Bad nights or bad bars? Multi-level analysis of environmental predictors of aggression in late-night large-capacity bars and clubs

Aims To clarify environmental predictors of bar-room aggression by differentiating relationships due to nightly variations versus across bar variations, frequency versus severity of aggression and patron versus staff aggression. Design, setting and participants Male-female pairs of researcher-observers conducted 1334 observations in 118 large capacity (> 300) bars and clubs in Toronto, Canada. Measurements Observers independently rated aspects of the environment (e.g. crowding) at every visit and wrote detailed narratives of each incident of aggression that occurred. Measures of severity of aggression for the visit were calculated by aggregating ratings for each person in aggressive incidents. Findings Although bivariate analyses confirmed the significance of most environmental predictors of aggression identified in previous research, multivariate analyses identified the following key visit-level predictors (controlling for bar-level relationships): rowdiness/permissive environment and people hanging around after closing predicted both frequency and severity of aggression; sexual activity, contact and competition and people with two or more drinks at closing predicted frequency but not severity of aggression; lack of staff monitoring predicted more severe patron aggression, while having more and better coordinated staff predicted more severe staff aggression. Intoxication of patrons was significantly associated with more frequent and severe patron aggression at the bar level (but not at the visit level) in the multivariate analyses and negatively associated with severity of staff aggression at the visit level. Conclusions The results demonstrate clearly the importance of the immediate environment (not just the type of bar or characteristics of usual patrons) and the importance of specific environmental factors, including staff behaviour, in predicting both frequency and severity of aggression.