Justifying harsh sentencing through the power of a single image

Any discussion of reducing prison populations must deal with the fact that crime rates, and particularly “violent crime” rates, have fallen dramatically during the time when federal and state sentencing practices have been especially harsh. In that vein, and with a big and obvious caution about oversimplification, the raw power of the following image is striking:

Image credit: FBI, Uniform Crime Reports, 201

Image credit: FBI, Uniform Crime Reports, 2012


PS Thanks to Professor Doug Berman. SeeFBI releases 2012 crime statistics showing stability in relatively low crime rates.

26 responses

  1. Ah yes, the old post hoc, propter hoc fallacy. What are we being asked to conclude here? That the draconian sentences caused the reduction? There is zero evidence for that.

  2. There is a HUGE cost in keeping inmates imprisoned, which cost filters down to taxpayers. Having worked in a correctional setting for many years, it is not simply a matter of locking up offenders that reduces crime rates. Yes, it keeps some people off the streets for awhile, but it still comes down to rehabilitation – changing the way offenders think and act, stopping cycles of abuse and addiction, treating them with some form of humanity and respect, and helping them understand the consequences to their victims – whether they are in prison or on probation. There are some people with psychological and personality disorders, and others who simply refuse to change, that will be habitual offenders, but it is proven that throwing offenders who have a chance to change their lives in with habitual criminals does not help. There has to be a concerted effort toward rehabilitaion and change, rather than locking people up and wanting to throw away the key.

    I realize there will be discussion on both sides…this is simply my viewpoint. I will be interested to read other viewpoints.

  3. The problem with such graphics is they encourage the mistake of equating correlation and causation, i.e., from the correlation of harsher prison sentences and lower crime rates we are to conclude that one causes the other. However, similar graphs have been drawn showing a correlation between the reduction in lead in gasoline and lowering crime rates, increased abortion rates and decreased crime rates, and so on. Yet, few would suggest that we should encourage abortion to decrease crime. Real answers are seldom simple enough to show in a graph.

  4. Richard,

    You assert that there is “zero evidence” for the proposition that “draconian sentences caused the reduction.” If that is true, then please explain what is causing the declining rate of violent crime?

    All the best.


  5. R.E.,

    Good point. But, if there is no correlation between declining crime rates and harsh sentences, what is causing the declining crime rate?

    All the best.


  6. The graphic is stark; but I’m having trouble getting past the fact that there are still an estimated 1.2 million violent crimes each year. That’s what — 1 in every 300 people?

  7. Dear Biker Chick,

    Thanks for your hopeful comment. I wish I shared your optimism. First, there is no way our country will spend the money necessary to address what you want to do even if we were to substantially reduce the number of people in prison. We are talking about huge sums of money. Second, if we ever got serious about getting at the root causes of crime many people would likely recoil in horror at what would be required to do. See, for example, my earlier post entitled “We can’t handle the truth.”

    All the best.


    PS I love your name, that is, “Biker Chick.” When Joan and I decided to get married in 1992, I was handling a massive Hell’s Angels drug conspiracy prosecution in Omaha. “Biker chicks” frequently attended the 28 days of hearings that I held on the wiretaps and bugs that the FBI employed. Anyway, I was riding up the elevator with the “Biker chicks” one day, when they congratulated me on my impending wedding. That had seen the legal notice of our marriage license in the newspaper. They were sincere and very nice. I recounted the story to our families. At our reception, one of Joan’s brothers gave us each a T-Shirt that said, “Forever together in leather.”

  8. CT,

    I share your dismay at the violence that is endemic to our society. That is one of the reasons why I am so conflicted about the current debate over reducing prison populations. There are very, very, very dangerous people out there.

    All the best.


  9. It’s funny that you say our country won’t spend the money even though the cost of outpatient treatment and counseling is much less than the cost of housing someone in a prison. Also, the “recoil in horror” would be far less with a little education, positive PR, and success stories such as this: http://www.npr.org/blogs/thetwo-way/2013/09/10/219295368/the-incredible-case-of-the-bank-robber-whos-now-a-law-clerk
    This guy did spend time in prison, but isn’t it great that given an opportunity and a positive focus, he turned his life around?

    PS: I was never a “BikerChick” in the gang sense. While I do have a couple of tattoos, I am among the legion of yuppie/newbie riders who delight in having sparkly Harley T-shirts and complain when the weather is bad. I am also bikeless right now, but I DO have my leather jacket!

  10. Judge,

    A few thoughts.

    A 14% absolute drop over 3 years — approximately 5% per year.

    Of those, how many could be reasonably attributed to repeat offenders; that is to say, the decrease was due simply to removing recividist individuals from access to the public and not to any halo/inter-individual deterrence? Removing recividist individuals, is, of course, one of several valid justifications for incarceration (the last bastion being the deontological rationale). (I know, this is law school shooting the bull stuff, but it has to be said.)

    Also, a similarly sized drop occurred between the five-year period of 1992 to 1997 (from 1.9MM to 1.6MM — 15%). My legal history is rusty (given that I was in high school then), but I don’t recall that there was as much strict sentencing in that time period as with a decade later.

