Sporting events and the spatial patterning of crime in South Africa: local interpretations and international implications.

Author:Breetzke, Gregory


Research in several countries has investigated the relationship between the occurrence of sporting events and crime (Braaf and Gilbert 2007; Drake and Pandey 1996; Fernquist 2000; Kirk 2008; Lin 2007; Sachs and Chu 2000). A wide variety of crime types has been considered and, overall, this body of literature has largely demonstrated that the occurrence of a sporting event is generally correlated with an increase in the number of crimes. However, the majority of prior research investigating the relationship between sporting events and crime has examined this relationship at a citywide or state level of aggregation. Furthermore, past research has examined this relationship in developed locations and has not considered whether the occurrence of sporting events could affect crime levels in the context of less developed locations.

In this study, we contribute to this literature by using a Geographic Information Systems (GIS) approach to model the sporting event/ crime relationship at much finer levels of aggregation (i.e., stadium vicinity) as well as across the city as a whole. The geographic location under investigation is the city of Tshwane (population 2.2 million; 1,248 square miles) in South Africa. The city has one major sports arena, the Loftus Versfeld stadium, which hosts both rugby union and soccer. We will examine whether crime levels change on game days compared to non-games days and will consider whether the outcome of a home game affects the relationship between sporting game days and the number and seriousness of offences at these multiple spatial levels.

Understanding the sporting events/crime linkage

One key spatial theory of crime that can be used to explain the relationship between the occurrence of sporting events and crime is the routine activities (RA) theory (Cohen and Felson 1979). RA theory is an opportunity-based theory that argues the need for three elements to co-occur before a crime can ensue: the presence of a motivated offender, the presence of a suitable target, and the absence of a capable guardian. The likelihood that these three elements will come together varies across time, due to the existence of routine activities that may affect individual behaviour. These routines may be daily (e.g., time periods within a day in which an individual is at work or in school), weekly (e.g., weekday versus weekend routines), or even yearly (e.g., annual school calendars that determine when youth do and do not attend school). The theory suggests that when individuals are not engaged in normal routine activities, the likelihood that the three key elements will converge increases, which may affect opportunities for crime and victimization.

The theory has, not unexpectedly, been revised and expanded over the years. Felson (1986) drew from Hirschi's (1969) control theory to incorporate the concept of handlers, individuals who have a social bond, or handle, that links them to an offender in such a way that it allows them to exert some control over that offender and prevent crime. Essentially, Felson argued, a web of informal crime control involves four elements: a handled offender (someone who can both commit crimes and be handled or controlled by a handler), an intimate handler (someone who is close enough emotionally to the offender to grasp this handle), a suitable target, and a capable guardian. Felson (1987) then incorporated a place element into the theory, describing the concept of a metroquilt that can interfere with the delivery of crime opportunities to offenders and suggesting that not all places are equally receptive or amenable to crime. Eck (1994) further expanded the theory to incorporate a third type of crime discourager (in addition to guardians and handlers) by developing the concept of place managers, individuals who control or monitor locations (e.g., homeowners, neighbours, private security officers, building managers, receptionists). Eck's concept presents routine activity theory as involving three objects that are necessary for a crime (a motivated or likely offender, a suitable target, and an amenable place) and three elements that directly supervise and control these objects (a handler, a guardian, and a place manager, respectively). Crime is least likely to occur when each necessary object is directly controlled by the appropriate element, thus interfering with and discouraging crime. Felson (1995) then divided each type of controller into four categories of crime discouragers, based on their level of responsibility. Personal crime discouragers have some personal responsibility (e.g., friends, family, property owner); assigned discouragers are employees who are specifically assigned to the task of crime discouragement (e.g., private security, hotel doorman); diffuse discouragers are employees who are not specifically assigned to crime discouragement (e.g., store employees, accountants); and general discouragers are unpaid individuals who lack both personal ties and occupational responsibility (e.g., strangers, bystanders, customers). This created a total of 12 distinct types of crime discouragers, based on the object of supervision and the level of responsibility (Felson 1995). For example, personal guardians are individuals with a personal responsibility toward a suitable target (e.g., a shopper watching his or her purchases), personal handlers are individuals with a personal responsibility toward a likely offender (e.g., a parent preventing a child from being truant from school), and personal managers are individuals with a personal responsibility toward an amenable place (e.g., a homeowner monitoring near his or her home).

Although routine activities tend to be fairly stable, they may be changed or disrupted by a variety of events, such as the occurrence of sporting matches. Other special and/or rare events that may affect routine activities and, by extension, crime include major holidays (see e.g., Cohn and Rotton 2003); unusual weather occurrences, such as Hurricane Katrina (see, e.g., Varano, Schafer, Cancino, Decker, and Greene 2010; Kirk 2008) and the recent earthquakes in New Zealand (see e.g., Mathewson 2011); social interactions that spark riots, such as those occurring in England in 2011 (see e.g., Birch and Allen 2012; McKee and Raine 2011); and even technological failures, such as the 1977 New York City blackout and the blackout in the north-eastern United States in 2003 (see e.g., Genevie, Kaplan, Peck, Struening, Kallos, Muhlin, and Richardson 1987; McPhee 2003).

Essentially, these types of events encourage, or even force, changes in normal routines, affecting criminal opportunities either positively or negatively by affecting the co-occurrence of the key elements needed for a crime. Within the routine activities framework, there are several ways that sporting events could affect crime levels. First, sporting events attract a large number of spectators to a stadium and to the area around it, thereby increasing the number of suitable targets within the stadium vicinity. Because many sporting events, including those held at the Loftus Versfeld stadium (capacity 51,000) in Tshwane, are often associated with tailgate parties (1) that begin several hours before the start of the game, these potential targets are at risk of victimization for a much longer period than just the duration of the game itself. At the citywide level, sporting events can bring family members and friends into close and sometimes intense proximity, either at local sports bars, taverns, and restaurants or at a family member or friend's home where people gather to watch the event on television. The concatenation of suitable targets, motivated offenders, and possibly alcohol may increase the opportunities for violence involving family members and acquaintances.

Second, a sudden influx of people into a relatively small area (i.e., the vicinity of a sports stadium) is likely to increase the number of motivated offenders. Activity nodes such as sports stadiums and shopping malls are well known to attract offenders, as the larger number of people in a confined area does, by its very nature, create an environment that increases opportunities for criminal activities (Brantingham and Brantingham 1995; Lin 2007).

