About . . . . . . Classes . . . . . . Books . . . . . . Vita . . . . . . . Links. . . . . . Blog

by Peter Moskos

July 1, 2017

Two-year increase in homicide

Over the past two years, homicide increased 31 percent in America's 52 largest cities.

The cities range from little Richmond (220,000 people) to big NYC (8.5 million), from comparatively safe San Diego (homicide rate 3.5 per 100,000) to dangerous St. Louis and Baltimore (rates of 50+).

Collectively 50.5 million people live in these 52 cities, or roughly one-sixth of America's population. Homicides increased 31 percent over two years (4,946 to 6,496, which is about 36 percent of all US homicides). 45.3 million people live in cities in which homicide rose; 5.3 million live in cities in which homicide decreased.

For graphic representation in the chart above, I removed cities with fewer than 40 or more murders in 2016 because a low n leads to overly dramatic year-to-year changes. This affected El Paso, Seattle, Portland, Raleigh, Omaha, Tucson, Wichita, Long Beach, Minneapolis, and Fresno. (I also dropped Bakerfield its 153-percent increase is either a crazy outlier or my numbers are wrong.) Of the 43 remaining cities, 39 saw homicides go up.

The cities that seem to be bucking the trend of greater violence over the past few years are Seattle, Portland, Fresno, Boston, Tucson, Columbus, and New York City. In terms of raw numbers, the cities with the largest increases in murders are Chicago, Houston, Baltimore, Memphis, Dallas, Milwaukee, Las Vegas, and Kansas City. Were one to take the eight cities with the biggest increase out of the equation -- and there's not any moral or statistical justification for doing so, but just for fun, because the Brennan Center likes doing this trick -- the rest of the cities have a collective 2-year 20-percent increase. That's 20 percent more than we've seen in a very long time. So, no. It's not "just Chicago."

Here are the top 52 cities and their two-year change in homicides, 2014-2016.

And the data in text form, for your cut-and-pasting needs. Albuquerque: +103% | Atlanta: +19% | Austin: +25% | Baltimore: +49% | Boston: -15% | Charlotte: +60% | Chicago: +80% | Cleveland: +33% | Columbus: -11% | Dallas: +49% | Denver: +97% | Detroit: +6% | Durham: +95% | El Paso: -19% | Fort Worth: +35% | Fresno: -17% | Hampton Roads, VA: +39% | Houston: +44% | Indianapolis: +10% | Jacksonville: +25% | Kansas City: +67% | Las Vegas: +44% | Long Beach: +28% | Los Angeles: +13% | Louisville: +73% | Memphis: +63% | Miami: +6% | Milwaukee: +68% | Minneapolis: +19% | Nashville: +83% | New Orleans: +17% | New York: +1% | Oakland: +10% | Oklahoma City: +96% | Omaha: -6% | Philadelphia: +11% | Phoenix: +28% | Pittsburgh: -20% | Portland: -23% | Raleigh: +29% | Richmond: +45% | Sacramento: +46% | San Antonio: +36% | San Diego: +53% | San Francisco: +24% | San Jose: +50% | Seattle: -35% | St. Louis: +18% | Tucson: -11% | Tulsa: +76% | Washington: +29% | Wichita: +31%

July 13 Update: A short while back I finally sent an email to one of the authors at the Brennan Center expressing my concerns about what I see as their deception. I received a brief reply stating, in part, "the statistics you cite [to wit: "Chicago accounted for more than 55 percent of the murder increase last year" & "A similar phenomenon occurred in 2015, when three cities — Baltimore, Chicago, and Washington, D.C. — accounted for more than half (53.5 percent) of the increase in murders"] refer to the increase in murders in a group of 30 cities that we study in our reports – not the national increase in murder." I replied:
Thanks for replying. But these clarifications do not negate the basic mis-truth of the highlighted statements (which have been cited and repeated which clarification). As as academic, I do not understand this. It's not enough to have the truth somewhere in a publication. The summary and abstract, especially when right beneath the title, need to be be accurate standing alone. Surely you understand that readers, especially journalists on deadline, may not have the time or statistical knowledge to parse data as I do. They read and quote the summary. And isn't that what you want them to do? So these need to be factually correct.

Chicago simply does not account for half of the increase in "urban murders." Personally, I would only feel comfortable saying "Chicago accounted for (roughly) 12 percent of last year's homicide increase" (assuming a national increase of about 2,000). But since we have the two-year data, why not use it? I would feel more comfortable saying "Chicago is (approximately) 9 percent of the nation's homicide increase over the past two years."

