Abstract
Although temperature aggression theory maintains that a high temperature engenders more aggressive behavior by irritating individuals, routine activity theory asserts that violent crime increases as temperature rises because of enhanced interaction among the public in outdoor settings. We investigate the effect of maximum daily temperature on whether crime victims are physically injured during the commission of an outdoor criminal offense in Cleveland, Ohio. We focus on violent crimes occurring outdoors because most U.S. households have central air-conditioning or room air conditioners. Two autoregressive integrative moving average (ARIMA) analyses provide support for routine activity theory because although maximum daily temperature has a strong positive effect on the frequency of violent crimes occurring outdoors, it has little influence on the physical injury of crime victims.
Keywords
Introduction
A substantial body of literature has accumulated that examines the relationship between temperature and violent crime. Despite some notable exceptions (Burke et al., 2017; Cheatwood, 1988; Rinderu et al., 2018), the general conclusion reached by researchers is that violent crime is more apt to transpire when the temperature is high. Studies find that high temperatures are associated with elevated violent crime rates (Cohn, 1990; Cotton, 1986; Dexter, 1899; Hsiang et al., 2013; Mishra, 2015; Rotton & Frey, 1985; Schinasi & Hamra, 2017; Tiihonen et al., 2017), increased murder rates (DeFronzo, 1984; Mavroudeas et al., 2018; Michel et al., 2016; Mishra, 2015; Rotton & Frey, 1985), and amplified robbery rates (DeFronzo, 1984; Sorg & Taylor, 2011; Tompson & Bowers, 2013). This documented positive relationship between temperature and violent crime is congruent with both temperature aggression theory and routine activity theory.
Several previous studies have utilized violent crime rates as a proxy for aggression (Hsiang et al., 2013; Simister & Cooper, 2005; Tiihonen et al., 2017), cultivating the notion of an underlying heat–aggression nexus. With the causal factors responsible for this underlying relationship still being debated, testing the presuppositions of theories purported to elucidate this relationship may furnish support for some theories over others. In this vein, the determination of whether temperature influences the likelihood of a crime victim being physically injured has theoretical relevance and will assist in testing a fundamental pillar of temperature aggression theory. Victim physical injury occurring during the commission of a violent crime is clearly relevant. Although both temperature aggression theory and routine activity theory postulate that a high temperature is predictive of violent crime, only temperature aggression theory hypothesizes that offenders will display an increased proclivity to demonstrate aggressive behavior with rising temperatures. This situation, in turn, increases the likelihood of violence (Anderson et al., 2000). This escalation in physical violence on the part of the offender is due to the temperature aggression theory’s assertion that a high temperature produces irritation and frustration.
In this study, we examine how maximum daily temperature affects the proportion of victims physically injured during the commission of violent crimes in Cleveland, Ohio, during 2016. We begin by discussing the relevance of temperature in temperature aggression theory and routine activity theory. We next explain the substantive differences in the underlying assumptions of these theories and discuss how these assumptions apply to the relationship between heat and aggression. Lastly, we demonstrate our efforts to appraise the applicability of temperature aggression theory and routine activity theory to the heat–aggression relationship. We focus on violent crimes occurring outdoors because most U.S. households have central air-conditioning or room air conditioners to cool the air and reduce temperatures within dwellings. Confirmed exposure to outside temperatures assures optimal conditions to assess the relationship between heat and aggression.
Theoretical Considerations
Temperature Aggression Theory
Two dominant positions exist in the literature to account for the relationship between heat and aggression. Rooted in the work of Quetelet (1969), temperature aggression theory proffers that elevated temperatures act to amplify the likelihood of frustration and impulsivity in individuals, which in turn engenders aggressive behavior. This thesis is most prominently identified in the work of Anderson et al. (2000), which postulates that the temperature aggression hypothesis is a conceptual framework wherein under select conditions, uncomfortable temperatures can provoke aggressive behavior. Furthermore, this aggressive behavior may evince itself in the form of violence because uncomfortable temperatures can foster discomfort and physical outbursts (Anderson et al., 2000). Temperature aggression theory also maintains that there may be an inverted U-shaped relationship between temperature and aggression because, at the tail ends of the temperature spectrum, the body’s ability to regulate temperature is overwhelmed and reorients the individual to prioritize bodily preservation (Anderson, 1989).
Temperature aggression theory finds a robust catalog of support from various academic disciplines. For instance, the link between heat and aggression exists in both animals (Takeshita et al., 2018; Vizzotto et al., 2015) and humans (Bell, 2005; Brewer, 2007; Kenrick & MacFarlane, 1984; Van de Vliert & Van Lange, 2019). Heat and high temperatures have augmented instances of police officers shooting potential crime suspects (Vrij et al., 1994), increased penalties related to overly aggressive behavior in National Football League (NFL) football games among home teams (Craig et al., 2016), and amplified feelings of hostility that indirectly produced aggressive thoughts (Rule et al., 1987). Elevated temperatures have also been documented to promote a heightened state of anxiety via a fight or flight response, which can provoke oscillations in aggression levels (Bell & Fusco, 1989).
The connection between heat and aggression is generally posited to be the result of uncomfortable levels of heat fostering feelings of hostility and agitation (Anderson, 2001). However, there is no consensus as to what mechanism is directly responsible for the connection between elevated temperatures and aggression. Simister and Cooper (2005) theorize that the heat–aggression nexus is hormonally driven, with high temperatures prompting impulsive behavior through the release of stress hormones. Similarly, Tiihonen et al. (2017) argue that elevated temperatures affect the serotonergic system because high temperatures modulate serotonergic transmission and likely increase impulsivity along with general human activity. Finally, Zillmann’s (1983a, 1983b) theory of excitation highlights the connection between external stimuli and an individual’s sympathetic nervous system. This theory assumes that when our excitatory nervous system is activated—in this case, through elevated temperatures—the reaction latency is nonspecific across emotions. Therefore, when negative stimuli are experienced by an individual and the sympathetic nervous system is provoked, misattributed anger often ensues because of the belief by the individual that one salient factor is responsible for his or her heightened feelings of arousal (Zillmann, 1983a). This situation can lead to criminal behavior becoming more physically violent, given that the perpetrator wrongfully attributes the sense of agitation to someone and not to the uncomfortable temperature.
Researchers have used the observed nexus between elevated temperatures and violent crime to gauge aggression levels via the violent crime rate (Baron, 1972; Bell & Baron, 1976; Bushman et al., 2005; Cohn, 1990; Harries & Stadler, 1988; Simister & Cooper, 2005; Tiihonen et al., 2017). The use of the violent crime rate as a surrogate measure for aggression levels remains widespread, despite studies showing that temperature does not affect the occurrence of violent crimes uniformly across crime types or climates.
Several studies underscore this inconsistency. Ranson (2014) analyzed monthly crime and temperature data from 2,997 U.S. counties, and found that temperature had a strong causal effect on the crimes of murder, manslaughter, rape, aggravated assault, and robbery. Schinasi and Hamra (2017) conducted a time-series analysis in Philadelphia, Pennsylvania, from 2006 to 2015 and found that the overall violent crime rate had an almost linear relationship with temperature, but that robberies had no relation to temperature after the temperature exceeded the average daily temperature. A study based on calls for police service in Minneapolis, Minnesota, found that high temperatures were associated with occurrences of domestic violence but had little effect on the occurrence of rape (Cohn, 1993). Hipp et al. (2004) analyzed several cities throughout the United States and reported that high temperatures increased violent crime rates the most in cities with varying seasonal climates. This finding casts slight aspersions on the consistency of temperature aggression theory, as individuals in uniformly warmer climates should be the most adversely affected by high temperatures.
Routine Activity Theory
Some researchers remain circumspect of temperature aggression theory because a positive association between temperature and violent crime is also compatible with routine activity theory. Cohen and Felson (1979) advanced the argument that crime patterns are the consequence of certain aspects of our sociostructural organization that coincide with everyday activities. Put simply, criminal activity occurs most frequently when suitable targets are available, capable guardians are absent, and when motivated offenders are present. Contrary to the rationale espoused by temperature aggression theory, routine activity theory employs a social and situational explanation for the observed fluctuations in criminal activity. Routine activity theory maintains that the underlying reason for why temperature and violent crime are interconnected is that a rise in temperature promotes outdoor activities among the population, which in turn enhances an individual’s vulnerability to criminal victimization.
