The rise of suicides within the US military has been one of the most tragic and noteworthy aspects (among many) of US efforts in Operation Iraqi Freedom (OIF) in Iraq and Operation Enduring Freedom (OEF) in Afghanistan. OIF and OEF have also been marked by the extensive use of suicide-attacks by insurgents. In the respective areas of medical/psychological and military/conflict-studies research these two types of phenomena have been rigorously documented and analyzed, but they also seem to have remained almost entirely separate and disconnected. As far as I’m aware, no published research has considered the possibility of a relationship between the two. While there are certainly reasons to consider suicide-attacks and military suicides as unrelated and distant phenomena, a connection between the two is also worth considering.
I’ve had a long-standing interest in thinking about the ways that US military service-members might be influenced and affected by the kinds of warfare they have been engaged in, the patterns of behavior they’ve been immersed in, and the relationships with members of local populations that they’ve built and maintained during their service. I think conflict research tends to be biased towards thinking about ways that US and international forces and officials influence local populations, and doesn’t think enough about influence that runs in the opposite direction. Much of my graduate research on the post-conflict state-building process in Bosnia looked at that influence (though admittedly not very rigorously). My interest was reinforced after the time I spent in Afghanistan and specifically after embedding for several weeks with Special Forces troops implementing the Local Defense Initiative program (now Village Stability Operations).
But I didn’t think much about suicide in the US military until walking across Harvard Square one evening last fall. That evening there was a smallish demonstration for a free Tibet, and the flier they were handing out mentioned the self-immolation of 6 women in Tibet over the past couple of months (31 since the beginning of 2011). When I read that it occurred to me that self-immolation had been a popular topic for the uprising in Tunisia, and I briefly wondered if a tactic like self-immolation could spread. The next day I saw this question in a headline, and then randomly this, and noticed picture 19, and I thought, “That woman probably wasn’t embedded in a signaling network that encouraged the attack-form of suicide. But it’s even more unlikely that she was unfamiliar with suicide. If suicides spread, I wonder what it would look like if you compared suicide attacks in Iraq and Afghanistan with suicide rates in the military”. At this point I knew the idea might have real merit, but that it might also be simply wrong. The only way to find out which was right was to consider the evidence.
In other words, it’s an empirical question. Empirical questions, of course, can be really hard to answer. So I talked to Schaun about the merits of pursuing the research (given that we both have full-time jobs focused on other areas of research). We decided it was worth it and I persuaded another collaborator to work on it with us. We now have most of the data we need to conduct some initial analyses. Before we do that, I wanted to present and explain our case for why it’s reasonable to look for a relationship between suicide-attacks and US military suicides. That’s what this post is about. More about the actual analysis and any findings will come later.
Since 2003 the US military has experienced an almost continuous rise in suicides among service-members and veterans. The Army suicide attending-rate in 2008 was the highest in 25 years (1). While there are plenty of other factors that should and are being considered, the rising suicide rate is plausibly related to service members’ military deployments and combat experiences. A substantial body of research has connected combat experience with suicidal ideation, as well as actual suicide (2, 3, 4, 5). One paper found results that suggested that all forms of combat exposure predict higher levels of capability to commit suicide. It also found that exposure to combat events involving aggression and high levels of exposure to death and injury independently demonstrate stronger associations with the capability for suicide than combat events that do not entail explicit exposure to death or aggression. However, there appears to be minimal research on the mechanisms that facilitate the influence of deployments and combat experience on suicide.
Both OIF and OEF have been characterized as counterinsurgency warfare. Unlike the more conventional warfare that occurs between nation-state militaries, counterinsurgency tends to unfold primarily between individual units and soldiers and individual insurgent groups and insurgents. At the same time civilians are generally at the very center of military thinking and actual operations. Counterinsurgency often becomes a very personal kind of conflict. For U.S. service-members in Iraq and Afghanistan, drinking tea with village elders and helping farmers dig wells are often integral parts of their deployments. Even the more conventional military activities of counterinsurgency like targeting can become extremely personalized (tracking the life-histories and movements and networks of individual insurgents – knowing exactly what they look like and even naming them).
