<< News Scan | Main | Exclusionary Rule >>

Research Notes


Broken Glass. Researchers at Auckland U. in New Zealand are studying forensic applications of fragments of broken glass on perpetrators, according to this report in the Dominion Post.

UK Clearance Rate. "Only one crime in 39 leads to a conviction, according to a startling Home Office study," says this report. This is far lower than the official clearance rate because so many crimes are unreported.

Steroids Cause Fraud? Most studies in criminology are correlational. That is, they study two variables out in the general population to see if they tend to go together. When a correlation is found between factor X and crime rate Y, we typically see a rush of dilettantes who don't know any better all proclaiming that the study proves that X causes Y. This fallacy is so basic yet so common that I call it the Fundamental Fallacy of Social Science.

Once in a while, though, the causal connection is so implausible (in researchspeak, "lacks facial validity") that everyone has to recognize something else is at work, right? Swedish researchers have found a correlation between steroid use and fraud, but not violence. While it is a plausible hypothesis that hormones could have an effect on impulsive crime, no one could jump to the conclusion, on a correlation alone without other very compelling evidence, that they cause an inherently cognitive, premeditated crime like fraud, could they? The Forbes story on the research is headlined "Anabolic Steroids May Boost Crime Rate."

The actual journal article reporting the study, of course, disclaims any proof of causal connection. Arch Gen Psychiatry. 2006;63:1274-1279


Doesn't everyone know that increased ice cream eating causes crime?

Indeed, the supposed link between anabolic steroids and violence is even weak. The highest correlation between violence, crime, etc. is past history of such an occurance (whether we're talking about an individual or a group). This has frustrated many researchers becasue this is considered a "static" factor that cannot be targeted for intervention (and leaves little room for proposing research porjects with big budgets).

IMHO, the strong correlation between past violence and future violence provides an obvious target for intervention: incapacitation with prison sentences and, in extreme cases, execution. James Q. Wilson noted this at some length in the 1970s in "Thinking About Crime," and sure enough, it worked. Even the soft-on-crime crowd (at least the more intellectually honest among them) had to admit that tougher sentencing was a substantial factor in the crime drop of the 90s.

I'm not sure if you intended it, Steve, but I love your Jabberwockian new word, "porjects."

Leave a comment

Monthly Archives