<< News Scan | Main | Poll on Sex Offenders >>

The Dangers of Small Sample Sizes

The Violence Policy Center has this study with the surprising finding that the State of Vermont, a very low-crime state generally, had the second-highest rate of women killed by men in 2008.  WCAX story here.

Does this mean anything? Probably not. The study only looks at one year, and there are only eight murders in the category in Vermont in that year.

Anomalous results pop up now and then just at random, particularly when sample sizes are small.  Why are so many recent Presidents left-handed?  Why are so many of the top poker pros named Phil?  Probably no reason but chance.  These are very small numbers of people.

This study would need to be done over many more years to see if there is anything worthwhile here.

Leave a comment

Monthly Archives