Monday, May 28, 2012

The Homicide Narrative: Revised To Incorporate Improvements In Trauma Medicine

Previously we cited
Murder and Medicine: The Lethality of Criminal Assault 1960-1999 -- Harris et al. 6 (2): 128 -- Homicide Studies
Available Here
http://www.wku.edu/~james.kanan/Murder%20and%20Medicine.pdf

I do not presume to take any credit away from these four gentlemen

ANTHONY R. HARRIS
University of Massachusetts Amherst
STEPHEN H. THOMAS
Harvard Medical School
GENE A. FISHER
University of Massachusetts Amherst
DAVID J. HIRSCH
University of Massachusetts Medical School
 
They deserve all of the credit for their original paper which I cite above, and the majority of the credit for any value I might add here in this post.
 
What I have done is take their original methodology and extend their data series up to 2010 (the UCR data is available online presently from 1960-2010).  In addition, I have produced a chart which does not appear in their work.  In this chart I produce an estimate of the homicide rate normalized to the level of medical technology present in 1960, which is our base year.  There are of course some complications that I neglected, for instance, the shift over time in the weapons used to carry out aggravated assault, but this plot should be sufficient for government work.  Or, in this case, it should be adequate to form a rough estimate of the effect of governance.






The raw numbers used are presented at the bottom of the post, if anyone would like to squeeze them further with Excel or similar software to produce more or better insights.  My laziness exceeds my arrogance.
We define, as in the original paper, the lethality rate as the homicide rate/(homicide rate + aggravated assault rate).  This gives us the estimate over time of how lethal aggravated assault is in practice.  As you can see in the data, this ranges from approximately 5.5% in 1960 to around 1.8% presently. 
Lethality Rate in 1960/Lethality Rate is the normalizing factor to adjust homicide to 1960 level medical technology and infrastructure.

What sort of a narrative can we draw out of this?  It looks to me like the adjusted homicide rate rises pretty steadily until a peak in 1992 and 1993, after which it drops off.  Incidentally, the delta between the two curves (actual and adjusted) is a pretty good estimate of the number of people who actually owute their lives to improvements in trauma medicine.  I don't see significant reductions in lethality rate after the year 2000, where the authors originally made their study.  I'm not certain what to make of that.  It strikes me as likely that there has been much more fudging in the aggravated assault rate since then (e.g., the recent stories in Wisconsin), but I don't think that can account for most of the apparent stagnation in lethality rate. 

Were there any justice, politicians would be loathe to pick on doctors as much as they do, considering how relatively good they've made them look (2010 homicide rate approximately the same as 1960 rate, despite around 3x the amount of aggravated assault!).  If you look in the paper's discussion on motor vehicle fatalities, you will see almost exactly the same story.  Nearly all of the improvement in recent years is the result of technology, not governance.

Year Homicide Rate Aggravated Assault Lethality Rate Lethality Rate in 1960/Lethality Rate Adj Homicide Rate