Third, when fans travel to a stadium or converge at bars or taverns to watch the event, their homes are often left unoccupied, leading to an absence of capable guardians and thus increasing their vulnerability to property crimes such as burglary. Similarly, motor vehicles parked in the vicinity of the stadium or elsewhere are unguarded and more vulnerable to crimes such as motor vehicle theft or theft from vehicle. While capable guardians in the form of police and stadium security often have a strong presence in and around stadiums, they are crime discouragers whose task is to focus on crime discouragement in and around the stadium (see Felson 1995). As a result, their influence could diminish as the distance from the stadium increases.

Obviously, the presence of large numbers of spectators in and immediately around the stadium may also increase guardianship. However, Reynald (2010) suggests that there are several key factors that affect guardians' willingness and ability to disrupt possible criminal opportunities. These include the willingness to supervise, the ability to detect potential offenders, and the willingness to intervene when necessary. It may be that these factors are not always present at sporting events. First, Reynald found that individuals who were willing to supervise and intervene tended to be motivated by a sense of responsibility for their neighbours and their community, something that is less likely to occur at a sporting event than in a residential community. Second, Reynald found that the ability to detect potential offenders appears to be determined, in part, by the extent to which possible guardians are familiar with their context and thus able to identify unusual events or people. The ability of the available guardian to detect suspicious and possibly crime-related activity increases his or her capability. However, as spectators at a sporting event are in a less familiar context (the stadium) and are surrounded by other spectators, most of whom are strangers, their ability to detect this type of activity is likely to be limited. Finally, Reynald found that the decision to intervene when suspicious activity is detected is affected by a variety of factors, including the sense of responsibility the guardian feels toward the community. It is likely that spectators at a sporting event will not feel the same level of responsibility toward other spectators that they feel toward their own residential community. In addition, it appears that, in general, capable guardians are only willing to intervene directly in a crime when their own personal safety is not likely to be endangered (e.g., by yelling at an offender attempting to break into a neighbouring property). Given the heightened emotions and possibly increased alcohol consumption frequently present at sporting events, guardians may feel that direct intervention may create too much personal risk.

Therefore, the convergence of these three base components of RA theory (suitable targets, motivated offenders, and the absence of a capable guardian) increases the number of criminal opportunities and hence the likelihood of a crime occurring. Studies evaluating RA theory in this context have sought to determine how human behaviour during and after a sporting event can affect an individual's risk of victimization (see Braaf and Gilbert 2007). More recently, however, the notion of both time and space have been incorporated into research of this nature, with researchers keen to examine the spatio-temporal differences affecting the probability of convergence of these three conditions on sporting game days compared to non-game days (see Lin 2007; Kirk 2008). In the majority of studies, RA theory is used at a citywide, macro level to test the stated hypotheses and to justify the results obtained, with varying degrees of success.

Prior research examining the sporting event/crime linkage

While the body of research investigating the link between sporting events and crime is not substantial, there has been a reasonable amount of research conducted around the world. For example, in the United States, Sachs and Chu (2000) observed an increase in domestic violence dispatches by the Los Angeles Sheriff's Department during selected American football game days, while Lin (2007) found increases in burglary and vehicular theft and a moderate increase in theft from a vehicle on an American football game day compared to a non-game day in Seattle. Rees and Schnepel (2009) also found an increase in assaults, vandalism, liquor law violations, driving under the influence, and disorderly conduct on American college football game days across a large swath of the United States. Adubato (2011) found that domestic violence arrests in Philadelphia were higher during the 8-hour period following the start of televised Philadelphia Eagles football games than on non-game days. Other studies have also found disproportionately higher incidents of assault and domestic violence on Super Bowl Sunday compared to other daily averages (Burton Nelson 1994; Vazquez, Stohr, and Purkiss 2005).

In Canada, Kirk (2008) examined the effect of hockey games on crime patterns at various levels of analysis in downtown Vancouver. Kirk found that the occurrence of Vancouver Canucks games did not noticeably alter overall crime patterns in the downtown core, although the temporal patterning of crime in certain neighbourhoods fluctuated depending on the time of the game. For example, False Creek North, a residential waterfront community in the south-eastern part of the downtown Vancouver peninsula, had lower reports of vehicular theft during hockey games and a dramatic spike in vehicular thefts in the hours immediately after games. Andresen and Tong (2012) found that the 2010 Winter Olympics contributed to moderate increases in assaults related to the Games in several key neighbourhoods, particularly the central business district, although there was no evidence to support a more pervasive impact on crime throughout the city as a whole. They argued that the substantial increase in the number of security personnel and the significant amount of police activity during the Games appears to have acted to mitigate possible increases in crime and victimization.

Sports outcomes also appear to be associated with crime levels. White, Katz, and Scarborough (1992) found that violent crimes against women increased in Los Angeles when the local football team won compared with non-game days but did not change following a loss. Fernquist (2000) found that making sports playoffs was significantly related to a decline in homicide rates in 30 United States metropolitan areas, while Rees and Schnepel (2009) found evidence that home game losses led to larger increases in the number of offences than home game wins. Recently, Card and Dahl (2009) found that upset losses in American football by the home team (those in which the home team lost after being predicted to win by a pre-game point spread of greater than three points) led to an 8% increase in police reports of male-on-female intimate partner violence across six states that are home to American football teams, although the spike in violence appeared to be extremely short-lived, appearing only in a very small window of time after the game ended. It was also affected by other factors such as the importance of the game (e.g., whether the home team was still eligible for the playoffs or if they were playing a traditional rival team).

Research outside North America has also shown a similar positive association between sporting events and crime. Kurland, Johnson, and Tilley (2011) examined crime occurring around five football stadiums in various parts of England and found that the numbers of crimes occurring at and immediately around the stadiums were elevated on football match days when compared to non-match days. Sivarajasingam, Moore, and Shepherd (2005) found a significant increase in assault-related presentations to the accident and emergency departments in Cardiff, Wales, on weekends when international rugby union and soccer games were played compared with weekends when there were no games.

American football, hockey, soccer, and rugby union are not the only spectator events that have resulted in an exploration of their effects on crime. The impact of global sporting events such as championship boxing matches (Phillips 1983), the Melbourne Cup horse race (Braaf and Gilbert 2007), the America's Cup yacht race (Hall, Selwood, and McKewon 1995), the European and World Cup Soccer championships (Cooke, Allan, and Wilson, 1999; Mattick 1999; O'Donnell, Mattick, and Mehta 2003; Goodall, Trevillion, and Muncie 2006) and the Summer and Winter Olympics (Decker, Varano, and Greene 2007; Jorm, Thackway, Churches, and Hills 2003) on crime levels have also been investigated, with researchers largely demonstrating a positive association between the occurrence of the sporting event and localized crime levels.


Crime data

The crime data used in this study consists of all crimes in Tshwane recorded by the South African Police Services (SAPS) during the 5-year period from 1 September 2001 to 31 August 2006. The SAPS data included the geographic location of the crime (x, y coordinate), date and time of day, and type of crime committed. The crime data provided by the SAPS had a geocoding hit rate of just under 93%, which is above the 85% minimum reliability for a geocoding hit rate for crime analysis specified by Ratcliffe (2004). It is unknown whether the non-matched crime records were randomly distributed, but given the volume of the crime data at our disposal, it is unlikely that the inclusion of this "missing" data would have significantly altered the results generated. The crime data were aggregated by day to be linked with daily weather data obtained from the South African Weather Service. A total of 882,524 crime incidents were recorded in Tshwane over this 5-year period, including 152,133 assaults, 57,126 cases of drunk and disorderly behaviour, and 97,864 burglaries.

A total of 139 home games (rugby union, n = 61; and soccer, n = 78) took place during this 5-year period at the Loftus Versfeld stadium. Of these, 82 games were won by the home team based in Tshwane, 33 were lost (i.e., won by the away/rival team), and 24 games ended in a draw. Tables 1 and 2 provide breakdowns of the occurrence of home sports game days by season and by result. In both instances, the home games days are also categorized by sporting code: soccer and rugby union. Soccer is a summer sport in South Africa, and as a result, there are more soccer games over the festive summer period, while rugby is a traditional winter sport. Despite this, however, there is still considerable overlap in occurrence between these two codes throughout the year. While rugby has a high win percentage of 70% (Table 2), soccer has a more modest winning percentage of 50%, although draws (which may not necessarily be a "bad" result) are much more common in soccer than rugby union. The average attendance for home sports games over this time period was 27,584 (7,033 on average for soccer and 20,551 on average for rugby union). Dummy variable (1, 0) coding was used to assess the differences between sports games days and other days as well as the differences between home sports game days won and other days and home sports game days lost and other days.

Control variables

Several control variables were included in the analysis in order to address the possibly confounding effects of seasonal trends and weather conditions on crime. Past studies have overwhelmingly demonstrated that crimes have temporal trends (see e.g., Cohn and Rotton 2000, 2003), while meteorological conditions such as temperature, relative humidity, and wind speed have also been shown to affect crime patterns (see e.g., Cohn 1993; Brunsdon, Corcoran, Higgs, and Ware 2009). These factors could potentially distort the relationship between sporting events and crime. For example, because most sporting events in Tshwane take place on a Saturday, an increase in crime on a game day could reflect the regular increase in crime on weekends, rather than an increase due to the sporting event.

Two sets of control variables were included to address the effect of weather and temporal trends on crime. The first set controlled for various meteorological parameters including precipitation, ambient temperature, relative humidity, wind speed, and air pressure. Daily averages were calculated for each variable and linked with the daily crime counts. Preliminary analyses identified several non-linear relationships occurring between the meteorological control variables. This evidence of non-linearity was not surprising because the relationships between meteorological parameters and crime can frequently be described by a quadratic or cubic equation (Cohn 1993). Following Cohen and Cohen's (1983) recommendations, these non-linear relationships were dealt with, where possible, by transforming the variables, based on the general shape of the underlying data. For example, assaults were found to be a curvilinear function of temperature; to control for this quadratic trend, the temperature scores were first converted into deviation scores and then squared. Lastly, a correlation matrix was employed to screen for associations between the meteorological variables themselves. The results revealed a low degree of intercorrelation between the meteorological variables, with the highest correlation occurring between precipitation and relative humidity (r = 0.455, p

The final data set contained 1,826 records with each record corresponding to a day (from 1 September 2001 to 31 August 2006). Information about each day was included in each record, including the number of crimes (overall crime, assault, drunk and disorderly, and burglary) committed that day; the average temperature, humidity, and wind speed that day; and dummies for month of the year, the day of the week, and whether that particular day was a pay day and/or public holiday. Finally, a sequence variable was assigned to each record, starting with one and ending with 1,826 to control for drift (Gottman 1981). The final list of variables used in the study is provided in Table 3.

Analytic strategy

Hierarchical multiple regression was employed to investigate the relationship between the occurrence of sporting games and crime across Tshwane as a whole and across the three buffered distances surrounding the Loftus Versfeld stadium. In the analysis, the four crime variables (i.e., overall crime, assault, drunk and disorderly, and burglary) acted as the dependent variables, and the weather and temporal variables were the independent controls.

A total of 48 multiple linear regression analyses were run, corresponding to 12 regressions at each level of aggregation (citywide, 1/2 mile, 1 mile, 2 miles). (4) For example, at the citywide level of aggregation, regression results were generated for all crime, assault, drunk and disorderly, and burglary for sports game days, for sports game days won, and for sports game days lost. For each of the 48 regressions, the weather and temporal controls were input first to create a baseline model. Next, the dummy variable for either sports game day or for sports game day won or sports game day lost was added to assess the independent effect that sports days had on crime at each level of aggregation. A map indicating the location of the Loftus Versfeld stadium is provided in Figure I and the results of the descriptive and regression analysis are summarized in Tables 4 and 5a-5d.


All crime

Table 4 provides descriptive data for crimes on game days versus non-game days for all crime and for each crime type separately at various distances from the stadium. The results indicate that overall crime increases throughout Tshwane on sports game days compared to non-game days (M = 500.60 versus M = 466.02). Approximately 34 more crimes are committed on sports game days than on non-game days. A one-way ANOVA found this difference to be statistically significant F(1, 1825) = 18.85, p


Game days attained significance in subsequent analysis at several spatial levels after controlling for weather variables, temporal controls, and autocorrelation (see Table 5a). From the coefficients in Table 5a it can be seen that crime is significantly related to game days at the half-mile and 1-mile buffer around the stadium, but not at the 2-mile buffer or citywide. This finding is particularly interesting as it provides the first indication that there may be a critical threshold distance from the stadium beyond which the occurrence of the sporting event may be unrelated to localized crime patterns. This result is supported by Kurland, Johnson, and Tilley (2010), who found that the number of crimes around Wembley Soccer stadium in London diminished as distance increased from the ground, with little difference if any beyond 750 metres from it. The figure provided in parenthesis in Tables 5a-d represents the change in magnitude in crime levels on sports game days. It is also interesting to note that the result of the game does appear to have some effect on crime levels: game days on which the home team won attained significance at all three buffer distances from the stadium, while game days on which the home team lost were only significant at the 1-mile buffer.