You've chosen to highlight a large percentage (55.5%) that is large only because of the self-selected limitations in your sample size ("in this group of cities"). Taking a percentage increase within a limited sample is not correct. For instance, were we to look at just the top five cities (NY, LA, Chicago, Houston, and Philadelphia), one could say Chicago accounts for 95.5 percent of the total increase in urban murders. While mathematically true, this would be substantively meaningless if not downright misleading. Or, to further illustrate this point, why not just take the top three cities? The numbers would allow us to say: "Looking at the three largest cities, Chicago accounts for 102 percent(!?) of last year's urban homicide increase." Of course, the numbers come out this way, but one city accounting for more than 100 percent of an increase is both conceptually impossible and mathematically absurd. Does this make sense? The larger sample you use lessons the magnitude of the absurdity, but not the nature of its existence.

Were one to take a larger sample, looking at the top 50-plus cities, then Chicago accounts for 38 percent of last year's increase. And 20 percent of the two-year increase. Were one to include all urban areas, of course, the percentage would be much smaller. But any arbitrary limit on the denominator is statistically dubious.

But back to my initial point -- what is highlighted (and cited in the media) is right there is the lede/summary/subtitle without qualification -- how is one expected to interpret, "Chicago accounted for more than 55 percent of the murder increase last year"? Do you think this is an accurate presentation of data?
Here our correspondence seems to end.

In addition to the UCR, here are some of my source. Corrections welcome.
http://www.kerngoldenempire.com/homicide-tracker http://www.city-data.com/crime/crime-Bakersfield-California.html

Correction, July 5: Numerous errors and typos have been corrected; the post has been updated.


Unknown said...

Are there evidence based social programs that reduce crime?

Peter Moskos said...

Yes. But I can't help but ask what you consider evidence? Real world trials? Peer-review publications?

And social programs? I assume so. (Actually, yes.) But, dare I be a bit cavalier, and ask what the hell does that have to do with policing?

It takes everything. But social programs alone don't solve the problem. Policing alone doesn't solve the problem either. But it sure helps mitigate the negative criminal effects of bigger social problems.

My point is that police can prevent crime. And in various guises it's some form of pro-active policing with community input and consultation that targets and deters (or arrests, as a last resort) specific known criminal offenders.

Jamie said...

Real world trials and Peer-review publication are probably good indicators. I know social issues are complex and applying scientific rigor to them can be difficult.

I would say that that policing could have a lot to do with social programs. Building connections between Police and troubled communities, possible career pathways for young people into the police force for example.

I understand your point that police can prevent crime. I became aware of your website after hearing your talk with Glenn Lourey on BloggingHeads. It was very insightful. I have been following the whole BLM issue for a while and understand your frustration with it.

I have noticed a lack of discourse over what programs or policies are working in terms of preventing the slide into criminality, particularly for young men. It would make sense to focus more on this side of things, not in stead of but in conjunction with effective policing.

Jay said...

In the 1960s (you probably weren’t around) the left took a similar position about the general rise in crime. It was, they said, based on police statistics, and these could not be trusted. They were right to some extent – even some criminologists today refuse to place any faith in UCR numbers other than homicide. Their criticism led to the creation of the NCVS and some good research on police. But it also may have discouraged good researchers from addressing the problem of crime itself rather than these related issues.

I don’t keep up with the crim lit, so I’m curious. Are there any explanations for the increase in murder? I know there’s the Ferguson effect. To support that idea, you’d need to show that cops withdrew, and in proportion to the increase in murder – that, for example, that Denver cops withdrew far more than did cops in Pittsburgh. You would also expect that rates of other crime would follow a similar pattern. Bakersfield, KC, Nashville, Albuquerque, and other cities that show big increases in homicide – have they also seen large increases in Robbery or Aggravated Assaults. After all, if active policing detets murder, it should also deter these other crimes. Do all these crime rates rise in parallel?

And if not a Ferguson effect, what? Who is putting any useful ideas out there?

Peter Moskos said...

Policing and social work must be done simultaneously. (Not necessarily in conjunction.) I've very pro-social programs, health care (mental health care in particular), and education. But that's not what I focus on professionally or in this blog. Some people and neighborhoods have the cards stacked against them. Police have to play with the cards as they are dealt. And even given a bad hand, police can still be a force for crime prevention and good.

