Nightly variation of disorder in a Canadian nightclub

Abstract

Objective

This paper aims to study nightly disorder within a single bar over an extended period, in order to analyse variations across time (n = 258 nights).

Methods

The security staff of a large Canadian nightclub agreed to note detailed information on every intervention in which they were involved. Bouncers wrote detailed narratives of each incident of aggression and incivility that occurred in the bar. Environmental characteristics (e.g. number of admissions and alcohol sales) were collected by one of the co-authors.

Results

“Hot nights” were observed. The number of problem events was particularly high on Tuesday nights, which had the highest number of customers admitted and higher alcohol sales. The average alcohol sale per customer was also higher during long weekends, and alcohol sales were positively related to problem events. Finally, path analyses revealed that the presence of more bouncers was a deterrent.

Conclusions

The level of disorder in a bar varies greatly over time. Contrary to what is often postulated, bars are not always high- or low-risk. The results strongly support responsible alcohol-serving policies and highlight the benefits of adequate surveillance.

Introduction

Drinking establishments have a very strong incentive not to collect detailed data on disorder within their walls. Their popularity and profitability rely on their reputation, which is mostly based on the general atmosphere and perceived safety within premises. The current study provides a useful exception, supplementing past literature on what happens within bars by following a single bar every night, over approximately 1 year.

Bars and other public drinking premises are associated with many forms of disorder. For example, the risks of being a victim of assault in a bar are two times higher than in any other commercial setting (Burrows et al., 2001). Drinking (and sometimes, excessive drinking) is promoted and unusual behaviours are largely tolerated. Security interventions are guided by two general principles: profit and safety. In other words, judgement calls are made on disorder within drinking premises, which is managed through relative control rather than traditional rule enforcement.

Many studies have shown that the level of disorder varies greatly between bars (Graham & Homel, 2008). Much has also been written on the importance of good management: aggressive acts are more frequent in bars with long line-ups (Graham et al., 2004), in crowded and small venues (Graham et al., 2006a), in bars where bouncers/security staff are too few, aggressive or poorly trained (Quigley et al., 2003), and where alcohol is served with little consideration for the client’s level of intoxication (Buka & Birdthistle, 1999; Gliksman et al., 1993; Stockwell, 2001). Moreover, bars attracting specific clienteles are higher risk: for example, bars visited by young men displaying “macho” attitudes (e.g. sexual competition) are more likely to have high rates of assault (Benson & Archer, 2002; Graham & Wells, 2003; Graham et al., 2000). Potential offenders who perceive low risks of punishment tend to misbehave, especially when alcohol lowers their inhibitions (Cohen & Felson, 1979; Graham et al., 2006b; Pihl & Peterson, 1993).

Such studies are useful to target high-risk venues. Three main hypotheses are derived from “between-venues” analyses. The first one suggests that the level of disorder fluctuates depending on the type of clientele. For example, a crowd of mostly young male patrons in a bad mood will be associated with more trouble than an older, easy-going female clientele will. The second hypothesis stems from the main characteristic of licensed premises: alcoholic beverages are served. The level of intoxication is thought to be closely related to disorder and aggressive acts, meaning that an appropriate monitoring of customers’ drinking should reduce related problems (Graham et al., 2005a, 2005b). The third hypothesis was summarised by Kathryn Graham (2009): “They Fight Because We Let Them”. Regardless of the clientele, security staff members should be easy-going, well-trained and able to keep everything under relative control.

All three hypotheses imply possible variations within premises. For example, the level of intoxication of a crowd is related to various contextual (e.g. day of the week and drink specials) and individual factors (e.g. clients’ age or occupation). However, a very limited number of studies have investigated the variations of incivilities and aggressive acts within bars. Graham et al. (2006a) provide the only quantitative assessment of factors associated with the level of disorder from one night to the other, based on observations in 118 large-capacity bars and clubs in the city of Toronto, Canada. They found that the proportion of clients aged less than 25 years was negatively associated with the frequency of violence, and that other clientele characteristics were not significantly related to the frequency or severity of aggression. On the contrary, Graham et al.’s results support a preventive security argument: the frequency of aggressive acts within a bar significantly increased with higher levels of rowdiness and permissiveness. While they found no significant relation between intoxication and the frequency of assaults, Graham et al. (2006a) also showed that there was a strong positive correlation between the level of permissiveness and the intoxication of customers. In other words, aggressive acts were more frequent when patrons were more intoxicated while rule enforcement was weak and inconsistent. Closing time is also crucial as the number of people hanging around after closing is significantly associated with the frequency and severity of aggressive acts. Surprisingly, none of the “staff variables” reaches levels of statistical significance (p < 0.05) in multivariate analyses, i.e. the presence and coordination of the security staff was not associated with the frequency of aggression.

