LAUSR

INTRODUCTION Causal relationships between alcohol outlets and crime are inferred from their statistical associations across neighbourhoods. However, many unobserved covariates may confound these effects. Recognising that outlet sales vary by time of day and day of week, we assess whether areas with more bars/pubs, restaurants or off-premise outlets have more crime during days and times when alcohol sales are greatest. METHODS Annual administrative crime counts, sociodemographic data and other area characteristics of 336 Census block groups in Oakland, California, USA, were related to outlet densities from 2000 to 2015. Bayesian space-time Poisson models were used to measure associations between outlet densities and crime during: (i) weekday daytime; (ii) weekday nighttime; (iii) weekend daytime; and (iv) weekend nighttime periods (four seemingly unrelated equations). Comparisons of parameter estimates across equations provided an assessment of outlet effects on crime across days and times within the same analysis units using the same constellation of confounding covariates. RESULTS Assault and driving under the influence crime incidents during weekend evening hours were more frequent in Census block group areas with greater numbers of bar/pubs. Burglaries were consistently greater in areas with greater densities of restaurants. DISCUSSION AND CONCLUSIONS The spatiotemporal signature relating densities of bars/pubs over weekend evening hours to assault and driving under the influence incidents suggests that these outlets are a critical source of these crimes across neighbourhoods. Prevention programs and policies that focus upon specific drinking establishments, days and times may be most effective in reducing assault and impaired driving incidents in neighbourhoods.