Sports game days attained significance in the analysis of assault only within a half-mile of the stadium F(24, 1801) = 4.15, p

Drunk and disorderly

Similar to the overall crime category, a significant main effect was obtained for sports game days at the half-mile and 1mile buffer around the Loftus Versfeld stadium, but not at the 2-mile buffer or citywide. The descriptive data presented in Table 4 indicate that drunk and disorderly behaviour is always higher on sports game days than non-game days, regardless of the distance from the stadium or the outcome of the game. Again, the result of the game does appear to have some effect on the magnitude of drunk and disorderly behaviour within the stadium vicinity with sports game days won, attaining significance at the half-mile and 1-mile buffer distances from the stadium. Similar to most other categories of crime, games lost did not attain significance in any model (see Table 5c).


Surprisingly, no significant effect was obtained for burglary on game days, nor were any significant effects found when controlling for temporal factors, weather, distance from the stadium, or the outcome of the game (see Table 5d). These findings may be partly explained by examining the descriptive data in Table 4, that show that mean rates of burglary on sports game days do not differ considerably from non-game days, and in some instances, crime rates are actually lower on game days than non-game days, specifically at the half-mile buffer (M = 0.16 versus 0.17) and citywide (M = 53.24 versus 53.95). However, this finding is interesting because RA theory would predict that when a residence is unoccupied (e.g., when the residents are away from home attending a sporting event), the lack of capable guardians increases the risk of burglary.

One possible explanation for this result is that burglary is a distinctly different type of crime than the others examined in this study. Both assault and drunk and disorderly behaviour are impulsive acts, committed on the "spur of the moment" and are not generally premeditated. These crimes are likely to be committed by fans who are attending the sporting event, who are watching it on television, or who are emotionally affected by learning the outcome of the game. Disagreements over disputed calls, arguments over particularly exciting plays, and the emotional involvement in the outcome, particularly when combined with alcohol, may make these sorts of behaviours more likely. Essentially, the sporting event itself, and the activities of engaged fans, may increase both offender motivation and target suitability. Burglary, on the other hand, is a property offence that is more likely to be premeditated than to be committed impulsively, with no prior intent or planning.

Recent studies of burglary (particularly, relating to target selection) have found that burglars employ a rational choice process of target appraisal and selection rather than impulsively committing crimes in response to immediate situational opportunities (see e.g., Bernasco and Nieuwbeerta 2005; Wright and Decker 1994; Wright, Logie, and Decker 1995). As Wright and Decker (1994) have pointed out, even individuals who engage in residential burglary spend the majority of their time engaged in other activities; burglars generally are not continually motivated to offend. Instead, the decision to commit a residential burglary is generally determined by immediate financial needs rather than long-range planning. Therefore, burglars who are also sporting fans may actually be less likely to consider engaging in burglary on game days because, rather than focusing on current financial needs, they are more likely to be involved in attending a game, watching it, or discussing it with other fans. Essentially, fan involvement in and engagement in the game reduces the likelihood of involvement in non-impulsive crime.

In addition, the effectiveness of other fans as capable guardians may vary by type of offence. Fellow fans may be effective as capable guardians to prevent burglary because burglars are more likely to make a rational choice to commit burglary and may be deterred by the presence of others. On the other hand, assault and drunk and disorderly behaviours often are impulsive, due more to immediate situational characteristics (including alcohol consumption and the events of the sporting event) than to a rational decision-making process; the offender's fellow fans may be equally affected by the situation and be unable to act as effective capable guardians to prevent the offender's involvement in these crimes or may be engaged in similar behaviours as well.


Much of the international literature investigating the influence of sporting events on crime supports the theory that this association is positive (see Braaf and Gilbert 2007; Drake and Pandey 1996; Fernquist 2000). The results of this study corroborate the prior research at finer spatial extents, but at a citywide level of aggregation the results provide less support for this view. The study found a localized effect in the area around the stadium; for some types of crime, this effect extended this up to a 1-mile radius around the stadium. However, the study did not find support for the macro level effect suggested by previous, less spatially refined, analyses of sporting events and crime; there was no evidence that sporting events significantly influence the number of criminal offences of any kind citywide.

However, this finding does not necessarily contradict RA theory. While RA theory does suggest that the impact of game days on individual routine activities would affect the expected convergence of the three key components of the theory (suitable targets, motivated offenders, and lack of capable guardians), this effect would not necessarily occur citywide but would be more likely to be concentrated primarily in and around the stadium.

Possible explanations for why sporting events do not significantly impact crime levels across the city of Tshwane as a whole requires an understanding of the underlying socio-demographics of sports fans in South Africa. Despite many attempts to utilize sport as a tool for reconciliation among the various racial groups in the country, sport in the post-apartheid era in South Africa is still racially polarized. Although more Blacks (5) have begun to participate in rugby union in recent years, the sport is still dominated by Whites and the majority of administrators, players, and spectators are White. Within this context, it is highly unlikely that many members of the Black population in Tshwane, who comprise over 50% of the city's residential population, would attend rugby union games in person. In addition, almost 40% of Black households in Tshwane don't own a television set (Statistics South Africa 2003), and even those that do may be unable to watch rugby union games, as they are broadcast in South Africa exclusively on satellite television which requires a monthly subscription fee that is greater than the total monthly income of approximately 10% of Black households in the city (Statistics South Africa 2003). Thus, many Blacks are excluded from watching rugby union games on television. Similarly, soccer is almost exclusively a Black sport in South Africa and very few White residents in Tshwane are likely to attend local soccer games. In addition, even though local soccer games are broadcast on free-to-air local television stations, it does not appear likely that many Whites watch these games in their homes or at local bars. As a result of this racial divide, the number of people watching sporting events (either rugby union or soccer) is essentially limited and, therefore, so is the convergence of people throughout Tshwane as a whole. In accordance with RA theory, this would reduce the likelihood of crime being affected by sporting events at a citywide level.