How we negate a history of racism and current issues of drug abuse and poor (or absent) family structure, I don't know. But that's for other people to work on and solve. I have opinions, of course, and an education. But I focus on the policing part of the equation.

And, it's worth point out once again that social factors have virtually nothing to do with the recent *increase* in murder. Violence has gone up as poverty has gone down nationwide. Do I think there's cause and effect there? No. But these issues need to be disaggregated. Reducing poverty is good for it's own sake. But it's not an effective violence reduction strategy in the short term. Nor do poor people magically start shooting each other because they're poor.

Peter Moskos said...

Those are good questions, Jay. I, as I think you know, generally avoid crime stats except for homicide. Right now there's nothing but the usual suspects coming from the left regarding crime. Implicit bias, structural racism (like that's new), police legitimacy, social justice, drug treatment (which was the liberal mantra circa 1990; it never happened and yet crime went down). Some of those are important. Others are nebulous concepts that can barely be defined much less quantified and improved. It all fits under the ideological umbrella of police are the problem and need to be contained.

The problem now as I see it is liberals are wasting precious brain cells denying the significance of the violence increase. How we can we get to violence solutions if we're still debating whether reality is real? That's why I find these posts (my posts) so frustrating. Saying "violence is up" is not what I want to be wasting my time on. It's step one. And I can't most on the left, as more young black men are murdered, are still in denial and citing the Brennan Center's unbelievable BS that there's nothing to see here but right-wing sensationalism.

As to the link between homicide and other violence. They're always tied together. They go up in sync. Absolutely. In part because of mutual causation. In part because of multicollinearity. An attempted failed murder is an aggravated assault.

The problem now is that homicides are way up and stats on other violence are only a little up. There's an increase in both un-reporting crime and un-tallied crime. (The latter simply because arrests are down.) But hell if it can ever be proven, since by definition it's not being tallied. (And the NCVS is so flawed I don't know if it's any help at all. Eg: http://www.copinthehood.com/2013/11/crime-is-up-no-wait-crime-is-down.html. I'd go with hospital admission as a better indicator, because I still have faith in the CDC. But you're not going to get property crimes.)

A semi-related correlation which is worth observing is that police-involved shootings are always correlated with citizen violence. But we're not seeing that right now. Police-involved shootings should be up, but they're steady. (In NYC, where the data is good, you can see them go up in the late 80s and way down in the 90s, in perfect sync with violence, with an additional downward trend overall). I think there are two confounding factors at work. Violence is up, but police are less proactive, engaging less. Fewer interactions. And these two variables function in opposite directions and I think are canceling each other out.

Gary Cordner said...

Nice work. What makes Seattle, Portland, El Paso, Tucson such outliers?

Andy D said...

The thing I am pondering is the three cities with decreases and New York with no change. I guess I haven't seen a lot of negative press about cops in Chicago, Boston, or Columbus, whereas a few of the cities with bigger increases have been in the news with negative police interactions (Albuquerque, Baltimore, Tulsa, Chicago...)

I personally buy the whole less proactive policing/more violence theory, but Jay has a great point in that to prove the hypothesis (or at least offer it good evidence) it seems that plotting proactive enforcement alongside homicide would need to happen (since drug arrests and traffic stops are good indicators of that, could that data, if available, be plotted over this data?)

I definitely understand the frustration level you have; you have been talking about the rise in violence for a long time now and many are still denying it. I don't run in terribly progressive circles (I'm a cop for God's sake!) but in the media I still hear it when they aren't talking about Trump Tweets(tm)

Peter Moskos said...

Gary: I don't think Seattle, Portland, El Paso, Tucson are outliers. They just seem to be at the end of a pretty normal distribution.

Peter Moskos said...

KC and Bakersfield (Bakersfield is actually off the chart) might be outliers. And probably because they have a small population.

Peter Moskos said...

For those who care, just looking at the numbers (each city equal) n=53, mean is .36, range is 1.88. Std Deviation 0.40. Skew is +0.74.

Peter Moskos said...

Kansas City wasn't *just* an outlier. It was an error. I put in the metro total for the past two years, which of course is a lot more. In reality the two-year total is only(?) up 67 percent. I've corrected the figures above. And I am removing Bakersfield from the chart as an outlier.

Gary Cordner said...

Re Seattle, Portland, El Paso, Tucson I should have been clearer that I was looking not so much at their 2-year trend as at their very low rates (compared to the other cities).

IrishPirate said...

Gary Gary Gary,

let's talk about the correlation that dare not speak it's name. The higher the percentage of black residents in a city the higher the murder rate will be. Thankfully, there are exceptions to this like NYC, but that correlation is there. Of course I'm sure the Brennan Center would declare me to be a vapid racist or perhaps even the Grand Beagle or Kleagle of the KKK. Now since I would proudly wipe my ass with the Confederate flag I imagine the KKK wouldn't want me, but perhaps they're more open minded about wiping one's ass with the CSA flag than I imagine. I doubt it though. Plus my sheets are fitted and the wrong color.

John McWhorter, black guy and scholar, wrote a piece for the National Review on race and IQ. Ignore the IQ bit, but look at the parts on how culture can be transmitted from generation to generation. He takes part of his argument from a linguist's perspective. http://www.nationalreview.com/article/449208/race-iq-debate-serves-no-purpose

Here's a big quote:

"As I have argued for 20 years now, a people are determined not only by external conditions but by the norms passed on in their culture. Many, including most academics, insist that culture is itself determined by external conditions, but this is an oversimplification. Norms and culture, once settled as habit, can persist long after the external causes that originally created them."

Now the quickest way to reduce the homicide rate is to get back to the way we were doing policing long ago. Say 2012 for example. That means cops engaging and harassing the bad guys. It's as simple as that.

In Chicago police killings have gone way down the last few years. This I would argue is a good thing. However, it comes at a price. Cops in Chicago may be killing five or so fewer people a year, largely black men, but that's because interactions with the bad guys are down. This has caused the number of murders to go up by say 300 a year. Play with the stats as you choose, but the homicide rate has almost doubled. Considering the rapid flight of blacks from the city the homicide rate in the black community has gone up even more. http://time.com/4635049/chicago-murder-rate-homicides/

So because of less police interaction with bad guys we're seeing cops kill five fewer people a year in Chicago. However, for every action there is a reaction and that reaction is 300 or so additional dead.

Let me get out my calculator. 300-5 = 295 additional deaths.

IrishPirate said...

Part Deux because I went over the character limit. Which is amazing because I generally lack character.

Now the sad truth is that there are people out there who think that is an acceptable number. They would see that as a good trade. Is BLM out there marching every time some black guy gets killed by another black guy? Of course not. Their shoes would wear out. Does BLM march nearly every time a Chicago cop kills a black guy. They seem to. Hell, we had one such shooting caught on video with the deceased waving a gun around and BLM still marched.

Now the long term way to reduce crime involves jobs, education, mental health counseling, etc etc etc. Short term it involves going back to the dark ages of policing when cops actually engaged bad guys.

So what we need to do is fine some old time cops who recall 2012 and get back to proactive policing. The alternative is thousands of additional dead a year.

Long term Chicago's murder rate will likely drop if only because the black population is fleeing so rapidly. However,the folks with the means to flee are also the folks who exert positive social control in their communities. In the short term we're seeing two related phenomena:

1. Blacks fleeing Chicago and the rate and actual number of black homicide victims going up.

2. Me writing annoying comments.

There's something in black culture and the black American experience that leads to more violence. Higher percentage of dumb ass teenagers carrying guns may have something to do with it. Destructive behavior also seems to be more "outer" directed than destructive behavior in the white community. We're not seeing a huge increase in homicides in areas where whites are overdosing on opiates, yet we're seeing a huge increase in death.

I don't have any short term solution to eliminating racial disparity in America. I don't believe there are any. I do believe there are long term solutions.

I do have a short term solution to reducing crime though. It's cops making the lives of criminals uncomfortable. Bangers need to fear being stopped and caught with a gun. They have to fear meaningful jail time if caught with a gun. Not the "get out of jail" quick schtick we have in Chicago.

So endeth the lesson.

Peter Moskos said...

Just to put some homicide rate numbers out there The US is 4.9 per 100,000 in 2015 (up to maybe 5.5 in 2016). And the average 2015 & 2016 average rates for Seattle, Portland, El Paso, and Tucson, respectively are 3.0, 4.1, 2.5, and 5.7.

Also, though I don't have the numbers off the top of my head, it's worth pointing out -- and I'd like to think this goes without saying here -- that violence in *white* America is shockingly high compared to comparable nations. America is a violent nation. It in our history, our laws, and our culture. It is just more extreme in black America.

IrishPirate said...

Sure, throw that "fact" in there before the liberal torches and pitchforks come out.

Of course violence in "white America" is higher than say in Sweden. Partly it's our history, laws, and culture. It's also our access to guns. The newish atrocity of a white guy shooting the black girl trying to merge into traffic shows how easy access to guns has an effect. Ever hear of giving someone the "finger". Less deadly. Better yet, chill out and let her merge. Friggin cars and their psychological effect on psychos.

Lord I love walking, biking and public transit.

Of the many things annoying about the Brennan Center's "take" on the homicide rise is that it ignores the real damage being done to the black community. This "it's a blip" BS has real consequences. Say 250 additional dead young black men in Chicago yearly. Over a 20 yr period that's 5000 dead plus say 8-10k children who are not born. Those are real numbers that have real demographic and social consequences. That's not even counting the flight out of Chicago by the black community. Empty neighborhoods have consequences too.

Of course activism does have some positive consequences. New trauma center being built at the University of Chicago to handle gunshot victims. That's a good thing. Better trauma care has had a discernible effect on homicide numbers which partially masks how bad the violence really is.

Sigh. Every positive effect has a negative effect.

IrishPirate said...

As Richard Nixon might say "Let me make something perfectly clear".

I don't believe there's something inherent in W African DNA that makes black Americans so much more violent than other Americans. I do believe that 400 years of being on the ass end of the American experience accounts for that difference.

As Joyce might say the reality of experience has forged in the smithy of America the created woes of the black community. Is that pretentious enough? Will the Brennan Center forgive me?

James Joyce. From "Portrait of the Artist as a Young Man":

"Welcome, O life! I go to encounter for the millionth time the reality of experience and to forge in the smithy of my soul the uncreated conscience of my race."

God, was he a pretentious jagbag or not?

Adrian said...


Peter Moskos said...

That's a fun little link. Thanks!

Unknown said...

Let me start by saying I agree with everything you said. The problem is that your offer no explanation about why it's gotten so much worse so fast. No one is arguing that black people have it much worse than in 2013, so why are black people shooting each other at such higher rates than before? Part of the answer, at least here in Baltimore, is that police are much less willing to get in the middle of the situation before the shooting starts because they know they won't be backed up by the Mayor, SA (who clearly despises the PD), and maybe even police brass. Why bother when you can just sip coffee at Royal Farms? BPD was a mess before Freddie Gray but it's gotten far worse since.

The bigger issue here in Baltimore is juvenile crime. The expungement rules must change. We are dealing with an epic crime wave perpetrated by 15-17 year old boys. They know there are no consequences and they are being recruited by drug gangs to commit all manner of crimes, most of which just go away. The SA issues a press release about the arrest, then ends up dropping charges when no one's paying attention anymore. It's a fiasco; enough for me to consider leaving the city after 35 years.

Peter Moskos said...

Of course the rise crime relates to police, or more specifically less pro-active policing that has come because of demands from the lunatic wing of the Democratic party (and in Baltimore the mayor, Mosby, and the US DOJ). I've wrote a lot about that after the riots.

But here's the problem: when I start talking about solutions to the rise in crime, hard as it is to believe, many people still deny that violence is up. Or if it is, in places Baltimore and Chicago (and it is) then it must be because of bad social condition. Of course that ignores that none of that social conditions got worse in the past year. So I took a step back here to at least get everybody on the same page. One can't talk about solutions until people admit there's a problem. But you're absolutely right.

Thos Wallace said...

"July 13 Update: A short while back I finally sent an email to one of the authors at the Brennan Center expressing my concerns about what I see as their deception."

Wow. Just wow. They have no personal integrity regarding their most widely distributed work product. And no sense of shame.

Although it is showing up all over. I assume you are aware of the replication crisis in social psychology, p-hacking, researcher degrees of freedom, etc. The good news is NHST (null hypothesis significant testing) is on its way out.

I had an odd experience regarding a widely publicized study by Stanford researchers. The problem with the paper would be analogous to the Brennan Center arguing that homicides are down. Of course, the data wasn't easy to find. Although it slipped in to the middle of a very lengthy, dense, technical appendix. After trying to follow up, no one seemed interested.

It's not like the current homicide rate is directly comparable to the 90's. Trauma centers are widely available and deaths per shooting are lower. In addition, auto thefts tend to be counted by insurers -- and they are down due to anti theft technology. 2000 era Hondas are still among the most stolen vehicles, because they are easy to steal. People use cash much less. Cell phones are easy enough to steal, I suppose, but can be bricked. The big jump in homicide rates are even more meaningful with other statistics, which were always suspect, even less meaningful. So, hell yes.

I admire your tenacity regarding this.