Tim Lambert (deltoidblog AT gmail.com) is a computer scientist at the University of New South Wales.
Paul Blackman wrote:
"Just a reminder that Pim's reliance upon Killias for showing a relationship between gun ownership and homicide is undermined by Kleck's critique of Killias for some deliberate distortions of data, inappropriate exclusions of some countries versus others, etc. Pim doesn't accept Kleck's refutation, and it seems pointless to recount it, but Kleck would find the data unimpressive."
At the heart of Kleck's "refutation" is a statistical error that a first year undergraduate should not make. I am surprised that someone with Dr Blackman's undoubted skills did not notice. Kleck computed the Pearson correlation coefficient without first excluding outliers.
Anyway, I note that Killias has repeated the study using the additional data obtained in the 1992 International Crime Survey.
He obtained similar results as before: Household gun ownership was associated with more gun homicides and gun suicides, and not associated with non-gun homicides or suicides.
