The Lancet special issue on Violent Conflict and Health (which included the Darfur mortality report — see previous entry in this blog) includes research undertaken in the Palestinian Territories (the West Bank and Gaza) concerning the association between political violence and intimate-partner violence (i.e., domestic violence).
It’s a truism that violence breeds more violence. There are several versions of this in social science research. The most obvious of these, some version of “if you live by the sword you shall die by the sword,” really doesn’t need hypothesis-testing. A less obvious version involves the association between types of violence. James Garbarino is probably the most eloquent “big picture” writer on this topic, tying together political, criminal, intimate-partner violence (or “IPV”), and child abuse. Garbarino argues that as violence within a community increases, so these other, “lower-level” violent acts increase (“lower-level” in the sense that they involve fewer people, not in the sense that they are less impactful). Garbarino has applied his theoretical analysis to war-affected communities — Sarajevo in the early 1990s — and “urban war zones” — American cities in the late 1980s and early 1990s. A number of sociologists and community psychologists have provided statistical analyses suggesting these phenomena in violent urban neighborhoods, but until recently I’d yet to see this phenomenon assessed statistically during a political conflict.
The authors of the current paper — who include researchers from universities in the US and Israel — use census data collected by the Palestinian National Authority, which evidently included responses to a rather comprehensive battery of questionnaires: measures of exposure to violence (at the hands of the Israeli Defense Forces and settlers), financial hardship, and even the Conflict Tactics Scale — the standard measure of how couples handle disputes, which has an emphasis on intimate-partner violence (hereafter: IPV). The authors analyzed data from a stratified random sample of 3510 women across the West Bank and Gaza, specifically their reports of the types of violence their husband had been exposed to, their own exposure to violence, their families’ economic situation, and IPV perpetrated by their husbands. Of course this data was collected in Arabic using Palestinian interviewers, and done in women’s homes with their husbands elsewhere (as far as I can tell, husbands were also interviewed, but these data are not part of this article).
So, what did they find? The odds that women who reported that their husbands had been exposed to “political violence” also reported that they were subjected to IPV were 1.5 (for psychological IPV), 1.9 (for physical IPV) to 2.25 (for sexual IPV) times that of those who reported that their husbands were not exposed to political violence. In plain English, exposure to political violence was associated with a higher likelihood of every type of IPV as well. So, IPV was explained by exposure to violence, right?
Well, partly, maybe. If I read the tables correctly, rates of IPV were uniformly higher than rates of exposure to violence. 371 women reported some form of sexual IPV and out of 3510, that’s 10.6%; 538 physical IPV, or 15.3%; 1302 psychological IPV, 37.0%. The number of women reporting exposure of their husbands to any type of political violence was 289 — 8.2%. In other words, 63% of the sample reported some form of IPV, but 8% reported exposure to violence… so clearly violence was not the primary cause of IPV.
Following this little back-of-the-napkin, my first thought was, “Wait a minute, they surveyed Gaza and the West bank, included things like being ‘insulted or cursed’ in their definition of violence and still only came away with a rate of exposure to violence of 8%?” I admit to being pretty ignorant about the Israeli-Palestinian conflict, but… 8% seems pretty low. Also, how about other sources of violence? With Fatah and Hamas fighting each other, there must have been some respondents who were exposed to non-Israeli-perpetrated violence, and that too would have had an effect on their family lives, I would imagine. [Note: the data were collected before last year's war between Hamas and Israel in Gaza, so presumably rates of exposure would be greater now.]
My second thought was, “Hm. Perhaps I’m getting distracted by the title, which includes the drama-loaded “political violence” and “intimate-partner violence.” In all fairness, it’s not really the purpose of this paper to push the explanation that the cause of IPV in the West Bank and Gaza was violence, so let’s not ding the authors for not pointing out that IPV was much more common than exposure to violence.” And so, ding them I will not.
I do, however, wish they had made more of another piece of data they collected: household financial data. It turns out that this too had an effect on IPV: odds for those reporting financial hardship were 1.4 for psychological IPV, 1.5 for physical IPV, and 1.6 for sexual IPV (compared to the 1.0 odds for those not reporting financial hardship due to the conflict). So what? Almost half of the women — 1695, or 48% — reported this hardship. Compare this to the 63% rate of IPV. Now we’re getting somewhere.
What’s the problem here? Well, it’s not the data, nor even the authors’ analyses. There are two things that are tricky here. First the effects of violence on IPV (1.5, 1.9, 2.25) are larger than those for economic exposure’s effects on violence (1.4, 1.5, 1.6), but exposure to violence is actually a much rarer occurrence (8% versus 48%). Effect sizes (here odds ratios) indicate the strength of an association, not the breadth of the occurrence of any one of the variables in the association. In other words, if you wanted to reduce IPV in the West Bank and Gaza and you wanted to bet that you would get have an effect, you would try to reduce exposure to violence; but, if you wanted to maximize your chances to reduce IPV for the most people, you would try to reduce economic hardship.
The second tricky thing here is really related to the first: political violence is much easier to sell than economic hardship, so the authors put it (and not economic hardship) in the title. After all, the special issue of the Lancet was not about the health effects of economic hardship, was it? (That’s not to say that the Lancet doesn’t do pretty well on that issue too; it does.) But the issue that draws our attention to conflict — violence — is often not the issue that affects the most people.