Joel Jacobsen

The year's end always brings stories about crime stats.  The New York Daily News has a story headlined "New York started with 8 murders in 1887."  If you're tempted to believe that number reflects the real homicide rate that year, they were building a bridge for your purchase. 

It might seem relatively easy to count up homicides in a densely-populated urban center, but the late Eric H. Monkkonen and Roger Lane have both documented the almost comical difficulty in ascertaining 19th century New York's (or, rather, New-York's) homicide rate.  The coroners records don't match those of the police, and neither tallies with the newspapers.

One difficulty common to the era is memorably depicted in Willa Cather's masterprece My Ántonia, with the coroner debating  whether a dead man was murdered with an ax or shot himself.  That seems incredible today, but the people determining cause of death in the 19th century weren't forensic pathologists working in laboratories, or even in laboratory conditions. 

Sometimes the coroner's jury literally held the inquest over the body in question, as in this 1859 case of a well-dressed young woman's body found floating in the water off Brooklyn: "Mr. Bennett, the Coroner, living at Bay Ridge, was notified, and soon arrived at the place.  A Coroner's Jury was impanneled on the beach, with Dr. Francis Mahan as the foreman."  The doctor-foreman, we are informed, "made but a casual examination, sufficient, however, to satisfy him that the deceased had been a mother."  The coroner identified the corpse by leaving it in a room and inviting the neighbors in. 

Lane or Monkkonen (I forget which) documents the way New York authorities classified almost all bodies pulled out of the water as accidental deaths.  There was no such thing as forensic evidence in those days.  Sherlock Holmes was cutting edge - it's a testament to Conan Doyle's skill as a writer that we don't even notice the gee-whiz technology that was a major part of the stories' original appeal.  (Though I prefer the Major Gerard stories, myself.)

Because 19th century prosecutions depended almost entirely on testimony or confessions, if you didn't have either of them it was little more than guesswork to return a verdict of death by homicide.  That's what the figure of 8 murders in 1887 means - 8 unattended deaths were, in the modern parlance, cleared.

Steven D. Levitt's Freakonomics blog at the New York Times has, with no indication of irony, recently been harping on the tendency of "the media" to misrepresent crime statistics.  When the Times congratulates itself for not being part of the media, you know something is up. 

While obviously it is unalloyed good news that New York's homicide statistics continue to drop, nonetheless you'd think someone who bills himself - or, more likely, passively went along with his publisher's marketing department billing him - as a "rogue economist" would want to give a little thought to what the statistics are measuring.

As I've pointed out many times, the lethality rate of knife and bullet wounds has dropped by two-thirds just since 1960.  (See post 34 and post 118.)  I ran across a particularly vivid illustration of this phenomenon the other day:

As Geddes approached the white car, expecting then to fight defendant, the driver side door opened. Geddes then heard a gunshot. Geddes saw defendant get out of the car, and he realized he was being fired at. The initial shot missed Geddes, but as he turned to flee, a second shot hit him in the buttocks and he fell to the ground. Geddes was hit twice more, and then he turned over on the ground and saw defendant approaching with the gun. As Geddes testified, "[defendant] was walking up on me, stood over top of me, [and] started shooting me."

 As Geddes further recounted, defendant, standing over Geddes, then continued to fire at him, hitting him several times, in the stomach, legs, arm and hand. Meanwhile, Geddes tried to roll away and cover his face from the gunfire.

The treating physician confirmed the obvious, that "without emergency treatment, Geddes would have died from his wounds."   The doctor counted eight bullet wounds.  It wasn't that long ago that eight close-range bullet wounds were not survivable.

The point is that the homicide rate is not the same as the rate of potentially-lethal violence, and the second is in many respects (though obviously not to the funeral attendees) the more important figure.  The great medical journalist Atul Gawande described this phenomenon from another perspective two years ago in the New England Journal of Medicine:

Combat deaths are seen as a measure of the magnitude and dangerousness of war, just as murder rates are seen as a measure of the magnitude and dangerousness of violence in our communities. Both, however, are weak proxies. Little recognized is how fundamentally important the medical system is — and not just the enemy's weaponry — in determining whether or not someone dies. U.S. homicide rates, for example, have dropped in recent years to levels unseen since the mid-1960s. Yet aggravated assaults, particularly with firearms, have more than tripled during that period. The difference appears to be our trauma care system: mortality from gun assaults has fallen from 16 percent in 1964 to 5 percent today.

We have seen a similar evolution in war. Though firepower has increased, lethality has decreased. In World War II, 30 percent of the Americans injured in combat died. In Vietnam, the proportion dropped to 24 percent. In the war in Iraq and Afghanistan, about 10 percent of those injured have died.

It's foolish to think, as Leavitt seems to, that better results from emergency medical care necessarily means that America is becoming significantly less dangerous for the vulnerable.

Vulnerability is the other obvious variable Levitt seems oblivious to.  Not all Americans are equally vulnerable to criminal violence.  For instance, this Bureau of Justice Statistics report says that in 1993, the national violent crime rate per 1,000 persons was 50.  By 2005,  the number had dropped by 58% to 21.  Exceptionally good news....

... until you look at the violent crime rate for those living in households with incomes under $7,500 - the destitute.  Their violent victimization rate in 2002-03 was 47.4 - almost the same as the national bad-old-days figure.  Still, by 2004-05 the victimization rate for the poorest had dropped dramatically to 38.1, just 55% above the national average.   

So does that mean it's just taking a little longer for increased safety to trickle down? Unfortunately, the victimization rate for those earning $7,500-$14,999 - not exactly Easy Street - increased by 5.6 over the same period.

The drop in crime rates heralded by Leavitt is, to an uncomfortably large extent, a drop in the rate at which middle class and wealthy people are victimized.  Which is a good thing, of course.  But ....

A lot of the year-to-year change is just random variation, like talking about the average temperature for this time of year.  Only long-term trends mean very much, but a comparison of long-term victimization rates needs to take into consideration changes in people's behavior over the same period of time.   Do people live the same way today as they did in 1960?  1970?

Just think for a moment about car alarms.  Or cell phones (very handy for calling 911 - another reason for the falling lethality rate).  Or gated communities.  Or security guards escorting you to your car in the locked parking garage.  Or indoor malls.  Or hitchhiking.  Or women other than prostitutes walking around alone at night.  Or metal detectors in airports, courthouses, even high schools.  Or elementary school lockdowns.   Or ...

We live much differently today than our parents and grandparents did.  Many of the changes are welcome signs of progress.  But not all.   One of the most significant reasons for the falling crime rate in America is that those of us who can afford it spend so much time, money and energy protecting ourselves.  That's one reason the crime rate has dropped much faster for the well-to-do: they can pay for the protection the state no longer provides. 

The responsibility for fighting crime has, to an underappreciated extent, been shifted from the community to the individual.  Now, why doesn't Freakonomics figure out some way to quantify that?