Yesterday, I wrote about crimes that don’t get counted in the official statistics and people who don’t get to live decent lives because powerful people work so hard to deny the terrible daily impact of crime.

A new anti-crime ethic is percolating in the neighborhoods and on-line.  This ethic, however, is being slandered because it flies directly in the face of the tired old excuse-making and crime-downplaying that has long been the status quo among politicians and criminologists.

One of the ironies of this status quo is that people who study crime seem intent upon denying its existence.  Every year, the FBI releases raw data on crime reports, and then, inevitably, like the swallows returning to Capistrano, a great cry emerges from the hallways of sociology departments:

Crime records are meaningless!  We need to understand criminals, not punish them!  Shame on people who worry about crime!  They’re hysterics, “perceiving” danger that does not really exist.

Well, OK, that’s a little exaggerated. Criminologists don’t shout.  They sit quietly at their desks and wait for journalists to call them so they can discuss their latest modeling techniques that show that crime simply is not the problem that the great unwashed public believes it to be.  And this is as predictable as the Capistrano swallows: not a season passes without some new statistical effort to “adjust” crime rates downward.

A great deal of the statistical jostling that occurs at the release of the Uniform Crime Reports has to do with the ranking of cities in terms of relative danger.  Nobody wants to be #1 in homicide, after all.  Nobody wants to be #1 in any crime rate, which is why the criminologists get crunching and the politicians start working the microphones as soon as the new UCR hits the streets.

It’s a cynical game, and a dangerous one. As I said yesterday, no politician should ever be allowed to declare victory over crime so long as people have to cover their front doors with metal bars in order to keep criminals from kicking the door down.  Criminologists in particular have a lot to answer for in their quest to excuse offenders, oppose incarceration, and downplay crime prevalence — more or less the official mantra of the profession since the 1970’s.

One of the more curious efforts at adjusting crime rates downward originates at Georgia State University, where Professor Robert Friedmann has gained national attention for arguing that crime data should be adjusted based on the number of poor, unemployed, black, and single-female-headed households in the city in question, a measure he equates with “social and economic disadvantage.”

Here is how Professor Friedmann’s formula works.  In 2007, Atlanta ranked 8th in the nation among large cities for prevalence of homicide based on population and the number of homicides reported to the Uniform Crime Reports.  Professor Friedmann took that raw data and “factored in” census information about the percentage of poor people, black people, unemployed people, and single-parent, female-headed households in Atlanta compared to other cities.  When you factor in these characteristics in a particular way, he argues, Atlanta is no longer the 8th most dangerous city in America: it drops to number 43.

Here is his mathematical model for “adjusting” Atlanta’s homicide rate:

The statistical model used to estimate adjusted homicide rates for the 63 large US cities was specified as follows:

Homrate = a + b1(Poverty) + b2(MdInc) +  b3(MaleUnem) + b4(Black) +  b5(FemHead), where


Homrate = Homicides per 100,000 city residents;


Poverty = Percentage of families with incomes below the poverty line;


MdInc = Median household income;


MaleUnem = Percentage of males age 20-64 unemployed;


Black = Percentage of the population black; and


FemHead = Percentage of families with children under the age of 18 headed by a female.

I am deeply uncomfortable with this theory: I think it presumes a certain level of pathology among minority populations and then measures nothing more than the deviations from this micro-engineered pathology rate.

Here are the before-and-after top ten rankings for 2007 (go here and click on table-1 for the full list):

  • Highest unadjusted homicide rates: Detroit, Baltimore, St. Louis, Newark, Washington D.C., Oakland, Philadelphia, Atlanta, Cleveland, Memphis.
  • Highest adjusted homicide rates: Newark, Baltimore, St. Louis, Oakland, Phoenix, San Francisco, Alburqueque, Washington D.C., Miami, Tulsa.

The clear winners of Dr. Friedmann’s approach are Atlanta, which drops from 8th to 43rd place, and Detroit, which drops from 1st to 23rd (well, prior to this week’s revelation that they under-counted murders by some 25%).  So, Atlanta alone benefits from this approach, at least in an academic sense.

At the cost, one might say, of a quantification of the soft bigotry of lowered expectations.  Instead of measuring incidents of crime in order to arrive at the crime rate, Friedmann measures the race of the population and then contextualizes crime within race.  Even murders committed by whites are reflected through the presence or absence of a black population, and while I cannot imagine that being the intent of this exercise, it is an unavoidable result.

Instead of blaming crime on the criminals who commit crime, Friedmann blames socio-economic “crime producing” factors.  He blames everybody, in practice, with the exception of the criminals themselves, who are partially acquitted of blame in the process.  This shifting of blame from “criminal actors” to “everyone” has been the main project of criminology since at least the mid-Seventies, which is probably why a theory like Friedmann’s didn’t raise more eyebrows, when what he is saying may actually be paraphrased: “Well, you must expect a certain amount of murder from poor, jobless, fatherless black people.”

One columnist in San Francisco expressed wry amusement that their 88 slayings placed them higher on the homicide list than Detroit for 2004 (the figures above are 2007):

A new study by the Improving Crime Data Project shows that San Francisco had the highest homicide rate among the 67 largest U.S. cities in 2004, when our fair burg racked up 88 slayings. In effect, the study posits that with our generally boffo quality of life — think fabulous ocean views and Gavin Newsom’s hair — we should have far fewer murders. Instead, we’re the new Baltimore. . .

“That’s the price you pay for living well,” quips criminologist Alfred Blumstein, a professor at Carnegie Mellon University in Pittsburgh, who worked on the project with Friedmann and Richard Rosenfeld, a criminal justice professor at the University of Missouri-St. Louis. Blumstein suggests that, given our relatively high median income of $55,000, low unemployment rate, and stable residential base, the number of homicides ought to decline. “It’s a problem you should be able to do something about.”

Detroit, for its part, was delighted to be knocked from #1 to #23 for 2007:

If you consider Detroit’s socioeconomic status, it’s not the deadliest city in the country after all, according to rankings released this week by the Improving Crime Data Project.

This was, of course, before the other bodies showed up off their books.  But there’s the rub, and it is far from amusing: all those people murdered in Detroit in 2004, and 2005, and 2006 and so on did not stop being dead just because some criminologists found a sophisticated way to justify away Detroit’s high rate of crime, or some police official found a clever way to avoid recording murders in which the victim lingered in I.C.U. for two days before dying.

All the homicide victims in Detroit would still be just as dead if the standard of living in San Francisco were to suddenly plunge below that of the Motor City.

You can play with statistics all day long, but crime is still out there, bearing down on people like that girl and her grandmother living in the war zone that is Thomasville Heights.  If any lesson is to be learned from the existence of studies that add “socio-economics” to crime statistics, it is that we have a greater emergency on our hands, not a lesser one.  If children are being born and raised in environments so destructive that it must be presumed that they will be far more likely to kill or be killed, that ought to be cause for alarm, not quiescence.

Join the conversation: