Blacks who commit homicide do so at a rate about 7.5 times larger than whites who commit homicide. The trend in black homicide correlates well with the change in overall homicide.
Going on a suggestion made by JH when we first examined the homicide statistics1, I put both races together on one plot to make comparisons fair.
First is the homicide rate for Blacks and Whites, over-plotted with the homicide victim rate for both groups, finally over-plotted with the overall homicide rate (taken from the post on guns and homicides).
Figure 1
Note that the blue line is scaled to the right vertical axis, and is the overall homicide rate (as estimated by the FBI and Census). The black lines are killers, and the red victims. The dashed are Blacks, the solid Whites. (See this post for separate pictures of each race, where the general decline of White homicide rates, and its slight upward bump in 1990 is more visible.)
Obviously, the overall homicide rate correlates nicely with the Black rates. In an odd coincidence, according to the FBI the number of Black victims is almost identical to the number of White victims.
Whatever is happening in Black populations is largely responsible for the decline in homicide rates.
One thing appears more or less constant: the rates for Black killers are much higher than for White killers, as this picture which shows the ratio of rates demonstrates:
Figure 2
Rates bob around, but average out to about 7.5 times higher.
Perhaps the difference lies in who is killing whom. Lumping all the Black and White homicides together, here are the percentages of Blacks killing Whites and White killing Blacks:
Figure 3
Racial animosity is on the rise, homicidally speaking. This comes as no surprise to anybody who isn’t an NPR listener.
Lastly, here’s the percentages for the races killing each other:
Figure 4
Fairly self-explanatory.
Update Fig. 2 with new axes limits.
Fig. 2b
—————————————————————————-
1The data are from the FBI (details here), the Census Bureau, and the Department of Justice (details here). For W/B,B/W homicides, see htus8008f19.csv in the DOJ link.
Briggs,
White and Black, or Black and non-Black? I link below to a fascinating collection of reports from the Chicago Police Department, and organization which appears to have some interest in terminal affairs.
I suspect that there is a lot to work with there, for example the departed (murdered) in Chicago tend to be predominantly folks who have been guests of the Judicial System at an earlier point. Most have been sent on their way by handgun.
One might ask what might be the annual local consumption of ammunition in the calibers preferred, (these guys are well known to be lousy shots – maybe a matter of seeing too many movies demonstrating an especially ineffective method for grasping a handgun), how many shots were required to dispatch the departed, collateral woundings and killings, numbers dispatched per event, and perhaps more importantly, percent of the killer/killee cohort as a part of the local population.
I was fortunate to sit through a course in criminology in 1963 and to hear that the 18-27 male cohort was the author(?) of most of the crime in the US. The larger this cohort was the more crime.
The lecturer, who was fond of the term “adjudicated”, suggested that some smart politician would recognize an oncoming reduction in the numbers of the miscreant containing cohort, devise some scheme to “reduce” crime and then when the stunted cohort came of age, take credit for his/her success.
I cannot recommend to highly a look at say the 2011 Chicago Murder Analysis Report which can be found at the above link, not the least because of the discontinuity between the content and the attractive cover.
One last bit. It does appear that more murders are the result of the activities of the local shooting clubs than any other cause. I wonder if the NRA has considered inviting their application for membership. It could be that with a little instruction and target practice, they might reduce the rate of collateral murders, and increase the percentage of forced departures of prior guests of the justice system.
It is clearly another example of how informative looking into something can be.
https://portal.chicagopolice.org/portal/page/portal/ClearPath/News/Statistical%20Reports/Murder%20Reports
Upstream from the linked page is a broader listing of publications of the Chicago Police Department including “Crime at a Glance.”
No-one could say these guys aren’t droll.
I would expect better data presentation from you than this, Matt.
Figure 2 should include the origin, so you can judge visually whether these results represent a roughly constant or a wildly varying ratio.
Figures 3 and 4 are misleading. If you care about inter-racial homicide, what really matters is whether the overall rate is going up or down, not how that compares to the overall homicide rate, which is dominated by black-on-black killing. The last figure (Fig 4) seems especially useless, because it suggests visually that there’s a mirror relationship between white-on-white killing and black-on-black killing. But, surely that’s another artifact of the fact that these stats are all dominated by black-on-black killing. When that goes down significantly, the percentages of the other categories will all rise, but that says nothing interesting about them at all.
Oh, I wondered why there was an inverse correlation between black-black and white-white murders, if they are % of all murders, dominated by black-black then of course they look inversely correlated.
And does white mean actual caucasian, or are hispanic being grouped in with that group? What about asian? It would also be interesting to see African American (used as African’s who are naturalized citizens) vs American blacks (used as American with some fraction of African), but I suppose that data isn’t gathered.
This report was last updated in 2005, but it’s still interesting.
http://colorofcrime.com/
An increasing white on black and black on white might correlate with increased intermarriage. That makes more black-white in-laws. Since most murders are family related …
Mathematician “La Griffe du Lion” has written a number of related pieces.
See the links below.