1960 5.1 86.1 0.055921053 1 5.1

1961 4.8 85.7 0.053038674 1.054344846 5.060855263

1962 4.6 88.6 0.049356223 1.133009153 5.211842105

1963 4.6 92.4 0.04742268 1.179204805 5.424342105

1964 4.9 106.2 0.04410441 1.267924275 6.212828947

1965 5.1 111.3 0.043814433 1.276315789 6.509210526

1966 5.6 120.3 0.044479746 1.257225094 7.040460526

1967 6.2 130.2 0.045454545 1.230263158 7.627631579

1968 6.9 143.8 0.04578633 1.221348207 8.427302632

1969 7.3 154.5 0.045117429 1.23945566 9.048026316

1970 7.9 164.8 0.045744065 1.222476682 9.657565789

1971 8.6 178.8 0.045891142 1.218558752 10.47960526

1972 9 188.8 0.045500506 1.229020468 11.06118421

1973 9.4 200.5 0.04478323 1.248705207 11.73782895

1974 9.8 215.8 0.043439716 1.287325456 12.61578947

1975 9.6 231.1 0.039883673 1.402103893 13.46019737

1976 8.7 233.2 0.035965275 1.554862371 13.52730263

1977 8.8 247 0.034401876 1.625523325 14.30460526

1978 9 262.1 0.033198082 1.684466374 15.16019737

1979 9.8 286 0.033130494 1.687902793 16.54144737

1980 10.2 298.5 0.033041788 1.692434211 17.26282895

1981 9.8 289.3 0.032764962 1.706733351 16.72598684

1982 9.1 289 0.030526669 1.831875361 16.67006579

1983 8.3 279.4 0.028849496 1.938371909 16.08848684

1984 7.9 290.6 0.026465662 2.112966356 16.69243421

1985 8 304 0.025641026 2.180921053 17.44736842

1986 8.6 347.4 0.024157303 2.314871481 19.90789474

1987 8.3 352.9 0.022978959 2.433576411 20.19868421

1988 8.5 372.2 0.022327292 2.504605263 21.28914474

1989 8.7 385.6 0.022064418 2.534444949 22.04967105

1990 9.4 422.9 0.021744159 2.571773516 24.17467105

1991 9.8 433.4 0.022111913 2.529001074 24.78421053

1992 9.3 441.9 0.020611702 2.713073005 25.23157895

1993 9.5 440.5 0.021111111 2.648891967 25.16447368

1994 9 427.6 0.020613834 2.712792398 24.41513158

1995 8.2 418.3 0.01922626 2.908576701 23.85032895

1996 7.4 391 0.018574297 3.010668563 22.27894737

1997 6.8 382.1 0.017485215 3.198190789 21.74769737

1998 6.3 361.4 0.017133533 3.263836675 20.56217105

1999 5.7 334.3 0.016764706 3.335641736 19.01315789

2000 5.5 324 0.016691958 3.350179426 18.42598684

2001 5.6 318.6 0.017273288 3.237429511 18.12960526

2002 5.6 309.5 0.017772136 3.146557801 17.62072368

2003 5.7 295.4 0.018930588 2.954005078 16.83782895

2004 5.5 288.6 0.018701122 2.990251196 16.44638158

2005 5.6 290.8 0.018893387 2.959821429 16.575

2006 5.7 287.5 0.019440655 2.876500462 16.39605263

2007 5.6 283.8 0.01935038 2.889920113 16.18355263

2008 5.4 276.7 0.019142148 2.921357212 15.77532895

2009 5 264.7 0.018539118 3.016381579 15.08190789

2010 4.8 252.3 0.018669778 2.995271382 14.37730263

Update:  Added Plot of Lethality rate change over time for Alrenous

5 comments:

Red said...

The 2000s are when they started to break up the projects and send poor blacks into non black communities via section 8. Cops are notoriously reluctant to take reports of black on any else crime and whites are very well practiced at avoiding black people these days.

What you are seeing is probably a drop in the crime levels in the black areas as the worst offenders are shipped off to whites areas where their prey is smarter and thus able avoid attacks and where the under reporting becomes very high. Blacks are also arrested at higher rates because whites will take the stand and blacks don't snitch on blacks.

http://www.theatlantic.com/magazine/archive/2008/07/american-murder-mystery/6872/

Jehu said...

Another alternative explanation in part is that a meaningful fraction of the best experts in trauma medicine are tied up in our 'invade the world' activities that started back up in earnest after 2001. Anyone know how true that is?

Alrenous said...

I've read that wars lead to improvements in trauma technique. We owe a lot of modern trauma medicine to WWII, I'm told. Note the sudden ending of innovation in the mid-seventies. If the data went back to the Korean war, you could confirm the effect.

The mideast wars should consume many of these surgeons, producing a short-term dropoff, and then the innovation mechanism should kick in, but I'm not seeing it. I don't know enough about ground conditions to even speculate if something's different, however. For example there's clearly a strong alternative mechanism active in the 80's.

My first guess, as always, is government. Innovation suffers from decreasing returns. Regulation grows exponentially. Apparently the lines cross in '93.

Jehu said...

Alrenous,
Yes, the authors discuss this effect in the post Vietnam period---you see fairly steep drops in lethality over that span.
Anecdotally, medicine for trauma has come a very long way even in really recent years. My sister-in-law, for instance, treated a patient who attempted to commit suicide with a shotgun in true Hemmingway fashion. He made a full recovery with not even insane cosmetic damage. A lot of prostethics, for instance, have gotten so good that there is serious talk of banning them in the Olympics in the track & field events.
BTW, the plot is of adjusted homicide rate and actual homicide rate, I'll add a plot of lethality.

Anonymous said...

Fewer young males is the most important effect on crime.