Despite significant insights gained from the work by Graham et al. (2006a), their analysis has some limitations. A major issue is that the data was not collected specifically to analyse changes within venues, but rather to assess the effectiveness of an intervention to reduce aggressive acts in bars and clubs in Toronto (Graham et al., 2004). As such, it covered only a 2-hour period before closing, on Fridays and Saturdays. These moments are high-risk periods, but incivilities and assaults can occur at other times. Also, the number of observations was limited to an average of 11.3 nights per venue over a 2-year period. Finally, several measures depend on the attention and judgment of the observers present during the evening. A total of 148 different observers were hired over the course of the project (Graham et al., 2005a, 2006c). Authors report that the level of inter-observer agreement was sometimes “lower than desirable” for key predictors (Graham et al., 2006a); despite rigorous training and selection, observers could not see everything and their comments remain the subjective reading of a situation (Graham et al., 2004, 2006a, 2006c; Purcell et al., 2003). Yet, Graham et al. (2006a) were able to demonstrate that there are not only “bad” bars but also bad nights: peaceful bars can experience turbulent nights, and vice versa.

The main objective of the present study is similar to Graham et al. (2006a): it aims to assess the influence of dynamic factors on the level of disorder and aggression within a bar. With its focus on a single bar, it provides a detailed understanding of the variation of disorder over a year of business. The data was collected by a small group of participant-observers and covers all opening hours.

Method

A 1-year long data collection took place between 27th April 2006 and 24th April 2007 in a long-running bar located in downtown Montreal. With the exception of minor renovations (fresh paint) and the appointment of two bouncers, there were no significant changes during the year under study. The bar had a legal capacity of 475 patrons but the number of customers inside often exceeded 600, in part to reduce the line-up in front of the venue. It was closed on Mondays and Wednesdays but open on all other nights from 8 p.m. to 3 a.m, for a total of 258 complete nights of observation. Loud mainstream music was played, and dancing was a popular activity.

At that time, one of the co-authors was an employee of the bar while completing a master’s degree in criminology. He had been employed for 4 years and was a member of the security staff (n = 12 bouncers). He led this data collection under the supervision of a Professor at the School of Criminology. He convinced the bouncers to note detailed information on every intervention in which they were involved. All staff members were informed of the goals of the research at the beginning of the project. The co-author would discuss the event at a later time with the staff member and complete a form. When the co-author took part in an intervention, a non-involved staff member was responsible for filling the form in order to reduce personal bias. Contextual information (admissions, special events and alcohol sales) was collected every night.

Dependent variables

The bouncers had to indicate the reason why they intervened. Thirty-three different reasons were noted during the year. In retrospect, two general categories clearly emerged: aggressive acts and incivilities. Aggressive acts include all acts of undesired physical contact and verbal threats (n = 242). Approximately 40% of incidents in this category involved assaults on staff members (primarily bouncers), the rest being conflicts between patrons. Incivilities include a variety of other forbidden behaviours, from consuming illegal drugs to throwing objects on the ground and having sexual intercourse inside the building (n = 547). In general, incivilities were not potential threats to other customers; still, offenders were expelled in 88.0% of the cases because they were visibly intoxicated or resisted the intervention.

Daily counts of aggressive events and incivilities were analysed. There were no security interventions on 15.5% of the nights, one or two on 38.0%, and more than five on 15.1%.

Independent variables

Different social contexts may have direct and indirect effects on problem events. Some social contexts can affect individual behaviours and are likely to be associated with the general mood of the crowd and security staff. For example, hockey is the most popular sport in Canada (audiences of more than 1 million viewers – on a possibility of about 8 million French-Canadians). A game monopolises the interest of many people and a victory of the local team brings supporters together, especially during the playoffs (1 = Montreal win, 0 = Montreal loss or no game). Long weekends also bring different crowds on Fridays, Saturdays and Sundays (1 = long weekend).

