Wednesday, May 26, 2004


Tim Lambert is a computer maven at the University of N.S.W. -- where I spent most of my teaching career many long years ago. He seems to have an obsession with catching out John Lott Jr. on gun-usage statistics and he has an occasional blog which seems to be mostly devoted to that. He does also however takes some interest in other scientific issues such as climate change. He has sent me copies of several of his past posts that I have not linked to because I thought that they were too intemperate. Not much light is generated by arguing with an angry man, in my view, and I think all conservative bloggers have to get used to ignoring rage-filled emails from Leftists. Leftists are very good at rage. It seems to be their principal emotion. I did nonetheless address on this blog what substance I could see in Lambert's various posts. Being a terrible tease from way back, however, I deliberately posted here recently another quotation from Lott. I wanted to see what heights of wrath Lambert might rise to this time. Sure enough, another email from Lambert popped into my inbox shortly thereafter. He seems to be learning, however, as his comment this time is principally factual. So, although I am no expert at all on the matter, I thought I might make a few comments on some of the points Lambert raises -- in case John Lott Jr. does not see fit to do so (I could say more about Lambert's whole post but I think that would be too tedious).

Lambert accuses Lott of selective use of statistics but almost any use of statistics has to be selective so the only interesting question is whether alternative selections of statistics show substantially different results. Lambert presents statistics to show that Australia's 1996 gun-control laws have been beneficial. He disputes Lott's claim that serious crime has risen since then. What Lambert's alternative statistics show to me, however, is more a pattern of no change than anything else. Deaths by firearm are surely the biggest issue but Lambert's table shows that the average rate of murder with a firearm before the bans was 0.37 compared with .30 after the bans -- with the figures in most individual years being .20 plus or .30 plus. Given statistical error and the range of influences which could have affected the averages concerned, the bans would seem to have achieved nothing significant -- a very poor result considering the vast expense in money and the significant loss of liberties associated with the bans. I note too that even the slight difference in averages observed seems to have been largely the result of a single very anomalous year in 1997 -- making the averages used a poor guide to any underlying trend. I think that for trend calculating purposes it would in fact be most appropriate to exclude both 1996 and 1997 -- and when one does that the "before and after" difference becomes very small indeed: .31 versus .28. The differences for other gun crimes also seem to be too small to assert a real underlying difference. And even Lambert admits a lack of a clear pattern when he notes that the "assault-with-firearm rate has increased". So Lott's statistical selection shows ill effects from the firearm ban and Lambert's selection shows no clear effect. It seems to me therefore that NEITHER set of statistics support the ban.

Lambert also weighs in on the global warming debate with a critique of another author who says that you can show either global warming or global cooling according to what set of statistics you use. Lambert makes some reasonable criticisms of the statistics which were alleged to show global cooling and recalculates the statistics concerned in a more orthodox way. His conclusion? "When analysed correctly, their data shows neither warming or cooling, regardless of which average is used". In other words, contrary to the Greenies, the earth is NOT warming up. That certainly fits in with all that I know on the subject! I must quote him on that again sometime.


No comments: