Are the reductions in homicide following the introduction of gun control statistically significant?

Frank Crary said (referring to changes in homicide rate in NSW and Qld):

The sharp drop you claim to see, is (in my opinion) not significant compared
to the background variations. While this may have been real effect of gun
control laws, in is at least as likely that it is a coincidence.

Could the changes be caused by noise? This calls for a t test, to see
if the average homicide rate changed.

The significant events were:
1920: NSW controls all guns
1927: Queensland controls handguns and NSW drops controls on long guns

So, I took three groups of years 1911-1920 (pre-control), 1921-1927
(NSW controls, no Qld controls), 1928-1937 (both controls)

The means:

           NSW  Qld
1911-1920  2.3  4.1
1921-1927  1.4  4.3
1928-1937  1.6  2.7

Probability chance could cause the change:
In NSW between 11-20 and 21-27: 0.0003
In NSW between 21-27 and 28-37: 0.09
In Qld between 11-20 and 21-27: 0.57
In Qld between 21-27 and 28-37: 0.002

Both states recorded significant declines in the homicide rate
following gun control. If some other factor caused the declines, then
it did not operate in both states simultaneously. (The change in NSW
after 1927 is of borderline significance, but it was an increase.)


More like this