime prevention efforts, policy, and preparedness.

Table of Contents

Page 1. Introduction

Page 3. Methods

Page 5. Results

Page 6. Discussion

Page 9. Conclusions and Recommendations

Page 10. References

Tables and Figures

Page 14. Table 1: Descriptive statistics for temperature and mortality data

Page 15. Table 2: Association between ambient temperature and homicide incidence with odds ratios reported per °C increase in temperature

Page 16. Table 3: Results for maximum temperature while excluding deaths reported as occurring on South African public holidays

Page 16. Table 4: Results for maximum temperature for deaths 2002-2013

Page 17. Figure 1: Counts of deaths recorded per year

Page 18 Figure 2: Association between maximum temperature and homicide incidence using categories of temperature in comparison to the reference group of <18 °C


Pages 19-20. Appendix 1: ICD-10 codes of deaths selected for this study

Page 21. Appendix 2: Exact numerical results for Figure 2

About this Master's Thesis

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
  • English
Research field
Committee Chair / Thesis Advisor
Last modified

Primary PDF

Supplemental Files