<< Blog Scan | Main | Why Holder Is Wrong >>

Crime Rate Confusion

This AP story says, "The Justice Department says the rate of violent crime in the United States held steady in 2008." The Bureau of Justice Statistics press release says largely the same thing. Going to the actual report, however, we find the statement, "The violent crime rate in 2008--19.3 victimizations per 1,000 persons age 12 or over--was statistically unchanged from the previous year's estimate of 20.7 per 1,000 persons." (Emphasis added.) The last column of Table 1 does the math for us and tells us that 19.3 down from 20.7 is a -6.9% change. So what is this "statistically unchanged"?

The lack of an asterisk by the percent change figure in the table indicates that the "Difference is [not] significant at the 95%-confidence level." That is, we cannot say with 95% confidence that the change is real change as opposed to random sampling error.  Does that mean we can say that the rate is unchanged? No!  A limitation on our confidence that X is true does not equal confidence that X is false. The chances are considerably better than even that the rate did indeed drop. The slightly larger property crime drop of -8.1% with the same sample was deemed "statistically significant," so the violent crime drop was probably close to, but not quite at, the 95% confidence mark. This illustrates once again why confidence intervals should be stated expressly, not just indicated as meeting or not meeting the 95% rule of thumb. Different confidence intervals may be deemed "good enough" for different purposes.

Meanwhile, at SL&P, Doug Berman asks for debate regarding implications of "the fact that crime is now at record low levels in the US...." Sorry, but once again "what we know for a fact ... just ain't so."  Crime levels are the lowest they have been during the time that the National Crime Victimization Survey has been operating, since the early 70s. However, the FBI data tell us that violent crime per capita is still close to triple what it was when the UCR began in 1960. In California, we have data back to 1952, and violent crime was another 1/3 lower in '52 than in '60.

We have made great progress in bringing crime levels down from the horrific peak of the early 90s. There is near universal agreement among knowledgeable people that tougher sentencing was part of that drop, while debate rages over how large a part.  We cannot rest on our laurels, though. There remains much to be done.

Leave a comment

Monthly Archives