Similarly, different nights attracted different kinds of people with different drinking patterns. An interesting feature of the bar under study is that nights were organised around specific themes that attracted diverse crowds (Table 1).

Table 1.

Description of business nights (mean; standard deviation).

Night of the week	Theme	Clientele (admissions)	Entrance fee	Alcohol (drinks/admission)	Aggressive acts per night	Incivilities per night
Sunday	“Frenchy”	Regulars 486 (212)	3$	Early deals 5.5 (1.3)	0.7 (0.9)	1.9 (1.1)
Tuesday	“Retro”	University students 843 (187)	6$	Cheap beer pitchers (48 oz.; 1,44 l) 9.5 (2.8)	2.2 (1.9)	4.1 (2.9)
Thursday	“Hits”	College students, locals 549 (169)	3$ (students), 5$	Cheap beers and cocktails 5.9 (2.5)	0.8 (0.9)	1.6 (1.2)
Friday	“XL”	Tourists, visitors 588 (131)	4$	Cheap distilled spirits 3.2 (0.7)	0.9 (1.0)	2.1 (1.6)
Saturday	“80s”	Tourists, visitors 682 (197)	4$	Cheap distilled spirits 3.2 (1.1)	0.7 (0.8)	2.4 (1.6)

On Tuesdays and Thursdays, most patrons were students from four major universities and a nearby college. Fridays and Saturdays brought mostly people visiting Montreal, while the majority of customers on Sunday nights were locals. The cover charge was highest on Tuesdays, but it remained the most popular night of the week. According to the bouncers, Tuesdays were by far the most turbulent nights. They highlighted that, on Tuesdays, the clientele was different from any other night, since they witnessed a lot of sexual competition between male patrons. A priori observations support their gut feeling: the 10 most-turbulent nights of the year were Tuesdays. A dummy variable (1 = Tuesday) was thus included in multivariate models.

An estimation of the average level of intoxication is given by dividing the total income from alcohol sales by the price of the top-selling beverage and the number of admissions on a given night (alcohol sales/(price of top-selling drink × number of admissions)). The calculation gives the approximate number of drinks consumed per customer on a given night. As shown in Table 1, customers bought more alcohol on Tuesdays (9.5 drinks) than on any other night. While it was impossible to determine precisely the types of drinks sold every night, beer was by far the most popular drink, especially on Tuesday nights. Alcohol sales per customer were surprisingly low on Fridays and Saturdays (3.2), which could potentially be explained by “barhopping” – customers spend less time at a given bar but visit and drink in other establishments close by (Felson et al., 1997). The bar under study was located on a popular street segment with a high concentration of drinking premises.

Control variables include the number of admissions and the number of bouncers.

Analysis

A path analysis is used. Analyses were performed with SPSS AMOS 19.0. A path analysis is used to test directed hypotheses using multiple models. This technique makes it possible to consider a set of relationships between independent variables and more than one dependent variable. Figure 1 shows the conceptual path model between social contexts and problem events. Direct effects are expected, as well as indirect effects through the number of customers admitted and the average number of drinks consumed by customers.

Figure 1.

Conceptual model for the impact of social contexts on problem events.

Results

Final path models for aggressive events and incivilities are presented in Figures 2 and 3. Standardised coefficients for significant relations (p < 0.10) are shown. Full models are presented in Appendix A. Various goodness-of-fit measures suggest that the models explain a significant portion of variance (see Appendix A for details).

Figure 2.

Path diagram for aggressive acts. **p < 0.01; *p < 0.05; †p < 0.10.

Figure 3.

Path diagram for incivilities. **p < 0.01; *p < 0.05; †p < 0.10.

Of the three social contexts under study, only Tuesdays have a direct relation with both aggressive acts and incivilities. This confirms the “Tuesday effect” suspected by bouncers. Tuesdays seem to attract troublesome customers. In contrast, long weekends are associated with a decrease in aggressive acts. Long weekends create a festive atmosphere in which patrons celebrate with each other, apparently leading to less interpersonal conflicts. This hypothesis is supported by the fact that aggressive acts also decrease when the local hockey team wins.

Social contexts also have indirect effects on problem events. During long weekends, patrons are in a festive mood, leading to more alcohol consumption, which in turn is associated with more aggressive acts and incivilities. The average number of drinks per customer is a significant predictor of aggressive acts and incivilities. This result strongly supports a responsible alcohol-serving policy (e.g. refusing to serve visibly intoxicated people).