OIF and OEF are also distinguished from many other conflicts involving the US military by the extensive use of suicide attacks as a military tactic of insurgents. Since 2003 there have been approximately 1,400 suicide attacks carried out in Iraq and Afghanistan (according to CPOST). Yet to our knowledge, that aspect of OIF and OEF has not received attention within research concerned with suicide rates within the US military. Presumably that’s because they’ve been assumed to be entirely unrelated. However, a well-known theoretical approach to suicide research, the Interpersonal Psychological Theory of Suicide (IPTS) seems to conceptually support our hypothesis that exposure to suicides carried out by insurgents has contributed to an increase in suicides within the US military.
The IPTS proposes that in order for a person to actually die by suicide, he or she must have both the desire to die by suicide and the ability to enact that desire. The capability piece was introduced partly to account for the fact that far more people think about suicide and intend to commit it than there are actual cases of suicide. For our hypothesis we’re focused on the latter element of the theory: capability. The capability to enact lethal self-injury is supposedly acquired through exposure and habituation to the fear and pain involved in self-injury. One paper has suggested that the capability to commit suicide develops as a function of repeated exposure to painful and provocative events. Research associated with the IPTS has attempted to evaluate the capability to enact suicide using such measures and tools as the Acquired Capability for Suicide Scale (ACSS) and Combat Experiences Scale (CES), a 23-item checklist of combat-related experiences.
Neither of those measures specifically incorporates exposure to suicide in combat as a possible contributing factor. This is probably because suicide attacks have not historically been a common feature of combat experience for US service-members. That has changed with OIF and OEF, and suicide has become both a sensational (in the sense of strongly affecting the senses) and relatively regular feature of life and combat in both theaters. Different types of combat experiences might differentially contribute to the acquired capability for suicide. Exposure to suicide plausibly carries relatively significant weight as a combat factor that contributes to service members’ capability to commit suicide. Such exposure might be subject to social influence and the social contagion effect and to the phenomenon of copycat suicides.
Social influence and social contagion are one of the most consistent and well-validated subjects of psychological research. Recent research has investigated a range of phenomena including obesity, happiness, knowledge, loneliness, and many other behaviors and affective states (attitudes, emotions, etc.) that can spread from person to person. The influence of social networks has also been shown with more specifically malign and self-harmful behaviors, including binge-eating, eating disorders, alcohol consumption behavior, smoking, and drug use.
In addition to these more general findings, a number of studies have shown that suicide behavior is susceptible to the influence of suggestion. This is commonly referred to as the Werther effect or copycat suicide. Usually copycat suicide is referenced in relation to media reporting and the effect that media reporting has on suicides immediately after a publicized story of suicide. A meta-analysis published in 1999 (and again in 2003) reviewed 293 findings from 42 studies and confirmed that suicides do tend to increase in the period immediately after media coverage of a suicide. But evidence for the influence of suggestion on suicides has also been found in more surprising cases. In 1979 the sociologist David P. Phillips noted that car accidents were the 5th leading cause of death in the U.S. and then investigated the hypothesis that some car accidents have a suicidal component (if they did, he hypothesized they should follow the patterns demonstrated by other kinds of suicide). His paper in the American Journal of Sociology found that three days after a publicized suicide, automobile fatalities increased by 31%. In addition, the more the suicide is publicized, the more the automobile fatalies increase, the age of the drivers was correlated with the age of the person described in the suicide story, and single-car accidents increased more than other types just after the publicized suicide. This kind of research has been updated fairly regularly, for example a paper in 2011 found that railway suicide incidents increased after media coverage of a fatal railway accident.
