A violent mix? The association between concurrent alcohol and cocaine use and violence amongst young people in England and Wales

Abstract

Aim: This study explored the association between concurrent alcohol and cocaine use and its predictive probability of self-reported violent behaviour. Methods: A series of logistic regression models were run on a sample of 3098 young people (aged 16–25) from a national self-report study. Findings: Current cocaine use was identified as predictive of violent offending (Exp B = 2.363, p ≤ 0.01). Cocaine use was also more likely in those reporting heavy episodic drinking. Whilst findings suggested an additive risk for both heavy episodic drinking and cocaine consumption, there was no evidence of a multiplicative risk associated with concurrent use on the probability of assault outcomes. Results also suggest that heavy episodic alcohol drinking was mediated by experience of violent victimization and having been involved in anti-social behaviour. Conclusions: Findings suggest that whilst interventions aimed at reducing drinking or cocaine consumption are likely to have some effect in reducing offending behaviour, attention should also be given to other individual level factors such as prior violent victimization and wider involvement in anti-social behaviour as these may be confounded as part of a wider substance misusing lifestyle or associated with violence as part as a wider repertoire of anti-social behaviour.

Notes

1. ‘Experience of violence was measured by asking whether clients were personally involved (either as a perpetrator or a victim) in an incident in which someone was pushing, grabbing, hitting, kicking, threatening with a weapon or being physically aggressive in any other way in the past year (Wells, Graham, & West, 2000)’ (Macdonald et al., 2008, pp. 202–203).

2. Audio-CASI allows respondents to listen to questions and possible answers via headphones before entering their response directly into a computer.

3. The model presented in this paper was run both with and without the weights applied. The models did not differ in the pattern of findings they yielded; thus, the weighted model will be presented here, as it optimizes the representation of the target population.

4. The variables entered into the modelling procedure were: being male, age, whether respondent was white or not, whether respondent was single or not, whether respondent had been a victim of violence, whether respondent had perpetrated ASB, whether the respondent visits a pub two/three times a month or more, whether the respondent visits a club two/three times a month or more and whether the respondent has received A-level qualifications or not.

5. An additional modelling step was performed, although not presented here first examining the effect of introducing age and gender which did not mediate the heavy episodic drinking coefficient.

6. Defined as an average of at least four ounces of pure alcohol per day.