In a post last month for Demos, Matt Bruenig argued that if one in fact cares about childhood poverty, a recent conservative position promoting marriage as a means of reducing childhood poverty rates is as cruel as it is misguided (http://www.demos.org/blog/4/14/14/single-mother-child-poverty-myth). In a nutshell, Bruenig notes that countries with low childhood poverty rates have not achieved this success via promoting marriage, nor are their rates of single-parenthood markedly different from those in the U.S.
One way to put this basic observation is that people promoting marriage are deliberately ignoring most of the variation that matters. Yes, it is true (ceteris paribus, of course!) that in the international sample Bruenig looks at, in every country, the children of single parents are more likely to be poor than the children of married parents living together.* But far more variation in childhood poverty rates is associated with different policies in the different countries, and not with the relationship status of the parents; indeed, even if the child poverty rate in the U.S. dropped to the level associated with children living with married parents living together in the U.S. today, it would still be substantially higher, indeed, quite grossly higher, than in the other countries Bruenig looks at.
When thinking about a policy decision, does it make more sense to look at the sources of variation within a population, at a particular time, or does it make more sense to look at the sources of variation between populations, or between different times? (One could, if one wished, put all this in terms of ANOVAs, but that seems a little much in this context…) The question comes up, I think, surprisingly often. And the focus on variation within populations, and the exclusion of variation between populations, seems a particular common and often problematic approach.
In the chapter of my first book (The Limits and Lies of Human Genetics Research) titled “Criminality and Violence: The Brain and the Gene,” I made a similar argument. Some researchers, noting that in the cross-cultural samples in question around half of all violent crimes are committed by a small percentage of young men (5-10% are the usual numbers cited), made claims about the possible gains that could accrue to being able to recognize these young men before they committed their crimes, given some perhaps misguided hope that those young men were particularly disposed to violence and also might be marked genetically or via some other biological markers for that predisposition. But I suggested, along with many others interested in these issues, that even if this were possible, given that the rates of violent crime varied by at least an order of magnitude (and perhaps by nearly two orders of magnitude!) between the areas studied (and over time within particular areas), this was surely the wrong way to look at the problem. If we actually cared about preventing violent crimes, we would surely better off looking to the differences in rates between the areas, and over time, than within them.
Similar arguments could be made regarding performance on IQ tests; indeed, similar arguments have been made. (For a nicely sarcastic one, see: http://mikethemadbiologist.com/2011/12/29/a-modest-yet-brave-proposal-alabama-whites-are-genetically-inferior-to-massachusetts-whites-for-realz/ )
Bruenig suggests, somewhat provocatively, that those who argue that marriage is a good solution to childhood poverty must in fact hate children (they must in fact prefer a state of the world in which there are more rather than fewer children suffering the effects of poverty), given the overwhelming evidence that marriage is not a good solution, and the overwhelming evidence in favor of other solutions as effective. Put (somewhat) less provocatively, one might note merely that those promoting marriage as a solution at least don’t really care about reducing childhood poverty rates; whether they in fact hate children or not, they clearly have no real interest (understood as a willing to do anything about it) in ensuring that children don’t grow up in poverty. One might argue that a focus on marriage and marriage success rates, as something we are unlikely to change meaningfully via any practical policies, is itself a defense of doing nothing. But the first defense of doing nothing here is to focus on the (smaller) within population variation; if we don’t even look at other systems, we don’t have to admit that change might be necessary.
The goals of promoters of genetic explanations (within populations / environments) for e.g. the so-called “gap” between the IQ test performance of Blacks and Whites within populations are, I think, all too clear; the explicit defense of the status quo is too glaring to ignore. And what do promoters of biological associations with violence get out of ignoring the variation associated with different times and places, and focusing only on the variation within particular populations at particular times? If violence and criminality are caused by the genetic predispositions of a small coherent of young men, then what could be gained by looking to other ways of organizing our worlds?
A final note, made in passing. With the publication of Wade’s A Troublesome Inheritance: Genes, Race, and Human History, arguments about the role of the (small) genetic differences between populations in explanations for the (large) differences in various outcome measures between those populations seem set to explode again. But which differences are being “explained” in these cases, and which differences ignored? And where the latter are often much larger than the former, what work is being done by shifting focus to some (smaller) differences, and away from other (larger) differences?
* Ignore here the obvious and well-support criticism that since financial troubles are bad for relationships, the causal arrow often, or usually, points from poverty to single-parenthood. This is true, but Bruenig is making a different point here.