Nation's Safest And Most Dangerous Cities

First of all, Morgan Quitno DOES rank MSAs as well as cities
http://www.morganquitno.com/met05a.pdf

But unlike Ms. Perkins, I do not think this is so useful.

The problem with city crime data isn't that it samples too small a piece of St. Louis's crime, but that it takes too big a piece. Using MSA data, as Ms. Perkins advocates, really tells you nothing about an area's safety because an large MSA ALWAYS contains an incredibly wide range of neighborhoods, from highly dangerous to very safe.

City data is SOMEWHAT better because at least you can compare like to like sometimes: to be fair, you can't easily compare St. Louis to a city like Jacksonville which has annexed most of its suburbs, but at least you can fairly compare St. Louis to other cities with smaller boundaries- not just by comparing St. Louis city to (for example) Washington, DC, but by comparing St. Louis's inner suburbs to those of other cities (e.g. Clayton and University City near St Louis vs. Arlington and Alexandria near DC).

But even city data is flawed because cities contain safe areas and very unsafe ones: for example, most of the north side of St. Louis city is dangerous, but I suspect St. Louis Hills at the city's southern tip would probably compare favorably with much of suburbia.

This flaw could be remedied if crime statistics compared individual neighborhoods; unfortuately, police departments rarely release such data, and often (I suspect) do not even tabulate them in formats that would be useful for someone trying to calculate crime rates.

Morgan Quitno's analysis is also flawed for an entirely different reason. Morgan Quitno ranks cities by overall violent crime. But not all violent crime is of equal public concern.

St. Louis was only 12th in murders. Murders matter because they are the most heavily reported crime; reported rates of other violent crimes are more likely to involve underreporting.

Robberies also matter because they are the most likely to involve strangers.

But a third common category of violent crime, assault, is usually a large part of a city's overall violent crime rate, but is of less concern to the overall public because assaults are more likely than other crimes to involve friends and relatives. So if you are trying to decide where to live, a neighborhood's robbery and murder rates are probably far more useful than overall violent crime rates.