Commentary on Norström & Pape (2010): Unleashing the beast within? Suppressed anger and changes in drinking and fighting

This paper by Norström & Pape 1 is important because it introduces two innovations to survey research on the link between alcohol and aggression: (i) assessing the role of suppressed anger and (ii) the use of change scores in assessing this link. These innovations, however, also raise some important issues worth considering. The suppressed anger explanation is consistent with Steele & Josephs' ‘alcohol myopia’ theory that aggression is most likely to occur in individuals when salient cues encouraging them to become aggressive are stronger than their inhibition against (control over) aggression 2. Although this explanation has intuitive appeal, it may not apply to alcohol-related aggression among young adults, especially young men, because they do not seem to feel ‘inhibition conflict’ about aggression when drinking 3-5. That is, it is possible that aggression occurs when young men drink not because alcohol and contextual cues release suppressed anger but because aggression when drinking is considered normative. Additionally, there is an alternative explanation for the present findings, specifically that alcohol-related aggression tends to occur among people who are more angry/aggressive generally than are other people, an explanation that has been supported by a number of studies (see review by 6). Thus, whether the link between drinking and aggression can be attributed to the ‘catalyzing effect of suppressed anger’, as argued by the authors, depends upon demonstrating that this link is due specifically to the suppression of anger and not to an angry disposition. It is especially important to clarify this link because the present findings are not consistent with experimental research showing no moderating role of suppressed anger on the link between alcohol consumption and anger expression 7. The authors partially addressed the issue of confounding of suppressed anger with anger generally by showing only a modest correlation between factor scores for suppressed anger and anger-out; however, a more convincing strategy would be to conduct comparable analyses for both suppressed anger and anger-out. The theory of suppressed anger could be supported further by demonstrating that it predicts alcohol-related aggression but not aggression generally, an analysis that was not possible in the present study but could be addressed in future research. As noted by the authors, the use of change scores has some advantages over cross-sectional analyses, in that such analyses can eliminate the role of common cause variables that make a similar contribution at both time-periods. Change scores also have some limitations, however, that should be recognized in adopting this approach. First, not only can change scores be affected by time-varying factors, but also they may be affected by stable conditions factors (e.g. predispositions) that contribute to changes in both drinking and aggression. This might be especially relevant for people in emerging adulthood. Thus, while change scores provide additional insight, it is important to avoid overestimating the extent that change scores can rule out competing hypotheses. Secondly, a potential problem for interpretation is that the proportion of fights involving alcohol is likely to be considerably higher at age 21–22 than at age 16–17. That is, the measures of aggression are not the same at time 1 (t1) and time 2 (t2) and may be related differentially to alcohol consumption 8 and other relevant factors. Thirdly, there are statistical considerations in interpreting correlations between change scores. In particular, the correlation between two difference scores is a complex function of the six correlations among the four measures as well as the post-/pre-ratio of their standard deviations 9, implicating a number of potentially different causal interpretations that need to be thought through. Other statistical concerns in using change scores include ceiling effects, regression to the mean, etc. For example, a person who has maximum scores on both intoxication and aggression frequency at t1 cannot show a measureable increase on either measure, while those at the low end of the scale have the most opportunity to increase their frequency—even though their final scores may be well below those who started out high and remained high. Finally, to adopt this approach in future research there needs to be a clear rationale for analytical decisions regarding scaling and other procedures for using changes scores and measures of suppressed anger. For example, it is not clear why the initial scores were log-transformed before differencing them. It would make more sense to assess the skewness of the raw difference scores and then use a transformation if necessary. Also, why were factor scores used to measure suppressed anger rather than simple sums of the items? Why were cross-sectional correlations reported using Pearson's r but the relationship between change scores reported as elasticities? In sum, this interesting and innovative paper makes a welcome contribution to research on the links among alcohol, aggression and anger suppression, but it also raises additional issues that need considering. None. I am grateful to Bob Gardner for sharing his statistical expertise and to both Samantha Wells and Bob Gardner for editorial suggestions.

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.