Thor Norström1, Hilde Pape. 1. Swedish Institute for Social Research, Stockholm University, Stockholm, Sweden. thor.norstrom@sofi.su.se
Abstract
AIMS: Is alcohol related causally to violence, and if so, is the effect of drinking contingent on suppressed anger such that it is strongest among individuals who are highly inclined to withhold angry feelings? We addressed these questions by analysing panel data using a method that diminishes the effects of confounding factors. DESIGN: We analysed data on heavy episodic drinking and violent behaviour from the second (1994) and third (1999) waves of the Young in Norway Longitudinal Study (n = 2697; response rate: 67%). The first difference method was applied to estimate the association between these behaviours, implying that changes in the frequency of violence were regressed on changes in the frequency of drinking. Hence, the effects of time-invariant confounders were eliminated. Analyses were conducted for the whole sample, and for groups scoring low, medium and high on a short version of the STAXI anger suppression scale. FINDINGS: Changes in drinking were related positively and significantly to changes in violent behaviour, but the alcohol effect varied with the level of suppressed anger: it was strongest in the high-anger group (elasticity estimate = 0.053, P = 0.011) and weakest (and insignificant) in the low-anger group (elasticity estimate = 0.004, P = 0.806). CONCLUSIONS: Alcohol use may be related causally to violence, but the effect of drinking is confined to individuals who are inclined to suppress their angry feelings.
AIMS: Is alcohol related causally to violence, and if so, is the effect of drinking contingent on suppressed anger such that it is strongest among individuals who are highly inclined to withhold angry feelings? We addressed these questions by analysing panel data using a method that diminishes the effects of confounding factors. DESIGN: We analysed data on heavy episodic drinking and violent behaviour from the second (1994) and third (1999) waves of the Young in Norway Longitudinal Study (n = 2697; response rate: 67%). The first difference method was applied to estimate the association between these behaviours, implying that changes in the frequency of violence were regressed on changes in the frequency of drinking. Hence, the effects of time-invariant confounders were eliminated. Analyses were conducted for the whole sample, and for groups scoring low, medium and high on a short version of the STAXI anger suppression scale. FINDINGS: Changes in drinking were related positively and significantly to changes in violent behaviour, but the alcohol effect varied with the level of suppressed anger: it was strongest in the high-anger group (elasticity estimate = 0.053, P = 0.011) and weakest (and insignificant) in the low-anger group (elasticity estimate = 0.004, P = 0.806). CONCLUSIONS:Alcohol use may be related causally to violence, but the effect of drinking is confined to individuals who are inclined to suppress their angry feelings.