29 people have been murdered in New Haven this year. This article claims that this is the highest count since 1994 and posits that the bad economy is to blame. A statistician would complain about the small sample but the broader point about recessions and crime (due to declining youth labor market opportunities) has been explored by Richard Freeman.
|
A statistician would complain about the small sample
Perhaps, but if they know anything about statistics they would be more likely to complain about multiple comparisons.
Apparently I don’t know anything about closing tags.
Lars: This is an awfully small sample to justify that inference, isn’t it?
Sample size isn’t the problem with that inference. It might be a significant increase in the murder rate (we aren’t presented with enough information to know). Even if it isn’t, I’d be willing to bet there are cities in the US who’ve experienced significant increases. There are, of course, also cities who’ve seen decreases. A social scientist or even an economist should know that looking at a single city doesn’t tell you anything, and this is a different conceptual problem than sample size.
There’s nothing inherently wrong with statistics and probability as tools. That you can accurately estimate things about the whole U.S. with a sample of less 2000 individuals is a pretty amazing thing, but there’s something seriously wrong with your epistemology, if that gets turned around to thinking you cannot learn anything “significant” from the murders of 29 people in a small city. There’s a depth of detail available, which can constrain interpretation, and the article seems to have located someone, who can make use of that data to bolster his interpretation. I applaud Professor Kahn for pointing to Freeman’s analysis, which makes use of a fuller context and depth of detail.
Matthew: “A statistician would complain about the small sample”.
Actually, no. Hint: “highest count since 1994 “. Not conclusive, but important. The article, of course, did not give the counts by year.
Matthew, you might want to make a habit of consulting with actual experts before writing on things. UCLA should have numerous experts in statistics, political science/public policy, and economics.
“I’d be willing to bet there are cities in the US who’ve experienced significant increases. There are, of course, also cities who’ve seen decreases.”
Which is why one looks at broad trends first, and also to causal factors at lower levels.