« Reports on possible SCOTUS replacements | Main | Supreme Court on death-qualified jurors »

New book by John Lott

Just in the mail: "Freedomnomics: Why the Free Market Works and Other Half-Baked Theories Don't." It is quite interesting and a good read, in part an answer to an earlier book "Freakonomics." These two books seem intended to remove economic's popular reputation as "the dismal science," by showing how statistical work can be applied to matters of popular interest. The key to that is often having the imagination to figure out the beginning data, where do you start comparisons?

A few of the most interest points (to me, anyway):

Interest groups contribute to campaigns of politicians who favor them. Does that prove politicians favor them because of their contributions, or rather that they contribute to politicians who favor them anyway? Which is cause and which effect?

Lott takes the voting records of 700+ House members, and asks what happened after they announced they were retiring? Contributions of course fell off by 85% -- they're not going to campaign again. But their voting record stayed the same. Even with no need for campaign funds or other support, they voted the same. The contributions didn't create their positions; their positions lead to contributions, is to say the call for "campaign reform" is badly undercut.

Campaign finance reform: the main effect is to level the playing field. Except that the playing field is already heavily stacked in favor of the incumbent. So the re-election rate goes upward wherever these measures are adopted.

A bit of history: private radio stations initially seemed doomed to failure, since nobody could figure out to make it pay. How can you charge for subscriptions, when anyone can listen in? For a time it looked as if government radio was the only possibility. Then AT&T discovered they could sell ads, and make it pay without dues or taxes.

Crime rates: a mystery to many was why rates began rapidly falling in the early 90s. Freakonomics attributed it to abortion -- potential criminals were being terminated early and the population of them was smaller. Freedomomics argues this was a bad analysis of the data, and denies the conclusion. It argues that increasing the number of executions depressed homicides, for one thing. (It points out that the odds of a policeman being murdered is less than the odds of a killer being executed, yet that is enough to influence officer decision -- to wear body armor, be cautious, etc. -- why doubt that the odds of execution influences killers?) Another factor was increased police on duty. Plus right to carry laws, and here Lott answers Freaknomomic's attack on his conclusion that right to carry reduces crime rates. (He also cites data by other economists suggesting that when right to carry is enacted, and crime rates drop, they often go up in adjacent counties in nearby states that didn't enact it, suggesting that the criminals moved, or relocated their crimes).

He has an interesting chart on studies of right to carry and its effects. Basically, 15 studies conclude right to carry reduces violent crime. 10 studies conclude it had no discernable effect. Zero studies conclude it increased violence.

All in all, a very great read, and demonstration of how statistical work can illuminate matters of great policy concern to us all. Oh, and here's the Amazon link to buy (I suspect it'd be useful if its Amazon ranking shot up quickly -- best-seller data isn't available for a while, and Amazon ranking is often used as a short term substitute):

· Crime and statistics