For service-members in Iraq and Afghanistan, stories about suicide are often a regular feature of everyday life. The first time I was driven through the streets of Kabul, I remember not being able to prevent myself from looking at every small white corolla (and they’re everywhere) and wondering for a second whether it would swerve over into us. When an attack happens you hear about it. And for those service-members that are out in rural areas and interacting with local populations, suicide is even more close – something you see, hear, and touch. You see young boys blow themselves up, or even just pretend to. Being exposed to suicide doesn’t mean you’ll commit it, far from it, but the addition of exposure to a confluence of other factors can make taking your life much more likely. And the effect of that exposure doesn’t necessarily go away just because the suicide was also murder.
Data (and basic research design)
I won’t say much about our research design or methods in this post, it’s already long enough. As we develop and implement our methods, I’m sure we’ll have a post describing them. I do want to briefly outline our data for those who are interested in the more technical aspects of the research question that I’ve outlined above.
Our initial analysis will focus on two primary measures: 1) Records of service-member suicides in Iraq and Afghanistan. These will ideally be broken down temporally by day, month, and year, and spatially by district, province, and country. 2) Records of suicide-attacks in Iraq and Afghanistan.
Unfortunately it’s almost impossible to get complete data on suicides in the US military. The DoD releases overall numbers, but not much more. iCasualties.org maintains a fairly reliable record of all overseas US service-member deaths in OIF and OEF. Each record is typically accompanied by a brief description of the cause of death. Many suicides are listed. However, the military also decides not to report the cause of death for many non-hostile casualties. I did media (e.g. lexis-nexis) and general web searches to gather additional qualitative data regarding those non-hostile casualties. This facilitated the identification of several more suicides. The remaining unidentified non-hostile casualties we will attempt to deal with during the analysis. They will constitute one of the most important and severe limitations of our initial analyses. We’re currently considering multiple methods of further clarifying the validating the data, and those methods will undoubtedly be at the forefront of future posts and discussions.
To show that suicide attacks influence service-member suicides, we’ll need to be able to discount the possibility that suicide attacks are not just proxy for the effects of soldier deaths caused by hostile activities, natural causes, accidents, etc., or just the overall stress of a conflict environment. Therefore, we need these three measures as well. Much of that information (including unit information) is also available in the icasualties dataset. We’re also exploring the possibility of acquiring data from the authors of several published studies. Our primary suicide-attack data will come from the Suicide Attack Database at the Chicago Project on Security and Terrorism.
A personal note
The ideas and hypothesis I’ve introduced above are extremely fraught and sensitive ones. I know many current and former members of the military who, based on my experiences with them, might react somewhat harshly and negatively to what I’m suggesting. Official military policy hasn’t historically been very open to discussion of suicide. I know that tendency also reflects the tendencies and preferences of many individual members of the military. I’m still not entirely sure what I think or how I feel about the subject. It can be a severely uncomfortable thing to think about. Unfortunately, I think such discomfort has too often prevented serious and necessary consideration of the subject.
Finally I want to note the title of this post. I’ve questioned whether it might seem too flippant, or too political. I want to be neither. I’m not sure I like the term enemy, nor do I think it generally has much merit. The term infers way too strongly the idea of a zero-sum reality. It characterizes conflict as the result of opposing essences and identities, rather than problematic conditions. It communicates too strongly the idea that a man is fought because of who he is – the enemy – rather than because violent interaction has emerged from a set of overlapping and shared conditions. Changing those conditions, not eliminating the enemy, will be the most effective means of success.
Ultimately I decided to keep the title because I think it conveys something of the contradiction that we’ve seen in Afghanistan. The contradiction between a strong and prevailing idea that insurgents are the enemy, and the opposing possibility that rather than being on opposite sides of a fundamental gap, US service-members and insurgents are operating in a shared environment and are affected by a set of shared conditions which lead them to tragic behaviors that we would ordinarily consider completely unrelated. I believe that possibility is worth investigating.