Using historical context to identify confounders, mechanisms and modifiers in aggregate level studies of the relationship between alcohol consumption and violent crime

The paper by Bye (this issue [1]), is a well-conducted study that both covers an impressive length of time and controls systematically for factors hypothesized previously as accounting for the aggregate-level relationship between overall alcohol consumption and violent crime. The introduction provides a reasonable rationale as to why divorce, marriage, fertility, unemployment, gross national product (GNP), public assistance and percentage of population aged 15–25 years might confound the link between alcohol consumption and violent crime. The problem is that none of the factors was found to be related to changes over time in both alcohol consumption and violent crime in Norway. That is, despite the rationale, these factors did not confound the relationship between alcohol consumption and violence—which leaves the question open as to whether important confounders were actually controlled. I would not dispute the likelihood that alcohol has a causal contributing relationship with violence [2]; however, I think the conclusion is premature that the relationship is due necessarily or wholly to an independent effect of alcohol based on the results of the present study. This conclusion requires demonstrating convincingly that the main important confounders are taken into consideration, which is difficult because identification of confounding factors is an inductive process. One way of increasing the likelihood that key factors are considered is to collaborate with historians and criminologists who are familiar with these time-periods in Norway. For example, there really needs to be some explanation for the steep rise in violent assaults from 1970 to 2000. It does not seem likely that this increase in violence is due entirely to increased alcohol consumption. Similarly, the increase in alcohol consumption during the period from 1955 to 1980 also remains unexplained. Does it relate to increasingly liberal policies regarding alcohol? If so, was this liberalism part of greater societal changes that might also include factors that would affect violence rates (i.e. suggesting the possibility that changes in both alcohol consumption and violence were caused by a common third factor)? Alternatively, changes in alcohol consumption and violent crime may have been due to separate unrelated factors (e.g. liberalism and enhanced enforcement, respectively) that simply happened to co-occur and result in a high (but spurious) correlation between alcohol consumption and violence. A good example of applying historical context to explaining alcohol use and violence is the work of Jessica Warner, explaining historical factors affecting youth drinking [3] and using time-series analyses to assess the relationship between violence and key changes in historical context [4]. Greater attention to the role of historical and other factors in the alcohol–violence relationship may also help to identify specific mechanisms. For example, existing aggregate-level studies have demonstrated that factors such as public drinking [5], outlet density [6], beverage type [5,7] and beverage type within country [8] are associated with the rate of violence over and above overall alcohol consumption. These findings indicate that the impact of an increase in overall consumption of alcohol may depend on how that increase is manifested. For example, if the increase in alcohol consumption is accompanied by a corresponding increase in drinking in bars, pubs or clubs, then violence might also be expected to increase, given the higher risk of violence in these settings [9–12]. On the other hand, if the increase in alcohol consumption is primarily in the form of an increase in wine drinking by an ageing population, a corresponding increase in violence is likely to be minimal [5]. Thus, historical context and drinking pattern data may be key both to identifying possible mechanisms of the alcohol–violence relationship (e.g. increased consumption > increased exposure to risky drinking environments > increased violence) as well as possible conditional or interactional relationships, as explored by Parker & Rehbun [13] in their evaluation of the extent that factors such as poverty, age and availability modify the alcohol–violence association. It is not only that key confounders may be overlooked by ignoring historical context; it may also be that hypothesized confounders in the current analyses were measured inadequately as historical variables. For example, I would argue that the percentage of population aged 15–25 years is a poor proxy for changes in routine activities over time. In particular, although the population has been ageing in many parts of the developed world during the latter part of the 20th century, this has also been atime of increasing activities outside the home as well as increasing urbanization, both of which are likely to contribute to an increase in risky routine activities that may be unrelated or even negatively related to the proportion of the population aged 15–25 years. That is, while at any given time young adults may be more likely than other age groups to engage in risky routine activities, changes in the proportion of people aged 15–25 years may not mirror important changes in routine activities over time. It is also possible, as suggested above, that changes in routine activities may modify rather than confound the alcohol–violence relationship. On the other hand, while the percentage aged 15–25 years may not be a good proxy for routine activities, the proportion of young adults would be expected to be an important factor in predicting drinking and violence. However, this effect may be gender-specific. It is well established that males in this age group are over-represented among violent offenders [14,15]. In addition, across cultures young adults, particularly young males, are more likely to engage in heavy episodic drinking [16,17]. This suggests that an overall increase in alcohol consumption would have a stronger impact on male violence than on violence generally, which has been found by several studies [7,8,18]. Thus, it would be useful to model assault rates separately for male and female offenders, if these figures are available. To conclude, these comments are not intended as criticism of this excellent and sophisticated analysis. Moreover, some of the issues raised here were already recognized by the author in her discussion of the findings. Rather, my main point is that for this research to reach its potential in identifying and understanding the relationship between changes in alcohol consumption and violence at the aggregate-level, future research needs to involve collaboration between alcohol researchers, historians and criminologists to ensure that the best explanations for changes over time in both alcohol consumption and violence are considered.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.