Pages

Monday, January 25, 2010

Stupid science by The Lancet

The Lancet is one of the world's most prestigious medical journals. In recent years, however, it has moved a bit from medical to political, and the results are that it has become a purveyor of idiotic social science studies as well.

The latest version is a study that looks at Palestinian Arab men who beat their wives ("intimate partner violence") and who is to blame for it.

Take a wild guess.

Here is the abstract:

Background

Intimate-partner violence might increase during and after exposure to collective violence. We assessed whether political violence was associated with male-to-female intimate-partner violence in the occupied Palestinian territory.

Methods

A nationally representative, cross-sectional survey was done between Dec 18, 2005, and Jan 18, 2006, by the Palestinian Central Bureau of Statistics. 4156 households were randomly selected with a multistage random cluster design, from which 3815 ever-married women aged 15—64 years were identified. We restricted our analysis to presently married women (n=3510, 92% participation rate), who completed a short version of the revised conflict tactics scales and exposure to political violence inventory. Exposure to political violence was characterised as the husband's direct exposure, his indirect exposure via his family's experiences, and economic effects of exposure on the household. We used adjusted multinomial logistic regression models to estimate odds ratios (ORs) for association between political violence and intimate-partner violence.

Findings

Political violence was significantly related to higher odds of intimate-partner violence. ORs were 1·89 (95% CI 1·29—2·76) for physical and 2·23 (1·49—3·35) for sexual intimate-partner violence in respondents whose husbands were directly exposed to political violence compared with those whose husbands were not directly exposed. For women whose husbands were indirectly exposed, ORs were 1·61 (1·25—2·07) for physical and 1·97 (1·49—2—60) for sexual violence, compared with those whose husbands were not indirectly exposed. Economic effects of exposure were associated with increased odds of intimate-partner violence in the Gaza Strip only.

Interpretation

Because exposure to political violence is associated with increased odds of intimate-partner violence, and exposure to many traumas is associated with poor health, a range of violent exposures should be assessed when establishing the need for psychosocial interventions in conflict settings.

Funding

Palestinian National Authority, Core Funding Group, Program in Health Disparities Research at the University of Minnesota.
So the study, meant to demonize Israel, was partially funded by the Palestinian National Authority - and the samples were chosen by the Palestinian Central Bureau of Statistics, which is paid by the same Palestinian National Authority.

Anyone see a problem here?

This is not theoretical. It appears that the PCBS chose people for the study that were not close to a random sample of Palestinian Arabs in the territories. Look at this table, entitled "Exposure to violence perpetrated by occupation forces or settlers in 2005:"
3% of the surveyed people claim to have had their homes demolished by Israel in 2005, and 6% to have had land confiscated. 8% claim to have had their home broken into by either IDF soldiers or "settlers." 1% of the husbands were "made fugitives." Almost 1% had family members killed by Israel in 2005!

These numbers are so out of whack with reality as to be laughable - but for the "social scientists" who only hear about house demolitions and land being stolen, they seem eminently reasonable. For all of the talk in the study about "adjusted multinomial logistic regression models" used, none of them looked at these numbers and concluded that it seems a bit strange to think that 15,000 adult men became "fugitives" in the territories in 2005.

So either the PCBS skewed the "random samples" in a huge way, or we have a large number of survey respondents who are simply liars in claiming to be direct victims of Israeli violence in 2005.

And people who are likely to lie in that way are also people who would blame all of their problems on others - and, very likely, the same kind of people to take their frustrations out on their wives.

But the Lancet study cannot imagine such a scenario. Instead, it relies on equally flawed previous studies and a healthy dose of political correctness to say things like
The feminist perspective is relevant to understanding the occurrence of intimate-partner violence because patriarchal ideologies and institutional practices underpin violence against women.28 Pre-existing gender inequalities are exacerbated and traditional gender roles are challenged in environments in which forms of collective violence persist.3 Occupation policies and interactions with occupation forces entail continuous humiliation for men and renders them unable to protect and provide for their families, potentially leading to frustration and violence against people with less power—namely, women and children.26, 29 From a resource-theory perspective, violence might be used to reassert men's socially established position of power in the family.30
And this:
From a psychological perspective, the frustration encountered in living under the control of the Israeli occupation could lead to aggression via negative affect.31 Various negative mental health sequelae have been associated with exposure to political violence in the occupied Palestinian territory32 that are also associated with an increased risk of perpetrating intimate-partner violence, such as depression and post-traumatic stress disorder.33 Humiliation, which takes place frequently in the occupied territories,32, 34 is associated with depression;35 hence, it might also be a mediator of the relation between exposure to political violence and intimate-partner violence.
The two footnotes that supposedly talk about how frequently Arabs are humiliated are a bit suspect. Footnote 32 refers to this book that has no chapters on humilation. Footnote 34 refers to a study of 10th and 11th grade students in Ramallah by Bir Zeit University that may mention the frequency of "humilation" among that sample but whose purpose was not to measure that number.

There is a more fundamental problem with the very way that the study was framed. It looks at "exposure to political violence" as a purely passive phenomenon. The Palestinian Arab men are judged, a priori, as being innocent victims of Israeli violence, who had done nothing on their own. The 1% of the sample who were "made a fugitive" or the 2% who were "detained" are assumed to be exactly the same psychological makeup as those who are more indirectly affected by Israeli policies. In fact, those who are directly exposed to such "political violence" are the ones who are much more likely to be the ones who cause political violence, i.e., members of terror groups. It is not surprising that people who are more violent towards Israelis would also be more likely to be violent towards their wives. Yet the study didn't even consider this very obvious observation!

In other words, this study is worse than worthless. It obscures more than it reveals and it unintentionally shows how a pre-existing bias can, and does, skew science and statistics. It is a case study in poor research techniques. It proves that social scientists can easily find the answer that they are predisposed to find, rather than seek the truth.