[Oberon]

I skimmed it. All it shows is that Lott's results are not robust to changes in specificataion and addition of new data (lot's dataset ended in 1992, this one in 1999).

It may be that the rates of violent crime, after dropping at the initial passing of CCW, stabilized for several years at the new, lower level. This would tend to appear to dilute the crime-lowering effect of CCW, if such exists.

[traditionalist]

It's mainly correlated ommitted variable bias, IMHO. There are so things that affect crime rates that are unobservable and correlated with the presence of concealed-carry laws that any rigorous statistical analysis of this question is close to being meaningless.

[traditionalist]

It may be that the rates of violent crime, after dropping at the initial passing of CCW, stabilized for several years at the new, lower level. This would tend to appear to dilute the crime-lowering effect of CCW, if such exists.

If this were true, then CCW should still be asssociated with lower levels of crime. While this relationship shows up in the 1977-1992 data, it disappears when you add 1993-1999 data.

[CHICAGOFARMER]

You are right on with the dilition theory. What Lott did was to index each state for a time period after the CCW bill was passed into law. This is a more accurate reading of the impact of CCW on each rather than a dilution be averaging accross time which a large study over a longtime will show. An average is just an average of the highs and lows.

For example Dr. Deming, the guru of Japan's resurgance after WWII, said if you have three groups who are averaged you will always have one group above the average and one performing below the average, but all doing well or not well when compared to the top 25%.

Dr Deming is wrong to manage to the average or the diluted mean average. You must manage to these who excell.