I take issue with a statistic quoted from Congressional Quarterly's 2007 publication 'City Crime Rankings' in your Naturejobs Regions feature 'Almost in bloom' (Nature 452, 122–124; 2008).

This statistic does a disservice to St Louis in ranking it as the second most dangerous city in the United States. Congressional Quarterly's comparisons are unreliable because they are based on the FBI's Uniform Crime Reports rankings, which are affected not only by the size of a city's population but also by that of the surrounding suburbs.

As the crime rate equals the number of police-recorded crimes divided by the city's residential population, cities with a small population relative to the surrounding suburbs will have inflated crime rates. The city of St Louis has about 10% of the metropolitan-area population (about 90% of the population lives in the suburbs), whereas the city of Memphis, for example, houses more than 50% of the relevant metropolitan-area population. Because assaults on suburbanites that happen in the city are included in the numerator but the victim is not included in the denominator, the crime rate of St Louis is inflated relative to that of Memphis.

If crime-rate comparisons are limited to metropolitan areas, the rankings can change dramatically — the position of St Louis moves to 120th (see R. Rosenfeld and J. Lauritsen Contexts 7, 66–67; 2008). Understanding the actual situation is important because, as noted in your feature, the way is then open for St Louis to become “a hip, feisty biotech hub”.