More significant results were obtained, however, when the unit of analysis was narrowed to examine a series of smaller buffer distances immediately surrounding the Loftus Versfeld stadium. Within both a half-mile and 1-mile radius of the stadium, overall crime, assault, and drunk and disorderly behaviour increased significantly on game days compared to non-game days. These results suggest that being physically near the stadium and large gatherings of sports fans increases an individual's risk of victimization. Essentially, passionate fans in close proximity to each other and to the stadium are more likely to engage in aggressive behaviour than fans watching the same event at home or at an entertainment establishment some distance away from the stadium. One reason for this finding could be that fans attending a sports game identify more closely with the team and are possibly more passionate about the sport, the team, and the outcome of the game than non-attending fans. This passion can spill over into aggressive behaviour, not only during the game itself, but also both before and after the game. On the day of the sporting event, the anticipation of physically attending the match, as opposed to simply watching it on television, may generate greater anticipation and excitement. The tailgating parties that take place outside the stadium on game days frequently begin hours before the start of the game, contributing to the fans' excitement and involvement by encouraging active participation in a variety of activities designed to engage fans in the sporting event and increase their support for the home team. These activities also increase social interaction among groups of passionate attendees. Similarly, after the game, the proximity to other excited or disappointed fans may increase the likelihood of aggressive behaviour under certain circumstances.

A second reason for these results could be that sports fans attending the game could be more primed to engage in aggressive behaviour than non-attending fans because they see the action "up close and personal." Spectators present in the stadium are an integral part of the sporting event and regardless of the result cannot simply turn the television set off and walk away. Announcers at the games often encourage spectators to cheer, to stand up and wave, and to participate in other activities that actively engage the spectators in the game, both physically and emotionally. In fact, Clark (2012) reports that, to increase American football crowd attendance (which has been in decline since 2005), the National Football League has recently "liberalized" its restraint on crowd noise by, among others, encouraging announcers to "incite a racket" when the opposing offence faces a crucial third down. The NFL also plans to allow fans in the stadium to view the same replays of calls under review as the game official sees them under the hood on the side-line, making fans feel even more part of the game. The concept of a priming effect was initially proposed by Berkowitz (1984) as a way of explaining how media violence may produce violent behaviour; he argued that the aggressive ideas suggested by a violent movie can prime other semantically related thoughts, heightening the probability that viewers will have other aggressive ideas shortly thereafter. Berkowitz focused on how violent media affect behaviour, and the fact that sporting events are not, generally, in themselves violent may explain why there does not appear to be an increase in crime and violence among fans who are watching a game on television. However, it may be that the presence of others in the stadium may have a priming effect and lower the activation threshold for aggression, while priming is less likely to occur among fans that are not actually present at the event. Regarding game outcome, the results of the study indicate that winning has a greater impact on crime levels in the immediate vicinity of the stadium than does losing. These results are consistent with some international evidence (White et al. 1992; Sivarajasingam et al. 2005) but contradict those of more recent studies (see Rees and Schnepel 2009). One possible explanation for this finding, that parallels interpretations made by White et al. (1992), is that levels of confidence, assertiveness, or patriotism are increased as a result of a win and, when this is combined with increased alcohol consumption, may lead to crimes such as assault and drunk and disorderly behaviour. Research has shown that winning is associated with greater alcohol consumption than losing (Neal, Sugarman, Hustard, Caska, and Carey 2005). In addition, Moore, Shepherd, Eden, and Sivarajasingam (2007) found that the success of a home team may increase aggression that affects alcohol consumption by spectators after the game. The results of this study suggest that, within the vicinity of the stadium, the euphoria of winning, possibly combined with alcohol consumption, could lead to increased interpersonal crime.

Finally, the study's limitations need to be considered. First, the underlying socio-demographic profile of the population (and neighbourhoods) within each spatial level of aggregation were not taken into account. For example, the population distribution underlying the three buffer intervals could vary, and this could have an impact on recorded crime. Although Breetzke and Horn (2009) have noted the relatively homogenous nature of the population (and neighbourhoods) in the area immediately surrounding the Loftus Versfeld stadium, it is likely that, for example, the population distributions within each buffer interval will vary significantly over the course of a sports day, due either to the movements of existing residents within these areas or to the influx of spectators. These movements could, in turn, influence the amount and distribution of the populations at risk in these areas and hence affect crime levels. The number of assumptions required, however, to estimate the socio-demographic characteristics of underlying populations at risk within each spatial level of aggregation throughout the course of a day was deemed too vast to make any claims with any degree of certainty. A second limitation relates to the fact that grouping both rugby and soccer together leaves it unclear which sport may be contributing to the observed increase in crime at various spatial levels. As noted in this research, sport is racialized in South Africa and the socio-demographic profile of sports fans of both codes could differentially influence crime levels. While we readily acknowledge this limitation, the main aim of this research was to examine the effects that home sports games have on crime levels at various spatial levels of aggregation in a city. We were also interested in examining the sporting event/crime linkage in the context of a developing state. The individual motivations, attitudes, and behaviours (criminal or otherwise) of various sub-sets of fans of different sporting codes were of less interest to us. Besides, by dividing sports games by codes, you reduce the magnitude of our variable of interest (i.e., home sports days), and in doing so, you lose statistical power in the regression models. Future research will examine the effect that both sporting codes have on crime levels over a longer period. A third limitation relates to the possibility that post-game levels of aggression are not affected by the end result (winning or losing) but rather by the actual outcome of the game relative to the expected outcome. In other words, if the home team was expected to win, but instead loses, the frustration among the home team fans would be greater than that felt if the team was expected to lose anyway. Future examination of the outcome of a game relative to the expected outcome and its effect of fans' behaviour both within the stadium vicinity and citywide is warranted. Fourth, it could be that more crime was recorded in the areas immediately surrounding the Loftus Versfeld stadium on game days simply because there was a greater police presence there than on non-game days and also a greater police presence on game days at and near the stadium compared to other areas of the city. Paradoxically, however, an increased police presence in areas immediately surrounding a sports stadium could serve as a deterrent and also limit the number of aggressive crimes that might otherwise have taken place. Finally, the study is limited to one sports stadium in one city in South Africa; whether the results are generalizable to the entire country is questionable. It may be more appropriate to view this research as a case study that provides empirical evidence to suggest the need for spatial analyses of the impact of stadiums and sporting events on crime in additional settings.

International policy implications

Sport is an emotional experience for both players and spectators. While spectators are not actively participating in the sporting event itself, they can often be emotionally attached to the players, the team, the sponsors, and even to the stadium where the sporting event is being held (see van Houtum and van Dam 2002). Watching a sporting event (live or televised) can generate a range of emotions, including aggression-inducing frustration; particularly if the individual's favoured team is playing poorly or losing. Research has shown that fans of losing teams are more likely to report negative affective states, increased aggression, and psychological distress (Wann 1993; Wann, Melnick, Russell, and Pease 2001; Banyard and Shevlin 2001). Under certain environmental cues or stimuli this aggression can lead to impulsive acts of crime (as shown in this study) or more widespread and serious acts of public violence. Perhaps the most recent example of the latter outcome is the 2011 Vancouver Stanley Cup Riot that broke out in downtown Vancouver on 15 June 2011. The riot, which led to over 140 people injured, one critically, was borne out of fans' frustration at the outcome of game seven of the Stanley Cup finals, which the Vancouver Canucks lost. Although the game was actually held in Boston, over 155,000 people are thought to have gathered in downtown Vancouver to watch the game on two big-screen televisions erected in the downtown area. By the end of the riot, over 100 people had been arrested and losses due to vandalism, theft, and damage to property are thought to have exceeded $4 million (Bailey 2011).

