Barbelith Underground > Switchboard > 'You could abort every black baby and crime would go down'

Well, yes and no. That is, yes, he's showing that you can prove things with statistics which are highly selective in the information they provide. The idea that all the children who had been aborted would have contributed nothing but cash to the economy, without there being any other impact on the economy from millions of head of extra population is a false proposition which uses a limited statistical viewpoint (where 1 person = a set addition to federal funds).

However, the example he provides does not map. First up, because the reductio ad absurdum here is not aborting black people, but aborting people. Using that statistical model, any reduction in population results in a corresponding reduction in criminality, as a certain number of the people not existing through abortion would be criminals and, since they have not been born, are now not criminals. As a statistical model, blackness or whiteness doesn't come into that.

So, the statistical position here depends on the idea that, ceteris paribus, a black person is more likely to commit crime than a white person. That's a statistical nonsense, because if we assume that the currently aborted in the previous example functioned only as economic positives, we also have to assume that the currently unaborted in this example function only as criminal positives - that is, again, that their colour is statistically irrelevant.

So, we go from pure statistics into the much murkier world of applied statistics. At which point the model becomes highly complex. What is crime? Would the specific abortion of all black people reduce it? Are we counting white-collar crime, for example? It's pretty clear that Bill Bennett was _not_ - his subsequent explanation that he was thinking about New Orleans demonstrates that. he was, in fact, thinking about the sort of violent crimes against property and person about which wealthy, white America seems to be disproportionately concerned, and associating those crimes with blackness rather than, say, with being stuck in the middle of a destroyed city.

So, Bennett's statement is bad statistics, and also bad broadcasting. He is saying "you say that that can be proven with statistics, but this too can be proven with statistics, and it doesn't mean that there is a compelling social argument on those grounds either for banning abortion or for making abortion compulsory for black people". However, he introduces the idea of truth, which is foolish, and also he associates crime with black people, which messes with his proposition of statistical accuracy as, as has been mentioned above, any reduction of population portrayed in those terms is likely to see a drop in crime - therefore, it is neither true nor statistically complete to select black people, and to do so shows an instinctive identification of crime as illegal stuff that black people do.

Since his first criticism of the statistical argument for banning abortion was that it was not statisticaly complete - on the grounds that many of those aborted would be the children of single-parent families, and thus would actually be a drain on funds rather than a boost to the economy (which seems to have been ignored in the larger brouhaha), to add a specific qualifier about race to the Freakonomic example is a rather odd decision.