The use of routine activity theory as a general lens to predict violent crime has proved moderately successful (Berthelot et al., 2015; Bunch et al., 2015; Fridell et al., 2009; Spano & Nagy, 2005; Stein, 2010). For example, routine activity theory has been linked to violent crime through spatial analysis and crime hotspots (He et al., 2017; Liu et al., 2016) and inequality indexes (Maume, 1989). However, the use of routine activity theory to explicate the relationship between temperature and violent crime has yielded only limited support. Hipp et al. (2004) provided support for routine activity theory by showing that temperature variation in moderate climate areas amplified the occurrence of violent crime. They theorized that violent crime would be most pronounced in areas with moderate climates because the opportunity of interaction would be maximized without the risk of uncomfortable temperatures. Similar seasonal oscillations in violent crime, with violent crime occurring in higher frequency during elevated temperatures, were also documented by others (Ceccato, 2005; Hu et al., 2017; Schinasi & Hamra, 2017).
Parsing the Difference
Although temperature aggression theory and routine activity theory are both supported in the literature, and each theory offers divergent conceptualizations of the causal factors responsible for the nexus between rising temperatures and violent crime, differentiating between them has proven exceedingly difficult because the directions of their respective predictions are identical. Each theory posits that a rise in temperature amplifies violent crime. Even the observation that violent crime is most pronounced in areas with moderate climates does not fully support routine activity theory because such a finding fails to explain increases in violent crime during the summer in geographical locations already experiencing hot climates (Hipp et al., 2004). Unseasonably warm temperatures can also muddy the distinctions between the two theories because individuals typically become accustomed to seasonal temperature norms and adjust their patterns of behavior and clothing accordingly. Nevertheless, research shows that there are limits to how flexible individuals are in responding to these unseasonably divergent temperatures (Sherwood & Huber, 2010). Thus, it is plausible that abnormally warm temperatures may produce elevated discomfort levels and higher rates of violence as reasoned by temperature aggression theory. It is also equally possible that uncharacteristically warm temperatures may elevate violent crime by motivating people to engage in outdoor social activities as proffered by routine activity theory.
In the current study, we use data drawn from the National Incident-Based Reporting System (NIBRS) and from other sources to help adjudicate between these differing points of view. The use of NIBRS aids in the investigation of the relationship between temperature and violent crime in several important respects. First, the use of the NIBRS enables the creation of a theoretically relevant measure of violence: whether the victim was physically injured during the commission of the crime. This measure, which has not been used previously by researchers, serves as a theoretically important outcome measure for our analysis because it helps us to more clearly differentiate between the claims articulated by temperature aggression theory and by routine activity theory. Although temperature aggression theory maintains that a high temperature irritates individuals and engenders more aggressive behavior, routine activity theory makes no such prediction. Routine activity theory merely argues that violent crime occurs more frequently as temperature rises because of enhanced interaction among the public in outdoor settings. It does not predict that high temperature increases the likelihood that an offender will physically injure his or her victim. In contrast, temperature aggression theory does make such a prediction because a high temperature is speculated to amplify aggressive behavior.
It is also important to acknowledge that many crimes deemed by researchers in prior analyses to be violent do not necessarily involve any physically aggressive behavior on the part of the offender. Take firearm crimes as an example. Victim physical injury is much less apt to occur when an offender uses a firearm because victims are more likely to acquiesce to the armed offender’s demands (P. J. Cook, 1980; Kleck, 1997). Conversely, when an offender does not use a weapon or uses a weapon other than a firearm such as a knife, victim injury is much more likely to transpire. Thus, it does not seem theoretically appropriate to deem all crimes involving a firearm as violent because many of these crimes involve much less physical violence directed at the victim than similar crimes committed without a firearm.
Second, the NIBRS affords us the ability to identify whether the physical injury experienced by a crime victim occurred outdoors. When one undertakes a cursory examination of the extant literature on temperature and violence, it is apparent that every study conducted to date combines indoor and outdoor crime in the same analysis. Researchers have employed violent crime, homicide, and property crime as their dependent variables, notwithstanding whether these crimes transpired indoors or outdoors. A significant problem with these dependent variables is that they fail to differentiate between indoor and outdoor crimes. The ability to distinguish outdoor violence from indoor violence is salient because upward of 87% of Americans have either central air-conditioning or room air conditioners to help cool the air (Energy Information Administration, 2011). The average American also spends the majority of his or her life indoors or in an automobile (Klepeis et al., 2001).