Increased understanding of the sporting event/crime relationship may help to inform social policies designed to improve security and reduce crime and violence at major sporting events such as the Stanley Cup. An in-depth, detailed discussion of the policy challenges created by major sporting events was provided by Decker, Greene, Webb, Rojek, McDevitt, Bynum, Varano, and Manning (2005), in their review of safety and security issues at the 2002 Winter Olympic Games in Salt Lake City, Utah. Admittedly, the Olympic Games occur on a much larger scale than the average sporting event, lasting for much longer periods and drawing exponentially larger crowds, while at the same time being much more short-lived. However, some of the lessons learned as a result of the Olympic Games may be applied to policy development for less exceptional events. Decker et al. (2005) particularly stress the importance of security centralization as a way of improving both communication and decision-making processes. While, during the Olympics, this related to the centralization of a temporary organization, the need to ensure that different groups and organizations can and do work together cohesively is equally important when attempting to improve safety and security within a city during any sporting event. Increasing communication and teamwork among county, city, and local law enforcement as well as private stadium security is essential. In addition, Decker et al. (2005) emphasized the need to treat the sporting event and the associated security with an integrated, theatre-wide approach. In the Olympics, this meant viewing the Games as a whole, rather than considering the need for security on a venue-by-venue basis. For a local sporting event, this might involve recognizing how the impact of the game may extend well beyond the confines of the sporting arena; tailgate activities that occur off stadium property, large "game-watching parties" held at sports bars and restaurants, and other game-related activities also have the potential to lead to violence and need to be considered in the development of policy and the coordination of security activity.

So, what are the implications of this research for law enforcement agencies tasked with policing cities during major (or minor) sporting events? In tactical terms, the intelligence garnered from knowing where certain types of crime are more likely to occur during (and after) a sporting event can be used to guide operational policing units to specific locations and individuals linked to criminal activities, potentially leading to the arrest of suspects and suspicious persons. In the context of this study, knowing that assaults and drunk and disorderly behaviour increase significantly in the areas immediately surrounding the sports stadium on game days can result in specialized units being deployed in these areas. In operational terms, knowing the impact of sporting events on the spatial patterning of crime in a city will inform several operational decisions by law enforcement agencies. For example, the routes for vehicle and foot patrols could be better delineated; the locations of roadblocks could be better identified; the locations of checkpoints and stop-and-search operations could also be better targeted. Because increased alcohol consumption, a common occurrence while watching sporting events, can also intensify frustration and increase the likelihood of aggression (Ito, Miller, and Pollock 1996; Pedersen, Aviles, Ito, Miller, and Pollock 2002), these checkpoints and stop-and-search operations should apply not only to weapons but to drugs, and especially alcohol. Indeed, one of the causes of the 2011 Vancouver Stanley Cup Riot was the ability of fans to enter the downtown fan zone without being checked for alcohol, despite several checkpoints being set up in the fan zone vicinity (Vancouver Police Department 2011). Essentially, the sports/crime linkage can help to inform the development of best practices crowd management techniques and assist police departments in advance planning when policing crowds at sporting events (Police Executive Research Forum 2006, 2011). From a strategic perspective, governments and other role players that are responsible for specifically addressing long-term solutions to crime can utilize the understanding of the sporting event/crime linkage to highlight various underlying factors affecting crime and measure ways to address them. This will most likely involve an examination of the growing young "hooligan" demographic that has been shown to be responsible for the majority of crime and violent behaviour during and after sporting events (see Lewis 2007). Lastly, from an economic perspective, knowledge of the connection between sporting events and criminal behaviour may assist the police in their efforts to limit injuries, deaths, and property damage caused by resulting from sporting-event-related crime. The use of effective crowd management techniques may significantly reduce the likelihood of riots and other criminal acts, thus reducing the cost to society as a whole (see Russell 2004).

This research contributes to the growing body of literature investigating the spatial relationship between the occurrence of sporting events and crime. In particular, we considered whether crime levels were affected by sporting events at varying spatial extents throughout a city. Most previous studies have been limited to examining the sporting event/crime relationship at a citywide spatial extent in the context of a more developed state. In this study, we found that that there is a positive association between the occurrence of sporting events and crime levels in contexts outside of North America and Europe. This positive association is, however, spatially variable, and it is possible that various ecological characteristics of neighbourhoods, such as deprivation and/or ethnicity, could influence this variability. We also employed a well-established spatial theory to explain the spatial patterning of crime in South Africa, with some measure of success. Finally, the results of this research can reliably inform the tactical, operational, and strategic processes that are put in place when law enforcement agencies design and implement regional policing plans for major sporting events. With such sporting events increasing in frequency and magnitude worldwide, any knowledge gained from local studies investigating the sporting event/crime linkage is relevant and applicable to law enforcement agencies in any country tasked with policing such events.

doi: 10.3138/cjccj.2012.E20


Adubato, Beth 2011 Fanning the flames: Televised, professional football games and domestic violence. PhD diss., Rutgers, State University of New Jersey, Newark, NJ.

Andresen, Martin and William Tong 2012 The impact of the 2010 Winter Olympic Games on crime in Vancouver. Canadian Journal of Criminology and Criminal Justice 54 (3): 333-61.

Bailey, Ian 2011, 20 June Bill for damages to be much higher than for the 1994 Vancouver riot. Globe and Mail. 1994-vancouver-riot/article584247/.

Banyard, Philip and Mark Shevlin 2001 Responses of football fans to relegation of their team from the English Premier League: PTS? Irish Journal of Psychological Medicine 18: 66-67.

Berkowitz, Leonard 1984 Some effects of thoughts on anti- and prosocial influences of media events: A cognitive-neoassociation analysis. Psychological Bulletin 95 (3): 410-27.

Bernasco, Wim and Paul Nieuwbeerta 2005 How do residential burglars select target areas? A new approach to the analysis of criminal location choice. British Journal of Criminology 45 (3): 296-315.

