About . . . . Classes . . . . Books . . . . Vita . . . . Blog. . . . Podcast

by Peter Moskos

December 30, 2015

There is absolutely NO NEED TO PANIC!

The latest Brennan Center report projects the 30 largest cities will see a 14.6 percent increase in homicide this year.

You know the last time the nation saw a 15 percent annual increase in the homicide rate?


Remain Calm. All is well.

But don't worry, they say in their best "you are getting sleepy" voice. There's no reason to concerned:
However, in absolute terms, murder rates are so low that a small numerical increase leads to a large percentage change. Even with the 2015 increase, murder rates are roughly the same as they were in 2012 — in fact, they are slightly lower. Since murder rates vary widely from year to year, one year’s increase is not evidence of a coming wave of violent crime.

A handful of cities have seen sharp rises in murder rates. Just two cities, Baltimore and Washington, D.C., account for almost 50 percent of the national increase in murders.

These serious increases seem to be localized, rather than part of a national pandemic, suggesting community conditions are a major factor. The preliminary report examined five cities with particularly high murder rates... and found these cities also had significantly lower incomes, higher poverty rates, higher unemployment, and falling populations than the national average.
Hmmmm, Statistical aberration are always a possibility and poverty and falling population makes me drowzzzzzzzzzzzy.

But when I snap out of it, I'm still concerned. Why do so many seem to be in denial about such a large increase in murder.

Can't we be politically correct and also ask what the heck is going on? When FBI director Comey said he was concerned, he received loud chiding from the political left and even a presidential rebuke.

If you think it doesn't matter, please let me know exactly what conditions need to be met, specifically how many more people have to die, before we are allowed to be concerned and move on from silly semantic debates. Shouldn't we better focus our efforts and, if you're so inclined, even your outrage?

No, don't panic. But frankly, I think it's OK to be a little concerned.

I have an idea! Instead of denying a dangerous increase in lethal crime, why don't we put on our thinking caps and ask what has changed this year with regards to policing and violent crime. But before you answer take a deep breath and then come back after a good night's sleep!

sources include:

(Correction: Originally I missed the the fact the Brennan Report was only talking about the rate in the 30 largest cities. This post has been changed to include that rather important detail.)

1, 2, 3, 4, 5.]


Concerned citizen said...

Prof. Moskos,

You ask, "Why do so many seem to be in denial about 2000 more murders?"

First, let me say, I thought you did a great job answering that question in an earlier article on this blog.**

However, it's not just the presence of denial that I wonder about; it's also the absence of compassion for parents and siblings of those 2000 additional murder victims, as well as the law-abiding people who live in the neighborhoods.

**"Many of those who are driving the 'there is no Ferguson effect' bandwagon still believe that police are largely irrelevant to crime prevention and, rather than having anything to do with crime prevention, serve primarily as agents of racial oppression."--P.M.

fh said...

I believe you have blocked me so I don't know if you understand that this data represents a fairly unsophisticated look at only the 30 largest cities (some of which do not have data such as SF). The 14.6% increase is an aggregate of only this data of which ~200 additional murders are in Baltimore and Washington, D.C. It says nothing about the overall number of murders in the country and the 14.6% increase represented by this data is only a 400 total absolute increase or spread over the entire 20k (I think that's the approximate number - literally too lazy to google)or so murders in this country a contribution of 2% to an overall national increase (whatever that ends up being). Whatever the increase or decrease for the nation, this will be a relatively small piece.

Crouchback said...

There's some interesting details in the table. The highest increase in murders by percentage isn't Baltimore or DC, it's Denver. Oklahoma City comes in 4th (after DC and Baltimore.) That might be worth some discussion especially if you're looking for evidence for or against an anti-police factor. I don't recall anti-police riots in Denver but perhaps they didn't make the news.

I think the current crisis in Baltimore is influencing interpretation of the data. For Baltimore, as Moskos has shown, there is very strong evidence that poor police relations have lead to a spike in murder. The murder rate shot up in the wake of the riots. The evidence is weaker for the rest of the country. Does anti-police sentiment explain Denver or Oklahoma City or Louisville or Charlotte?

Also, overall crime rate is down for the cities in question. Now Professor Moskos has stated he doesn't think NYPD is cooking the books so either everyone else is cooking the books or somehow murder rates went up while other crimes went down. I find that hard to fit into an anti-police narrative. Cops were too demoralized to catch killers but not to catch petty criminals? The public was willing to inform on petty criminals but decided not to help catch killers? That suggests other factors.

The estimate of 2,000 more murders assumes the 14.6% increase will hold for the overall US population. We won't know for several months.

To be honest I find the lead poisoning hypothesis more convincing than good policing. Good policing might explain drops in NYC crime in the 1990s but we also saw dramatic drops in Washington, DC crime in the 1990s. That was with Mayor Marion Barry and a police force that some years killed as many people as the NYPD with about a tenth of the manpower. I'm not saying good cops don't matter, I'm saying like teachers they can be overwhelmed by other factors.

fh said...

Came back to see if my comment got posted. I will eat my hat if homicides increase 10% this year; I will be very surprised if they increase even more than 5% (despite this 30 city sample contributing a 2-3% increase all by itself (the denominator is only 13k). That chart is right there (didn't even have to google it). The decrease in homicides is amazing and I wish we knew why; definitely not because of policing in my opinion.

As to Baltimore, I don't argue with Peter that there has been an effect and it is due to policing. We philosophically disagree about who to blame. He blames the political establishment and Mosby who he thinks is unjustly beating up on cops and that cops are responding appropriately. I am the exact opposite and blame the cops who I see as selfish and corrupt. In my mind they have decided to stop all policing - both the good and bad kind - once they were put on notice that they would be held accountable for the bad kind. We will never agree I'm afraid even with the exact same data in front of us.

direwolfc said...

Peter - you definitely need to update this post to reflect the fact that the report was for the 30 largest cities, and that Baltimore and DC *do* account for almost half the increase.

My main concern is that the +60% reported for Baltimore actually understates how bad things have been since the riots in Apr. In reality we are seeing a ~100% increase in homicides that has gone unabated since it started. Next year 500 homicides is not out of the realm of possibility if things don't drastically change for the better.

Moskos said...

Yeah, I did update the post. Somehow I missed the important fact that the study is only predicting for the biggest 30 cities. That certainly matters and does change things. Sorry about that.

fh, it's funny you think you were blocked. If you posted something that pissed me off, I don't remember it. But what's funny is that you think I *can* block people from posting. I don't think I can (but yes, I can delete posts. But I rarely do so).

Adam said...

FWIW, FiveThirtyEight analyzed the 60 largest cities and found a 16% increase.

Moskos said...

If this were some kind of "Ferguson effect," I would expect the increase to be less in whiter suburbs and rural areas. Still, the 60 largest cities have 50 million people. That's a pretty good sample size.

If the 60 largest cities go up 16 percent, I would be very surprised to see anything less than a 10 percent increase nationwide. 12 or 13 percent wouldn't surprise me. But that's just a guess.

fh said...

We disagree about this and I'll guess we will see. On the other hand, I think your belief that proactive policing matters in preventing crime (except strangely crime is not up in those cities) should lead you to a different conclusion about the overall number vs. big cities. Proactive policing really only happens in big cities. The population density just does not lend itself to proactive policing anywhere else (at least the kind that prevents crime). So if this is a nationwide 10-12% increase then I have to say that something is going on that we fundamentally do not understand nor even have a hypothesis for.

I believed I was blocked because I was under the impression that you always indicated when a comment was deleted but in your "choose your adventure" post, a comment I made which made it through moderation (are you claiming that the notice I get saying my comment is being moderated is not accurate?) was then silently deleted. It was there one day and gone the next without any indication it had ever been there. Given the anomalous nature of the deletion (no notice), I assumed that it had been mistakenly released from moderation (where blocking would generally occur) and then deleted silently to correct the initial moderation failure. My mistake BUT I do wonder what kind of glitch would have resulted in silent deletion.

Unknown said...

I am fascinated by the significant increase in homicides in large cities BUT flat or even decrease in overall crime. On the one hand, I think Prof. Moskos is right that the numbers for the largest cities show that there's *something* going on that is leading to the increase in homicides, and that the Brennan Center's attempts to explain them away as either not worrying or statistical noise is unconvincing. On the other hand, the Ferguson Effect hypothesis seems contradicted by the flat or decrease rate for overall crime. Need a hypothesis that explains the evidence, but don't know of one yet.

It might be interesting to see what the crime rate is for other categories of crime. Is the increase in homicides also reflective of an increase in other violent crimes such as rape and robbery? Or, do other violent crimes show no increase, like the overall crime rate?

Moskos said...

If homicides are up, I simply do not other believe crime is flat. They're always correlated. So I think it's a data problem.

The most obvious hypothesis, one that certainly holds for Baltimore, is that A) cops make fewer on-view arrests (hence, crime goes down statistically and B) people aren't calling police when there *is* crime (hence, crime does down statistically). It's mostly about what is reported, and a bit about what cops choose to do.