Thus, it stands to reason that if the temperature is related to violence occurring outdoors but not indoors because of the use of air-conditioning within dwellings, prior studies that amalgamated indoor and outdoor crime when measuring their dependent variable are endemically inconsistent with the implicit rationale associated with temperature aggression theory. Unless the dependent variable is constructed in such a way to differentiate outdoor violence from indoor violence, interpretation of the statistical results generated in prior studies will be problematic. This issue is addressed explicitly in the research reported here because only violent crimes occurring outdoors are examined. Indoor crime is used as a control variable in the analyses. Although not a perfect method, separating indoor and outdoor crime is the only available technique that comes close to reliably identifying whether the perpetrator, on some level, was exposed to elevated temperatures at the time the crime transpired.
Data
The daily data used in this analysis reflect a 1-year period (January 1 to December 31, 2016) for Cleveland, Ohio. The city of Cleveland is the second-largest city in Ohio and possesses a population of almost 400,000 residents (U.S. Census Bureau, 2010). Cleveland was selected for analysis because it is a racially diverse city, and due to the availability of daily crime data that could be used to distinguish between indoor and outdoor criminal activity. We used data calibrated in daily intervals because they help to attenuate the confounding of history effects that can threaten internal validity. In addition, because we analyze the impact of temperature changes on violent crime within a single jurisdiction over time and not changes in violent crime levels across jurisdictions, biases resulting from geographical differences in reporting practices are eliminated.
Dependent Variables
We analyze two dependent variables. Both these endogenous variables were derived from the NIBRS, which is an official incident-based reporting system that collects crime incident data from law enforcement agencies throughout the United States (National Archive of Criminal Justice Data, 2016). The first endogenous variable, violent outdoor crime, is measured as the daily number of reported violent crime incidents that occurred on a highway/road/alley/street/sidewalk, park/playground, field/woods, or lake/waterway/beach. The second endogenous variable is operationalized as the daily proportion of violent outdoor crime incidents where the crime victim sustained a physical injury. The proportion of outdoor violent crimes with physical injuries is used because the frequency of outdoor violent crimes and the frequency of outdoor crimes with physical injuries are highly correlated. As the number of violent offenses occurring outdoors rises, so does the number of outdoor crimes with physical injuries. Our use of the proportion of physical injuries instead of the frequency of physical injuries helps to circumvent this problem. Physical victim injury is defined as any apparent broken bones, possible internal injury, loss of teeth, severe laceration, unconsciousness, and other major and apparent minor injuries. The location-type code in the NIBRS is used to identify outdoor and indoor crime incidents.
Independent Variables
The variable of theoretical interest is the maximum daily temperature reported in Cleveland. During the period of observation, the maximum daily temperature in Cleveland ranged from 12 °F to 95 °F. The maximum daily temperature was utilized as it best represents the greatest possible effect temperature had on any given day. In accordance with temperature aggression theory, higher temperatures should provide the highest potential provocation for physical injuries to occur during the commission of a crime. Previous studies have found high maximum daily temperatures to be correlated with higher violent crime rates (Cotton, 1986; Michel et al., 2016). Data for the maximum daily temperature were obtained from Weather Underground, a commercial weather company that provides real-time and archived weather statistics across the United States (Weather Underground, 2019b). Weather Underground utilizes a robust network of weather stations that include about 2,000 Automated Surface Observation System (ASOS) stations maintained by the Federal Aviation Administration (FAA), more than 26,000 weather stations managed by the National Oceanic and Atmospheric Administration (NOAA), and more than 250,000 Personal Weather Stations (PWS), which are a part of Weather Underground’s expanding PWS network (Weather Underground, 2019a).
Two variables were used as controls in the analyses. The first control variable was the amount of rainfall measured in inches within a 24-hr period in Cleveland. The expectation is that there is a negative relationship between the amount of rainfall and crimes occurring outdoors because people tend to move indoors to avoid the rain (M. R. Cook et al., 2012; Horrocks & Menclova, 2011; Michel et al., 2016). Data for the number of inches of rainfall were also obtained from Weather Underground (2019b). In addition, because the temperature is expected to have little influence on indoor crime due to the widespread use of air conditioners in the United States, we were able to use violent indoor crime as a second control variable. Violent indoor crime is a robust control variable because deleterious social conditions such as poverty or high unemployment would be expected to affect both outdoor and indoor crimes similarly. In contrast, the maximum daily temperature should mainly influence outdoor crime. In addition, our use of indoor violent crime as a control series also negates the need to include only tangentially related control variables such as day of the week and holidays in the analysis. If outdoor violence is higher on weekends and holidays, it stands to reason that indoor violence will also be higher on weekends and holidays.
We measured violent indoor crime in two ways, depending on the equation estimated. In one equation, violent indoor crime is defined as the daily number of reported violent crime incidents that occurred at a residence/home, and robberies that occurred at a bar/nightclub, convenience store, department/discount store, grocery/supermarket, liquor store, or restaurant. In the second equation, violent indoor crime is defined as the daily proportion of reported violent indoor crime incidents with physical injuries. The means, standard deviations, and definitions for all the variables used in this study are reported in Table 1.
Description of Study Variables, 2016 (N = 366 days).
Note. Violent crime incidents include murder/nonnegligent manslaughter, kidnapping/abduction, rape, sodomy, fondling (indecent liberties/child molesting), robbery, aggravated assault, and simple assault. Physical injury includes any apparent broken bones, possible internal injury, loss of teeth, severe laceration, unconsciousness, and other major and apparent minor injuries.
Descriptive Analyses
We initially constructed a figure depicting the relationship between maximum daily temperature and outdoor violent crime over the 1-year observation period. A cursory glance at Figure 1 clearly shows that as the maximum daily temperature rises, the frequency of reported outdoor violent crime also increases. This finding is analogous to the results generated in previous research.
Outdoor violent crime incidents by maximum temperature.
We next constructed a figure of the relationship between maximum daily temperature and the proportion of victims physically injured in outdoor crimes. We are specifically interested in determining whether offenders become more violent in their crimes as the maximum daily temperature rises in Cleveland. Figure 2 readily shows that there is no visually discernable association between maximum daily temperature and victim injury. This finding suggests that high temperature may not necessarily be irritating offenders to the point where they have an enhanced proclivity to injure their victims during the commission of a crime physically.
Outdoor violent crime incidents with injuries by maximum temperature.
Although the positive bivariate relationship depicted in Figure 1 and the null finding shown in Figure 2 are both interesting, they do not inform us about whether other factors are also predictive of daily violent offending. To further delve into this issue, we estimated two multivariate autoregressive integrative moving average (ARIMA) models that included controls for daily rainfall and violent offending transpiring indoors.
ARIMA Results
The univariate ARIMA models for outdoor violent crime and physical injury in an outdoor crime were constructed through an iterative model-building strategy (McCleary et al., 2017). A univariate ARIMA model accounts for the stochastic processes associated with a series. Both series were stationary in variance and level. Once both series were determined to be stationary in variance and level, we inspected each series’ autocorrelation function (ACF) and partial autocorrelation function (PACF) for autoregressive processes and moving average processes. Our assessment of competing univariate models that were based on a visual examination of their ACFs and PACFs suggested that we use a (0,0,0) (0,0,1) outdoor violent crime model and a (1,0,0) outdoor physical injury model. The residuals for both these models were white noise (Ljung & Box, 1978).
The results for the maximum-likelihood ARIMA analyses appear in Tables 2 and 3. The conventional p < .05 level is employed to assess statistical significance. In line with the predictions derived from both temperature aggression and routine activity theory, Table 2 shows that there is a strong positive relationship between maximum daily temperature and the frequency of violent crimes occurring outdoors. As maximum daily temperature rises, outdoor violent crime increases. This finding is not that unexpected considering the results produced in prior research and our preliminary bivariate analysis depicted in Figure 1. Further examination of Table 2 reveals that both control variables are essential in explaining outdoor violent crime. There is a negative relationship between the amount of rainfall and violent criminal behavior occurring outdoors. As daily rainfall increases in Cleveland, violent crimes occurring outdoors decrease. In addition, as one might expect, there is a positive relationship between indoor and outdoor violent criminal activities.
ARIMA Model Predicting Violent Outdoor Crime.
Note. ARIMA (0,0,0) (0,0,1). Ljung-Box Q (17) = 27.041, p > .05. ARIMA = autoregressive integrative moving average.
p ≤ .05. **p ≤ .01. ***p ≤ .001 (two-tailed test).
ARIMA Model Predicting Violent Outdoor Crime With Injuries.
Note. ARIMA (1,0,0) (0,0,0). Ljung-Box Q (17) = 22.947, p > .05. ARIMA = autoregressive integrative moving average.
p ≤ .05. **p ≤ .01. ***p ≤ .001 (two-tailed test).
Table 3 reports the results for the relationship between maximum daily temperature and the daily proportion of reported physical victim injuries in outdoor violent crimes. Table 3 establishes that the effect of maximum daily temperature is trivial in magnitude and not of substantive importance. This null effect was graphically illustrated in Figure 2. In addition, in contrast to the first estimated equation reported in Table 2, the results in Table 3 reveal that the coefficient for the rainfall variable is essentially zero in the equation. The amount of daily rainfall has little impact on the physical injury of crime victims in outdoor crimes. Further examination of Table 3 also shows that indoor physical injury is salient in a positive direction. As the physical injury of crime victims increases indoors, it also rises outdoors.
Discussion
Although a comprehensive and diverse body of empirical work has accrued that examines the effect of temperature on violent crime, this research produces few substantive conclusions. Some empirical studies advance the tenets of temperature aggression theory (Bushman et al., 2005; Cohn & Rotton, 2005), whereas others evince evidence supportive of routine activity theory (Hipp et al., 2004; LeBeau & Corcoran, 1990; Rotton & Cohn, 2003). Based on these inconsistent findings, we felt that additional empirical research was necessary before a defensible position could be actualized regarding whether temperature aggression theory or routine activity theory has a greater value in elucidating the frequently observed positive association between temperature and crime.
The primary purpose of the present study was to expand upon prior research by exploring whether temperature influences the physical injury of crime victims. This outcome measure has not been used previously. It is of theoretical relevance because temperature aggression theory and routine activity theory make different predictions regarding whether higher temperatures influence whether a crime victim will be physically injured during the commission of a crime. Recall that a central aspect of temperature aggression theory is that a high temperature amplifies the likelihood that a crime victim will be physically injured because it influences offenders to become more aggressive when committing a crime. In contrast, routine activity theory makes no such prediction. It merely argues that warmer temperatures increase the frequency of violent crime by facilitating interaction among people in outdoor settings—the more interaction among people, the more likely that violent crime will occur. There is no suggestion by advocates of routine activity theory that the crime itself will be more violent when the temperature is high. One might also question why a high temperature would influence a significant amount of violence when most people in the United States are not continually exposed to the negative physiological or psychological consequences produced by a high temperature because they spend most of their time indoors in air-conditioned dwellings.
Although no single study can fully resolve this debate, the present study endeavored to unravel the theoretical claims proffered by temperature aggression and routine activity theory by investigating the effect of maximum daily temperature on victim physical injury in outdoor crimes. Initial ARIMA results showed that a rise in maximum daily temperature increased violent criminal offending transpiring outdoors, thereby supporting the predictions of both temperature aggression theory and routine activity theory. However, further analysis cast some doubt on the validity of temperature aggression theory because as the maximum daily temperature rose in Cleveland, the proportion of outdoor crimes that resulted in a crime victim suffering a physical injury remained unchanged. This finding bolsters the argument made by routine activity theory, which proffers that the positive relationship between temperature and crime observed in many prior studies, and this study, is being produced by an increase in the physical interaction among people in outdoor settings.
Our findings are in accordance with previous studies that found routine activity theory to be a useful explanatory tool for understanding the relationship between weather and crime. Despite different conceptual approaches, our findings yielded similar results to studies undertaken by Hipp et al. (2004) and Rotton and Cohn (2003). Although both studies could not discount temperature aggression theory entirely, they argued that routine activity theory was superior for explaining violent criminality. For example, both studies allude to routine activity theory being more adept at explaining the relationship between rising temperatures and violent crime, particularly in seasonal climates (Hipp et al., 2004; Rotton & Cohn, 2003).
Although this study furnishes support for routine activity theory, it is not without its limitations. The question remains as to whether the evidence presented here suffices to sufficiently discredit temperature aggression theory in explaining violent criminal behavior. Our analysis was restricted to a single city during a specific historical period. We focused on Cleveland because we needed a city with a reasonable number of daily outdoor crime incidents where the offender physically injured the victim. Otherwise, there would be insufficient variability in the dependent variable. This study should be replicated in other cities that possess a consistently hot climate, as even warmer climates show mild oscillations in violent crime rates when the temperature rises (Hipp et al., 2004). Thus, it remains to be seen how well our findings generalize to other geographical locations.
We also must acknowledge that the aggregate nature of our data precludes the identification of potential microlevel variables that might condition the relationship between maximum daily temperature and violence. For instance, men and women have minor but critical differences in their physiological responses to both heat (Kenney, 1985; Notley et al., 2017) and stress (Matud, 2004; Verma et al., 2011). Women have higher core body temperatures (Mackowiak et al., 1992; Ogoina, 2011), which is enhanced for women on birth control (Baker, Mitchell, & Driver, 2001; Baker, Waner, et al., 2001), and slower metabolic rates (Buchholz et al., 2001; Tooze et al., 2007; Westerterp & Elbers, 1999) than men that all act to make women better able to tolerate higher temperatures. A study of the neuroendocrine systems in the body also revealed that physiological stress has a higher likelihood of triggering a fight or flight response among men. In contrast, stress activates a more conciliatory response, on average, among women (Verma et al., 2011). When one considers that men are probably more susceptible to higher temperatures than women and that they also typically exhibit higher levels of aggression (Giancola et al., 2009; Lightdale & Prentice, 1994; Wickens et al., 2012), it seems reasonable to speculate that elevated temperatures will influence the aggressive behavior of men more than women because of physiological differences between the sexes.
Similarly, the relationship between heat and violence has the potential to be influenced by age because a person’s age is linked directly to his or her physical capacity to deal with stress (Colinet et al., 2015; Pandolf, 1997). As we age, the body’s thermotolerance tends to decrease (Sørensen & Loeschcke, 2002; Stapleton et al., 2015), making age a potentially salient factor in how hot temperatures may influence violent behavior. Based on this research, future studies may wish to conduct multilevel analyses to determine whether an offender’s sex, age, or other demographic characteristics condition the relationship between maximum daily temperature and violent criminal behavior. Such studies will undoubtedly advance our understanding of both temperature aggression theory and routine activity theory.
Finally, we were unable to control for other situationally based variables that may have influenced the impact of maximum daily temperature on behavior. Our analysis makes the implicit assumption that there is a minimal crossover from outdoor temperatures to indoor violence. Although not the norm, it is possible that some indoor violent crime injuries may have been the result of individuals traveling indoors because they experienced uncomfortable outdoor temperatures. These repositioning individuals might have temporarily maintained their high level of irritability indoors despite no longer being exposed to uncomfortable outdoor temperatures. In addition, despite its documented relevance in the literature (Foran & O’Leary, 2008), the effect of alcohol consumption on violence could not be assessed due to data limitations.
Although by no means conclusive, the current study has allowed us to gain further insight into the nexus between temperature and violence. The nature of the data allowed us to better distinguish between temperature aggression theory and routine activity theory. Our results cast doubt on the contention that temperature aggression theory is beneficial in explaining the positive correlation between maximum daily temperature and violence. Although we find that maximum daily temperature influences violent crime, it has little effect on the physical injury of crime victims during the commission of a crime. Routine activity theory furnishes an arguably more plausible theoretical framework for understanding why a rise in temperature amplifies violent behavior. It is noteworthy to emphasize that the current study is only preliminary. Like all research endeavors, the present study has flaws. We hope that this study serves to inspire additional empirical research on this topic in the hope of engendering a clearer understanding of the frequently reported association between temperature and violence.
Footnotes
Acknowledgements
We would like to thank the editor and reviewers for their helpful comments.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