Birch, Sarah and Nicholas Allen 2012 'There will be burning and a-looting tonight': The social and political correlates of law-breaking. Political Quarterly 83 (1): 33-43. http://

Braaf, Rochelle and Robyn Gilbert 2007 Domestic violence incident peaks: Seasonal factors, calendar events and sporting matches. Stakeholder paper no. 2. New South Wales, Australia: Australian Domestic and Family Violence Clearinghouse.

Brantingham, Patricia and Paul Brantingham

1995 Criminality of place: Crime generators and crime attractors. European Journal on Criminal Policy and Research 3 (3): 5-26.

Breetzke, Gregory Dennis and Andre Carl Horn 2009 A geodemographic profiler for high offender propensity areas in the city of Tshwane, South Africa. Environment and Planning, A 41 (1): 112-27.

Burton Nelson, Mariah 1994 The Stronger Women Get, the More Men Love Football. London: Women's.

Brunsdon, Chris, Jonathan Corcoran, Gary Higgs, and Andrew Ware 2009 The influence of weather on local geographical patterns of police calls for service. Environment and Planning, B: Planning and Design 36 (5): 906-26.

Card, David and Gordon Dahl 2009 Family violence and football: The effect of unexpected emotional cues on violent behavior. Working paper no. 15497, National Bureau of Economic Research, Cambridge, MA.

Clark, Kevin 2012, 30 June Game changer: NFL scrambles to fill seats. Wall Street Journal: A1.

Cohen, Jacob and Patricia Cohen 1983 Applied Multiple Regression/Correlation Analysis For The Behavioural Sciences. 2nd ed. Hillsdale, NJ: Erlbaum.

Cohen, Lawrence E. and Marcus Felson 1979 Social change and crime rate trends: A routine activity approach. American Sociological Review 44 (4): 588-608. 10.2307/2094589.

Cohn, Ellen G 1993 The prediction of police calls for service: The influence of weather and temporal variables on rape and domestic violence. Journal of Environmental Psychology 13 (1): 71-83. S0272-4944(05)80216-6.

Cohn, Ellen G. and James Rotton 2000 Weather, seasonal trends and property crimes in Minneapolis, 1987-1988. A moderator-variable time-series analysis of routine activities. Journal of Environmental Psychology 20 (3): 257-72.

Cohn, Ellen G. and James Rotton 2003 Even criminals take a holiday: Instrumental and expressive crimes on major and minor holidays. Journal of Criminal Justice 31 (4): 351-60.

Cooke, Matthew W., Teresa F. Allan, and Sue Jo Anne Wilson 1999 A major sporting event does not necessarily mean an increased workload for accident and emergency departments. British Journal of Sports Medicine 33 (5): 333-35.

Decker, Scott H., Jack R. Greene, Vince Webb, Jeff Rojek, Jack McDevitt, Tim Bynum, Sean Varano, and Peter K. Manning 2005 Safety and security at special events: The case of the Salt Lake City Olympic Games. Security Journal 28 (4): 65-74. 10.1057/

Decker, Scott H., Sean P. Varano, and Jack R. Greene 2007 Routine crime in exceptional times: The impact of the 2002 Winter Olympics on citizen demand for police services. Journal of Criminal Justice 35 (1): 89-101.

Drake, Brett and Shanta Pandey 1996 Do child abuse rates increase on those days on which professional sporting events are held? Journal of Family Violence 11 (3): 205-18.

Eck, John E. 1994 Drug markets and drug places: A case-control study of the spatial structure of illicit drug dealing. PhD diss., University of Maryland, College Park, MD.

Felson, Marcus 1986 Routine activities, social controls, rational decisions, and criminal outcomes. In The Reasoning Criminal, ed. Derek Cornish and Ronald V.G. Clarke, 119-28. New York: Springer-Verlag. 10.1007/978-14613-8625-4_8.

Felson, Marcus 1987 Routine activities and crime prevention in the developing metropolis. Criminology 25 (4): 911-31.

Felson, Marcus 1995 Those who discourage crime. In Crime and Place, ed. John E. Eck and David Weisburd, 53-66. Washington, DC: Police Executive Research Forum.

Fernquist, Robert M. 2000 An aggregate analysis of professional sports, suicide, and homicide rates: 30 U.S. Metropolitan Areas, 1971-1990. Aggression and Violent Behavior 5 (4): 329-41.

Genevie, Louis, Seymour R. Kaplan, Harris Peck Elmer L. Struening, June E. Kallos, Gregory L. Muhlin, and Arthur Richardson 1987 Predictors of looting in selected neighborhoods of New York City during the blackout of 1977. Sociology and Social Research 71: 228-31.

Goodall, Martin, Pam Trevillion, and James Muncie 2006 Lessons learned from the domestic violence enforcement campaigns.

Gottman, John M. 1981 Time-Series Analysis: A Comprehensive Introduction For Social Scientists. New York: Cambridge University Press.

Hall, Colin M., John Selwood, and Elaine McKewon 1995 Hedonists, ladies, and larrikins: Crime, prostitution and the 1987 America's Cup. Visions in Leisure and Business 14: 28-51.

Hirschi, Travis 1969 Causes of delinquency. Berkeley, CA: University of California Press.

Houtum, Henk van and Frank van Dam 2002 Topophilia or Topoporno? Patriotic place attachment in international football derbies. International Social Science Review 3: 231-48.

Ito, Tiffany A., Norman N. Miller, and Vicki E. Pollock 1996 Alcohol and aggression: A meta-analysis on the moderating effects of inhibitory cues, triggering events, and self-focused attention. Psychological Bulletin 120 (1): 60-82.

Jorm, Louisa R., Sarah V. Thackway, Tim R. Churches, and Marcia W. Hills 2003 Watching the games: Public health surveillance for the Sydney 2000 Olympic Games. Journal of Epidemiology and Community Health 57 (2): 102-8.

Kirk, Michael G. 2008 The spatio-temporal effects of spectator events on crime. Master's thesis, Simon Fraser University, Burnaby, BC.

Kurland, Justin, Shane D. Johnson, and Nick Tilley 2010 Late weekend kick-offs at premier football matches and their possible link to increased levels of alcohol-related ASB and disorder. University College, London. box/UCLfootballreportforGOL.pdf.

Kurland, Justin, Shane D. Johnson, and Nick Tilley 2011 The 'Footprint' project: Research brief. University College, London.

Lewis, Jerry M. 2007 Sports Fan Violence in North America. Lanham, MD: Rowman.

Lin, Chien-Min 2007 The impact of football games on crime: A routine activity approach. Master's thesis, University of Maryland, College Park, MD.