The number of customers admitted in the bar is higher on Tuesdays than on any other day. As expected, there is a strong positive relation between admissions and problem events (aggressive acts and incivilities). A deterrent effect is also suggested since there is a significant negative relationship between the number of bouncers and the number of problem events detected, when controlling for admissions. The effect is stronger for incivilities than aggressive acts. As one of the bouncers put it, “there’s only so much you can do”. The presence of a sufficient number of bouncers may have a deterrent effect on most patrons, but some aggressive acts and incivilities are difficult to prevent, especially if intoxicated individuals commit them.

Discussion

This paper provides an in-depth investigation of the variations of disorder within a bar over a year of operation. It is based on detailed information collected by a small team of observers in charge of the security of the bar. The results further demonstrate the impact of social contexts on aggressive and unruly behaviour. They show that the level of disorder in a bar varies over time. Contrary to what is often postulated, bars are not always high- or low-risk.

The analysis also expands the work of Graham et al. (2006a) by considering not only the highest-risk periods (Friday and Saturday nights), but also all opening hours. In this case, Tuesday nights were by far the busiest for security staff. Tuesdays had more admissions and more alcohol sales per customer. In addition, bouncers described customers as being generally more rowdy on Tuesdays; this statement is supported by direct relations between the variable “Tuesday” and both measures of disorder.

Despite these results, none of the bouncers felt that security was an issue in the bar under study. During informal interviews, all bouncers pointed out an ethical dilemma they faced. On the one hand, security staff must keep the bar safe and make sure that patrons follow rules. On the other hand, their jobs and wages depend on the popularity of the bar, which is mostly reflected by alcohol sales. Serving alcohol brings monetary benefits, but the more the customers are intoxicated, the more they have inappropriate behaviours that threaten other patrons and employees. One of the bouncers summarised the situation nicely: “My job is to maintain an environment that is safe enough for customers to spend more”. This feature distinguishes bouncers from police officers. Bouncers are primarily handlers dealing with individuals threatening the economic survival of the bar, while police officers are service-minded guardians (Graham et al., 2005a). One of the bouncers even described a situation when he expelled a client whose behaviour was inappropriate but gave him VIP passes to come back another day.

Limitations

The analysis presented is based on observations made by bouncers in charge of security in the bar under study. While it offers unprecedented insights into the work of bouncers and first-hand data on disorder in a bar, the data is limited by its very nature. Bouncers knew the general purpose of the study and had every incentive to give an impression of professionalism and competence. As such, it is possible that they underreported a number of “less-justified” interventions or less-serious infractions.

Another limitation relates to the fact that the analysis focuses on a single bar. It was not designed to determine the impact of environmental variables, which has been evaluated by other authors (for a review, see Graham & Homel, 2008). However, the results presented in this paper are consistent with previous findings (Graham et al., 2006a). Still, the study should be replicated to validate whether the findings apply to all types of institutions. The bar we studied was relatively smooth compared to other clubs. It had a reputation of good management and was not particularly considered as a threat by Montreal police services.

Finally, although the proportion of variance explained (0.421 for aggressive acts and 0.533 for incivilities) is satisfactory by social sciences standards, the models could be improved. Some variables were not measured as accurately as desirable. In particular, the clientele could be described in more detail and other contextual information (e.g. waiting time before entrance) should be included in further analyses.

Appendix A. Full path model (standardised coefficients)

	Alcohol sales	Customers	Bouncers	Aggressive acts	Incivilities
Hockey win	0.047	−0.025	–	−0.108*	−0.068
Long weekend	0.119†	0.093†	–	−0.089†	0.007
Tuesday	0.071	0.528**	–	0.496**	0.574**
Alcohol sales	–	–	–	0.112*	0.088*
Customers	–	–	0.597**	0.297**	0.348**
Bouncers	–	–	–	−0.225**	−0.378**
Squared multiple correlations	0.021	0.288	0.356	0.421	0.533
Chi-square	–	–	–	70.110	66.806
Degrees of freedom	–	–	–	9	8
Significance (p)	–	–	–	0.685	0.457
CFI	–	–	–	0.944	0.925

^†p < 0.1;

^*p < 0.05;

^**p < 0.01.