Letting loose the dogs of war

 To Sky News early this morning, to talk about whether soldiers returning from Iraq and Afghanistan are more violent than members of the general public, with whom they must now live in society. The MacManus et al. (2013) paper came out of embargo today, and it has an interesting story to tell, as well as raising an issue about what we mean when we “control for” correlated variables.

MacManus et al. did a proper study, taking 13,856 randomly selected soldiers and checking their Police records on the national computer database.

http://press.thelancet.com/violentoffending.pdf

In brief, the key thing about male violence is that it drops sharply by age, probably in line with orgasm frequency, testosterone and muscle strength. Young males fight, older males reminisce.

   

Being deployed into the theatre of war did not make them more violent later on (2.4%), but serving in a combat role did (6.3%). Being exposed to traumatic events increased later offending, in a linear trend (1.6% with one trauma, 4.1% with up to 4 traumas, 5.1% for up to 16 traumas). Violent offending was strongly linked with alcohol intake (9% for heavy drinkers, 2.2% for the rest). Those with post-traumatic stress disorder were more violent (8.6%) than those without the disorder (3%). For those with the disorder, those with the hyper-arousal cluster of symptoms (an extreme sensitivity to signals of danger) were the most violent. In an interesting validation of self-reported aggressive behaviour, those with high scores on an aggression questionnaire were more violent (6.7%) than those with zero scores (2.5%).

If you look at what predicts post-service violent offending, it is rank and pre-service violent offending. In a meritocratic army, rank is strongly associated with intelligence. Those with A levels were half as likely to offend (30.8%) than those with only GCSE’s or less (69.2%).  A standard theory of violence is that it represents a severe lack of negotiating skills. It is a high cost way of achieving a personal benefit. Few mutual benefits are obtained by fighting.

The authors have followed the entirely proper technique when they look at the effects of deployment and combat, which is to correct for predisposing variables. For example, if the most aggressive soldiers are put into front line combat, which makes sense from a military point of view, then they are likely to show more violence in later civilian life, even if combat itself had no effect on them. So, you have to correct for their predisposition to violence. In fact, their tables show adjusted hazard ratios which have been adjusted for age, education level, pre-service violent offending, rank, service, engagement status, and serving status. Of course, in real life you cannot “adjust” for education levels. Those with less education and lower intelligence have fewer skills and achieve less powerful solutions to problems. The authors, correctly, have adjusted so as to make a point about the impact of combat (their adjustments reduce the real effect).

Other authors tend to go overboard with adjustments. For example, they “adjust” life achievements to take into account levels of education, and then find that intelligence does not predict much. They neglect to mention that levels of education are strongly influenced by prior intelligence: those of low abilities drop out sooner. 

What do we do about the dogs of war, who have been sent out to kill on our behalf? I think they are owed a duty of care. The UK does not have a dedicated, stand-alone Veteran’s Service. At the very least ex-service personnel hey need to be tagged by the Health Service and associated services so that their problems can be seen in the context of their military service.

What can we do about the deleterious effects of alcohol? Currently we are in the midst of a debate about alcohol pricing, and it looks like the plan to introduce a minimum floor price will be dropped.  From a libertarian point of view you should be free to intoxicate yourself as you chose. From a social point of view the costs of alcohol and drug induced violence are far too high, and restrictions are prudent. Increase cost, restrict access.

Finally, how do reduce the overall rate of post-service violence? We are having another debate about reducing expenditure on the military, as part of a general round of cuts. (Most of these do not turn out to be cuts at all. They are noble statements that at some time in the future the rate of increase of public spending will be brought down somewhat).

We must make a virtue of necessity. Here is a chance to make a real cut. Forget about recruiting thousands of soldiers. Just recruit a few well educated, clever, abstemious, officers.  That will impress the enemy.