Tim Lambert (deltoidblog AT gmail.com) is a computer scientist at the University of New South Wales.
What on earth do you mean by 'the "nothing else happened" parameter"?
Andy Freeman said:
Lambert's model is for a transition between two stable situations with some "noise". He uses it to argue that gun control explains the transition. Yet, he doesn't bother to show whether or not anything else happened at the relevant time, whether or not the situations were in fact stable, and so on.
You call this a "parameter"?? This is a bizarre usage even by your standards. Your claim that I have not shown that the situations were stable is false. The homicide rate was roughly constant in the period before gun control and in the period after gun control. Whatever effects other factors may have had on the homicide rate in these periods are lost in the noise.
I do not claim that the data proves that gun control caused the change, but it is more plausible than any other explanation offered so far:
In the book from which I extracted the original data, Wallace suggests that the change was caused by a decline in infanticide. Indeed, there was a decline in infanticide, but not sufficient to explain the observed change -- i.e. if we exclude infanticides from the data the drop is still statistically significant.
It might be that demographic change caused the decline i.e. a decrease in the percentage of young men in the population. However, the demographic change associated with WWI when 40% of the males 18-45 enlisted, is far far larger than any other change during the period, and the homicide rate 1915-18, while lower than the surrounding years is still higher than the post gun control rate. Furthermore, correcting for the demographic effects of WWI destroys the "gradual decline" model -- it does not fit the adjusted homicide rates at all well.
Perhaps changes in unemployment and poverty caused the decrease. This seems unlikely since the Depression does not seem to have affected the homicide rate at all.
Maybe the decrease was caused by increasing urbanization. This seems unlikely since urbanization increased gradually over the period 1900-1970, not abruptly around 1920.
If some other factor is supposed to explain the decrease, it must explain why the rate did not decrease in the adjacent states of Queensland and South Australia until they introduced gun control (in 1927 and 1929 respectively). It must also explain why the rate for homicide by "other means" did not decrease significantly after 1920, but the firearm homicide rate did.
"Ahem" indeed. The data isn't complete, so its "fit" is basically meaningless. For one, the step model explanation is inappropriate without "nothing else changed" data, which hasn't been presented.
Gee, Andy what has changed your mind? On Wed, 18 Dec 1991 Andy Freeman said:
Why doesn't Lambert tell us about pre-control crime and murder rates and trends in Oz and compare them to post-control rates and trends? If gun control actually worked in Oz, the introduction of controls was associated with a good change in the rate trend. Could it be that there wasn't a good change associated with the introduction of controls?
I provide the data that Andy requested and now he tells us that it is meaningless unless I provide data about "everything else" that happened. If you felt that the data you asked for was meaningless, why did you ask for it? Could it be that you consider data to be "meaningless" only if it does not support your beliefs?