Both A and B would certainly be influenced by the some variation of the Ferg Effect (I think we really need a new name) and the narrative that cops are the bad guys.

Moskos said...

As to comments, only comments on old posts get "moderated." That's to keep out the spam and the hard-core kooks and the bots. The rest, I'm pretty sure, go straight through... *unless* blogger thinks they're spam, in which case I eventually fish them out. That is not a rare occurrence.

I do remember deleting one comment on the choose-your-adventure post. That was was purely for editorial quality control. I don't remember for sure, but I thought it was poorly written or I was too stupid to understand what the point was and, as it was the first comment, didn't want the comment thread evolving from it.

Concerned citizen said...

PM: "If homicides are up, I simply do not other believe crime is flat. They're always correlated."

The Brennan report disagrees: "....using murder as a proxy for crime overall is mistaken."

I'm going with Prof. Moskos on this one.

Matt Ashby said...

I think the answer to the question you pose depends on exactly why you’re interested in the number of homicides occurring, although I can see why at least two groups of people wouldn’t be that worried.

The first group of people is interested in current homicide trends because they want to predict the number of homicides likely to happen in the future. These people are likely to be very worried about noise in the data (as the Brennan Center report points out) and so are probably not going to be worried until they see three or four years of data that suggest that the drop in homicides since 1991 has come to an end. It’s perfectly legitimate for those people to say they’re not worried, or at least not-yet worried, because of their specific interest in long-term trends. How many more people have to die for them to be worried? Presumably several hundred more each year than in the most-recent past years.

The second group of people is interested in preventing homicides, and for them a one-year increase in homicides might be much more worrying. However, given that the number homicides represented by the sorts of percentage increases discussed above is much smaller than the total number of homicides, I would expect these people to be far more concerned by 13,000+ people being murdered every year than by (say) 1,000 more being murdered one year than another. Yes, 1,000 more people dying is a tragedy but 13,000 more people dying is a much larger one, so that’s where I would expect these people’s interest to lie. As such, people in this group might actually not be that interested in year-on-year changes, since they’re quite small relative to the huge numbers of people being killed each year in total. How many more people have to die for them to be worried? None, because the US homicide rate being so high means they’re probably pretty worried already.

There will be other groups of people with other interests that influence whether or not they are worried by any one-year changes. People interested in the relationship between policing and crime (e.g. many academics) probably are going to be interested, but that doesn’t mean that other groups can’t legitimately say that any change in homicide numbers since last year isn’t worrying.

Unknown said...

"The most obvious hypothesis, one that certainly holds for Baltimore, is that A) cops make fewer on-view arrests (hence, crime goes down statistically and B) people aren't calling police when there *is* crime (hence, crime does down statistically). It's mostly about what is reported, and a bit about what cops choose to do.'

I know you said it's only a small part of it, but I don't see how cops making fewer on-view arrests makes any difference. If a crime is committed and reported, then whether the cops saw it happen and made an arrest shouldn't make any difference. Your hypothesis in (B) is interesting. Isn't there also a big annual crime victim survey? If lower crime is an artifact of less reporting to police, then I'd think the crime survey would reveal that.

Moskos said...

Because a lot of arrests happen on-view (without a call for service) Every time a cop makes a discretionary arrest, a crime (as a result) is recorded. No arrest, no crime. A huge amount of crime isn't recorded. Basically many (most?) misdemeanor arrests wouldn't be recorded as s crime were there not an arrest. And probably half of felony crime isn't reported to police (or so "they" say).

And yeah, in theory the NCVS should rise above those issues. But it doesn't. I don't trust it either, though. See, for instance http://www.copinthehood.com/2013/11/crime-is-up-no-wait-crime-is-down.html

Moskos said...

Pragmatic, that's a good analysis. But isn't there a difference between saying, "statistically I need a few more years of quantitative data before I'm willing to go on record as saying there's a real trend" versus "there's no reason for people to worry about murder being up because it isn't up until the numbers say so with statistical significance"? I mean, isn't the latter -- putting statistical significance over what is actually happening -- the epitome of head-in-the-sand academic ivory tower?

The idea that other people shouldn't be worried until you have data to prove some thing isn't random (with 95 percent certainly) is crazy, especially when people are dying.

The thing is -- and while I would want scientific statistical certainly with regards to cause and effect (for drug trials or something) -- it's kind of grad-school absurd for quant people to say something can't be happening until *they* have data to show so.

Does the homicide increase of 2015 not happen until the UCR tells us it did in 2016?