    Third, between the years of 2004 and 2006, both absolute violent crime and violent crime per capita *increased* by 5%. Does that also reflect changes in sentencing?

    Last, I’m pretty sure we don’t have numbers yet, but do you think that the FSA of 2010 and _Dorsey_ , which are arguably “lenient sentencing”, will cause any kind of increase in the rate of violent crime?


  11. Dear J. Law,

    Here is my answer in brief:

    The last time “violent” crime rates were this low was in 1963. See, e.g., Daniel B. Wood, US Crime rate at lowest point in decades. Why American is safer now, Christian Science Monitor (January 9, 2012). Mr. Wood wrote:

    The declines are not just a blip, say criminologists. Rather, they are the result of a host of changes that have fundamentally reversed the high-crime trends of the 1980s. And these changes have taken hold to such a degree that the drop in crime continued despite the recent recession.

    Because the pattern “transcends cities and US regions, we can safely say crime is down,” says James Alan Fox, a criminologist at Northeastern University in Boston. “We are indeed a safer nation than 20 years ago.”

    He and others give four main reasons for the decline:

    *Increased incarceration, including longer sentences, that keeps more criminals off the streets.

    *Improved law enforcement strategies, including advances in computer analysis and innovative technology.

    *The waning of the crack cocaine epidemic that soared from 1984 to 1990, which made cocaine cheaply available in cities across the US.

    *The graying of America characterized by the fastest-growing segment of the US population – baby boomers – passing the age of 50.

    I believe Mr. Wood has it about right. All the best.


  12. Judge,

    Thank you for your thoughts. Those four reasons make sense facially. (I am curious, however, about the contribution of each factor to the overall decline.)

    Reasons 3 and 4 are systemic and are policy-neutral. In addition to being systemic, Reason 2 is also, I think, unarguably commendable. Reason 2 tends to also be more a priori, focussed on detection and prevention.

    It is only Reason 1 that is fundamentally policy-driven — specifically, a posteriori policy. It appears to be justifiable only consequentially, and accordingly raises issues of cost-benefit tradeoffs. I’m sure the grant rate is too low to be of any use (not to mention the changing standards with the AEDPA), but one wonders whether the base rate of 2255/2254 petitions should (would) have increased in the same time.

  13. Dear J. Law,

    I agree with you about the value of cost and benefit analysis regarding harsh federal sentencing and the increase in the federal prison population. That said, to do such a thing right would be an enormous and complex task that should be conducted independently by an entity that has no policy ax to grind.

    I don’t have long term data that is easily available to me on 2255 motions or 2254 petitions. Consequently, I won’t hazard a guess regarding a connection between these post conviction actions and harsher sentencing.

    All the best.


  14. Judge,

    Replying to a number of comments, you asked what could be causing the drop in crime rates, if not the harsh sentencing. Recent research suggests that a large portion is due to reduced lead levels (mainly from unleaded gas) in the environment. There is a strong correlation between lead levels and crime rates with a 20 year lag. The basic mechanism is that environmental lead affects the developing brain of children.

    This correlation has been found across countries and across regions in the US (which regulated leaded gas at different times). This article provides a good summary of the evidence (despite it’s “leftist” origins):

    P.S. I greatly enjoy your blog for a number of reasons, one being able to see the human side of the justice system i.e. that those who work in that area realize that they are dealing with the lives of human beings and can greatly affect those lives (perpetrators, victims, and the community at large). My only interaction with that system has been to serve on juries, which can lead to somewhat different ideas about how the justice system functions.

  15. Yet to be determined. To the extent there is a correlation (and no one is suggesting there is not in this case, only that correlation does not prove cause), there likely exists a common cause for both. Perhaps an attenuated common cause as the “true” cause works through other factors to get to one or both of these outcomes.

    For instance, what else was going on in the time interval of the graph? A Commenter suggests that it’s odd that the drop in violent crime rate also occurs during a Panic and failed recovery (OK, my characterization of his “great recession”). Perhaps the drop in rate is from no one having the money to go out, stay home, and so are less exposed to violent crime. That seems unlikely, but the data presented, here on on the FBI’s site, don’t address that at all.

    Eric Hines

  16. Mixing apples and oranges to no small degree (the Brits have a different definition of “violent crime”), but their violent crime rate is 1 in 2,000. We’re not so bad; although, we can do better.

    Eric Hines

  17. Not a Lawyer,

    I have seen that study before. I have two responses.

    First, you are no doubt also familiar with the hypothesis that increases in abortion rates drive down crime rates as shown by fitting two statistical curves together. I think the lead poisoning hypothesis is of a similar character. Don’t misunderstand me. I do not denigrate the difficult and significant statistical work that provides the foundation for both suggestion, but, in the end, it is statistical rather than empirical in nature. Put in simple terms, I trust criminologists more than I trust epidemiologists or wizards in econometrics.