Mathewson, Nicole 2011 Sex crimes increase, with quakes blamed. Press. http://www.stuff. thquake-2011/6682161/Sexcrimes-increase-with-quakes-blamed.

Mattick, Anthony P. 1999 The Football World Cup 1998: An analysis of related attendances to an accident and emergency department. Scottish Medical Journal 44 (3): 75-76.

McKee, Martin and Rosalind Raine 2011 Riots on the streets. BMJ (Clinical Research Ed.) Case report no. 343: d5248.

McPhee, Martin 2003 Wasn't so calm after all: Shootings, break-ins rose in the city during the blackout. New York Daily News. http://www.nydailynews. com/archives/news/calm-shootings-break-ins-rose-city-blackout-article-1.523078.

Moore, Simon C., Jonathan P. Shepherd, Sam Eden, and Vaseekaran Sivarajasingam 2007 The effect of rugby match outcome on spectator aggression and intention to drink alcohol. Criminal Behaviour and Mental Health 17 (2): 118-27.

Neal, Dan J., Dawn E. Sugarman, John T.P. Hustard, Catherine M. Caska, and Kate B. Carey 2005 It's all fun and games ... or is it? Collegiate sporting events and celebratory drinking. Journal of Studies on Alcohol 66: 291-94.

O'Donnell, John J., Anthony P. Mattick, and Raj Mehta 2003 The Football World Cup 2002: Analysis of related attendances to an Irish emergency department. Irish Medical Journal 96: 90-91.

Pedersen, William C., Fredy E. Aviles, Tiffany A. Ito, Norman N. Miller, and Vicki E. Pollock

2002 Psychological experimentation on alcohol-induced human aggression. Aggression and Violent Behavior 7 (3): 293-312. 10.1016/S1359-1789(01)00044-1.

Phillips, David P. 1983 The impact of mass media violence on U.S. homicides. American Sociological Review 48 (4): 560-68.

Police Executive Research Forum 2006 Police Management of Mass Demonstrations: Identifying Issues and Successful Approaches. Washington, DC: Police Executive Research Forum.

Police Executive Research Forum

2011 Managing Major Events: Best Practices from the Field. Washington, DC: Police Executive Research Forum.

Ratcliffe, Jerry H. 2004 Geocoding crime and a first estimate of a minimum acceptable hit rate. International Journal of Geographical Information Science 18 (1): 61-72.

Rees, Daniel I. and Kevin T. Schnepel 2009 College football games and crime. Journal of Sports Economics 10 (1): 68-87.

Reynald, Danielle M. 2010 Guardians on guardianship: Factors affecting the willingness to supervise, the ability to detect potential offenders, and the willingness to intervene. Journal of Research in Crime and Delinquency 47 (3): 358-90.

Russell, Gordon W.

2004 Sports riots: A social-psychological review. Aggression and Violent Behavior 9 (4): 353-78.

Sachs, Carolyn J. and Lawrence D. Chu 2000 The association between professional football games and domestic violence in Los Angeles County. Journal of Interpersonal Violence 15 (11): 1192-1201.

Sivarajasingam, Vaseekaran, Simon C. Moore, and Jonathan P. Shepherd 2005 Winning, losing and violence. Injury Prevention 11 (2): 69-70. http://

Statistics South Africa 2003 Statistics South Africa: Census 2001. Pretoria, South Africa: Statistics South Africa.

Vancouver Police Department 2011 Stanley Cup Riot Review. vpd-riot-review.pdf.

Varano, Sean P., Joseph A. Schafer, Jeffrey M. Cancino, Scott A. Decker, and Jack R. Greene 2010 A tale of three cities: Crime and displacement after Hurricane Katrina. Journal of Criminal Justice 38 (1): 42-50. j.jcrimjus.2009.11.006.

Vazquez, Salvador, Mary Stohr, and Marcus Purkiss 2005 Intimate partner violence incidence and characteristics: Idaho NIBRS 1995 to 2001 Data. Criminal Justice Policy Review 16 (1): 99-114.

Wann, Daniel L. 1993 Aggression among highly identified spectators as a function of their need to maintain positive social identity. Journal of Sport and Social Issues 17 (2): 134-43.

Wann, Daniel L., Merrill J. Melnick, Gordon W. Russell, and Dale G. Pease 2001 Sports fans: The psychology and social impact of spectators. New York, NY: Routledge.

White, Garland F., Janet Katz, and Kathryn E. Scarborough 1992 The impact of professional football games upon violent assaults on women. Violence and Victims 7: 157-71.

Wright, Richard and Scott Decker 1994 Burglars on the Job: Street Life and Residential Break-ins. Boston, MA: Northeastern University Press.

Wright, Richard, Robert H. Logie, and Scott H. Decker 1995 Criminal expertise and offender decision making: An experimental study of the target selection process in residential burglary. Journal of Research in Crime and Delinquency 32 (1): 39-53. 10.1177/0022427895032001002.


(1) Tailgate parties generally take place in the immediate vicinity of a sporting arena, generally the stadium parking lot, and usually begin several hours before the start of a game. These parties are so named because they are often held out of the open "tailgate" or "liftgate" at the back of a station wagon, sports utility vehicle, or similar vehicle, although such vehicles are not required for a tailgate party. Tailgate parties generally involve barbequing and alcoholic beverages, particularly beer, as well as considerable outpourings of fan support of the home team.

(2) While precipitation may be an important modifier of human behaviour, Tshwane has a relatively low annual rainfall (5-year average = 533.08 mm/yr). Further analysis indicated that average rainfall levels were lower on sports game days (M = 1.19; SD = 4.94) compared to other days (M = 1.49; SD = 5.15).

(3) Dates of public holidays are 1 January (New Year's Day), 21 March (Human Rights Day), 24 March (Family Day), 27 April (Freedom Day), 1 May (Workers' Day), 16 June (Youth Day), 9 August (National Women's Day), 24 September (Heritage Day), 16 December (Day of Reconciliation), 25 December (Christmas Day), 26 December (Day of Goodwill).

(4) From this point onwards in the manuscript, we refer to the area from the stadium to the half-mile buffer distance as the half-mile buffer; the area from the half-mile buffer distance to the 1-mile buffer distance as the 1mile buffer, and the area from the 1-mile buffer distance to the 2-mile buffer distance as the 2-mile buffer.

(5) The South African population is still officially classified into racial groups. Black Africans represents the descendants of west and central African populations, the Indian population group represents the descendants of south Asian populations, and the Colored group includes the descendants of the indigenous Khoisan population, imported Malay slaves, and...

To continue reading