Why Most Serial Killers are White Men
http://www.lagriffedulion.f2s.com/serial.htm
Politics, Imprisonment and Race
http://www.lagriffedulion.f2s.com/prison.htm
Racial Disparities in School Discipline
http://www.lagriffedulion.f2s.com/discipline.htm
Aggressiveness, Criminality and Sex Drive by Race, Gender and Ethnicity
http://www.lagriffedulion.f2s.com/fuzzy.htm
The Color of Death Row
http://www.lagriffedulion.f2s.com/DP.htm
Crime in the Hood
http://www.lagriffedulion.f2s.com/hood.htm
Tom P,
There’s a good reason R has the defaults axes it does; Fig. 2 is plotted with these defaults. A quick glance at Fig. 1 and then at Fig. 2 shows that these series are roughly similar for Blacks. There is a little more noise in 2, true, but the only conclusion I made, and the one I don’t think you dispute, is that Blacks commit homicide at a much higher rates than Whites. True?
Figs. 3 and 4 are lifted directly from the table I pointed to. If you think these “especially useless” I suggest emailing the DOJ and telling them to stop compiling this data.
But then your statement “what really matters is whether the overall rate is going up or down” is false. Suppose the Black and White homicide were perfectly flat, and even equal. It could be that Blacks just kill blacks or that they kill only Whites, and vice versa, or it could be that Blacks only kill Whites and vice versa. There’s no way to tell who is killing whom just looking at the overall rate.
Ferguson.
Fascinating link.
Briggs.
Could you elaborate a tad more on why figure 2 is ok? I also would have preferred a axis starting at 0 (and stopping at 10) for data going from roughly 6 to 9.
Bill S,
Sure. Why stop at 10? Why not 20? Or 100? Or, since the range can be infinite (only Blacks could be killers, say), why not as high as we can go?
Plus, since I only want to establish that the rate is about 7 times higher, I’ve done that with this chart. At least, I don’t think I hear anybody (not just here, but anywhere) claiming anything different.
But, in the sake of domestic harmony, see above shortly for an update.
Replaced my own brake pads before reading latest blog.
Decided 56 year old pencil pushers should not do that.
Using Crown Royal to ease the ache in lower back.
Do not claim full faculties.
But the question is honest. I would use 1 to roughly 110% of peak because the data is a ratio.
But what I do is based on experience presenting data to customers. I was hoping you knew some rules on presenting statistical data and would elaborate.
Bill S,
Trade you graphing secrets for a shot of that Crown Royal.
Matt,
I read recently (don’t have a link inhand) that when accounting for single parenthood, blacks ad whites commit murder at the same rate. Thoughts?
Swade016,
Sounds plausible, anyway. Data?
What I’d like to find are the same stats above, but going back pre-WW II. Then we’d have some better indication of how social policy influenced the numbers.
The rule of thumb for scaling recommended by Joiner was that the window of the time series be 1.5 to 2.0 times the range of the data. It’s an aesthetic choice: random variation looks random, assignable variation looks assignable. On Shewhart control charts, the control limits ought to be about an inch apart. (Lit. “rule of thumb”) This is useful in industrial situations so that one does not spook the herd.
It would be interesting to see the effect of punishment policies on crime severity. With 3 strikes law and death penalty one might be incline to eliminate witness if things goes bad. Though a vast majority of killings are gangs related.
Related to the underlying reasons for the trends noted:
Violence—Suicide is Contagious: http://www.micronesiaproject.org/2010/11/i-somehow-feel-that-i.html
Violence is Contagious: http://www.wired.com/wiredscience/2013/01/violence-is-contagious/
Bystanders effect on interpersonal violence makes a difference: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1298601
Watching violence—especially by youth with developing brain neurology—leads to desensitization to violence: http://www.rcgd.isr.umich.edu/aggr/articles/Huesmann/2007.Huesmann&Kirwil.WhyObservingViol.CambridgePress.pdf
…but then, that knowledge is nothing new: http://killology.com/book_stop_summary.htm
http://www.killology.com/ is a website by a retired military officer dedicated to the psychological effects of having to kill [e.g. by policy or military] and related effects. Humans appear to be “hard-wired” such that the act of one human killing another human appears to be intrinsically distasteful & instinctively unnatural under any circumstance, including unquestionably legitimate situations such as in various self-defense scenarios. This has been observed in a number of military engagements (especially in WW-II, and via archeological & re-enactments of US Civil War engagements) – the findings led to means of desensitizing military recruits between WW-II and the Vietnam conflict. Many of those de-sensitization techniques have found their way into numerous video games and television shows. Anyone interested in this would benefit by reading, “On Killing, : The Psychological Cost of Learning to Kill in War and Society” by Dave Grossman (http://www.amazon.com/exec/obidos/ASIN/0316330116/qid=973195800/104-1614587-2282360 ).
“It would be interesting to see the effect of punishment policies on crime severity. With 3 strikes law and death penalty one might be incline to eliminate witness if things goes bad. Though a vast majority of killings are gangs related.”
If you want to look at the effects of severe punishments on murders, see Singapore, Saudi Arabia, or Bahrain.
Pingback: Firearm Homicide Rates Around The World: Comparing Against USA | William M. Briggs