    Second, let’s assume that the hypothesis is true. That does not rule out the probability that other factors including such things as harsh sentencing play a significant part in reducing crimes rates. Thus, Ronald Bailey, the award-winning science correspondent for Reason magazine, points out that.

    Drum [an advocate of the lead hypothesis] is right that exposure to lead increases the chances that a person will suffer the sorts of neurological damage that lowers their intelligence and lower intelligence is well-known to correlate with increased criminality. Reducing such exposures has no doubt contributed to our happily falling crime rates. But it is likely that other factors including more policing, more incarceration, less crack, increased concealed carry, and other such efforts to control crime have contributed as well.

    Ronald Bailey,Lead Poisoning Causes Crime? Reason Magazine (January 4, 2013).

    Thanks for taking the time to comment. And, by the way, I often find the views of people who are not lawyers to be especially illuminating.

    All the best.


  18. Sometimes the best available answer to a question like this one is “We don’t know.” As discussed below we can likely do better than that in this case, but have not arrived at a conclusive answer. The data does show a correlation between the harshness of sentencing and a lessening in violent crime. However, correlation is not causation. (Data also shows a correlation between height in the general population and the cost of a haircut – on average taller people pays less for haircuts than shorter people, – but being tall does not cause the price of a haircut and the price of haircuts does not cause a person to be tall.) The danger is that we might (and probably have) imprisoned thousands of people at immense financial and human costs, costs that go beyond taxes and affect the family and communities of the prisoners, e.g., more one parent homes, based on a mere correlation. We ought not to be saying, “We don’t know what the cause of lessening crime rates is, but based on nifty graphics we’ll continuing doing what we’re doing despite the costs.”

    We can get closer to an answer by controlling for variables other than the one involved in the correlation and testing alternative hypotheses. If, for example, we look at all jurisdictions with similar age demographics, e.g. all jurisdictions that went through the baby boom and then baby dip of the last 60 years and find that crime has gone down regardless of the severity of sentences in those jurisdictions then there is more likely to be a causal relationship between the change in demographics and the lessening crime rates than between the severity in sentencing and the lessening of the crime rates. Similarly, if we look at jurisdictions with the same harshness of sentencing and see a variation in crime rates that correlates with age demographics, there is more likely to be a causal relationship between the change in demographics and the lessening crime rates than between the change in sentencing and the lessening of the crime rates.

    It is particularly difficult to see a causal correlation between federal sentencing and overall decreases in crime rates if one looks at the variations in decline in violent crime rates among states and cities. Recognizing the possible disparity in actual sentences described in your posts about the use of enhancements in by the US Attorney offices in Nebraska and Iowa, it seems that federal sentencing is, on average, just as harsh in one jurisdiction as it is in others, yet the crime rate in New York has dropped faster than that of many other cities despite having dropped its prison population by a quarter since 1999 (“The curious case of the fall in crime,” The Economist, July 20, 2013,) while the violent crime rate appears to be ever rising in Chicago, even though the residences of both cities are subject to federal jurisdiction. Clearly something other than simple harsh of federal sentences must be at play.

    At this point those who study this topic can posit several factors – changing demographics, improved policing, better technology, incarceration, and even rehabilitation, and so on. Each may play a role. If we grant that there is actually a causal relationship for each of these, we must still ask, as others have suggested here, “At what cost?” Harkening back to my first post, even if we could show a causal relationship between abortion rates and crime rates (this seems to work fairly well with U.S. data, but does not pass muster across borders,) we would not be willing to bear the cost involved in encouraging abortions.

    Thanks for sparking this discussion.


  19. Dear Bob,

    Thank you for your thoughtful and frank comment. Additionally, I agree that we ought not rely on “nifty graphics” as an explanation for a complex phenomenon which is why I warned against oversimplification before reproducing the powerful image that prompted this discussion. Most significantly, I agree with you that we should ask about costs. In that regard, I ask only that we take a hard-headed and unemotional look at both sides of the cost equation.

    All the best.


  20. Your honor,

    If you attribute harsher sentencing to a decrease in crime rates, what aspect of the sentencing do you believe caused the decrease? In other words, is it through a general knowledge that there are harsher sentences for crimes, thus decreasing their incidence? Or do you believe it to be longer incarceration keeping recidivists from engaging in further crimes? Or something else?

  21. Dear Vicious,

    I do believe that harsher sentencing plays a part in reductions in crime rates. General deterrence (when the outsider decides that “I won’t commit the offense that Joe Doe Defendant got tagged with ’cause the penalty is too steep”) is a part of it, but I would guess that incapacitating offenders for very long periods of time plays a more significant role in reducing crime rates than does general deterrence.

    As I noted earlier, there are other factors at play as well. To be clear, I don’t claim to be able to quantify the impact of each factor. Indeed, I am not even sure there is a primary factor.

    But, if I had to guess, I would guess that very long prison sentences for roughly two generations of young people (incapacitation) coupled with very aggressive and relatively sophisticated law enforcement practices were, relatively speaking, the most important drivers in the reduction of crime rates.

    All the best.


%d bloggers like this: