« More details on the Monica Conyers case (and its future?) | Main | Accused murderer of Chelsea King avoided going back to prison after 2007 parole violations »

March 12, 2010

Important and worrisome new multivariate analysis from USSC of post-Booker sentencing patterns

I am pleased to see that the US Sentencing Commission has released this potent and important new sentencing data analysis, which is titled "Demographic Differences in Federal Sentencing Practices: An Update of the Booker Report’s Multivariate Analysis."  I am not pleased (though also not especially surprised) about what that analysis reveals about some post-Booker sentencing trends.  Specifically, according to this new USSC report:

This report focused on three separate time periods which together spanned the time between May 1, 2003, and September 30, 2009.  The Commission found a correlation between the length of sentences imposed on some groups of offenders and the demographic characteristics of those offenders.  These differences were not present in all time periods under study and differed in magnitude in the time periods in which they were observed....

Based on this analysis, and after controlling for a variety of factors relevant to sentencing, the following observations can be made: 

  • Black male offenders received longer sentences than white male offenders. The differences in sentence length have increased steadily since Booker

  • Female offenders of all races received shorter sentences than male offenders. The differences in sentence length fluctuated at different rates in the time periods studied for white females, black females, Hispanic females, and “other” female offenders (such as those of Native American, Alaskan Native, and Asian or Pacific Islander origin). 

  • Non-citizen offenders received longer sentences than offenders who were U.S. citizens. The differences in sentence length have increased steadily since Booker. 

  • Offenders with some college education received shorter sentences than offenders with no college education. The differences in sentence length have remained relatively stable across the time periods studied. 

  • The data were inconsistent as to the association between an offender’s age and the length of sentence imposed.

Any serious student of the history of prosecutorial and judicial sentencing discretion knows that with any increase in discretion power will come some increase in disparate outcomes and the risk that disparities are influenced by non-legal factors that we might wish would not (but always does) influence how imperfect humans exercise their discretion.  These realities always demand careful assessment and sober reflection, but they also always demand a careful reaction and sober consideration of what remedies are possible and what remedies might prove worse than the disease.

Though I will need to review this new report and its data closely before developing a detailing reaction to these findings, I will start by suggesting that my own anecdotal post-Bookerexperiences suggest that economic realities and the disparate efforts of prosecutors and defense counsel may account for worrisome disparities more than the determinations of sentencing judges.  I fear there my be some systematic and structural biases, often influenced by socio-economic realities, that can result in prosecutors charging and bargaining a bit harder on certain types of offenders and that can result in defense counsel developing better mitigating arguments for certain types of offenders. 

March 12, 2010 at 12:05 AM | Permalink


TrackBack URL for this entry:

Listed below are links to weblogs that reference Important and worrisome new multivariate analysis from USSC of post-Booker sentencing patterns:


i think your last sentence is dead on. all in all, i like the post-booker world---it gives my clients who are more deserving a chance at a lower sentence that they never would have had under the guideline regime. i never have quite bought the entire "mandatory" guideline claim, but at the very least the old guideline system had district judges scared to act independently and thoughtfully. now, there is a fighting chance in many cases that i can convince them to so act, and they sometimes do, certainly many more times than before booker, gall, and kimbrough, and i represent a lot of clients who are non-citizens.

we need to work on the remaining structural problems and biases, for sure. i do not trust the ussc, however. they want to be back in control and if they get there i expect they will, as before, in the name of reducing disparity, reduce discretion, reduce fairness, and increase sentences. the commission's mostly hateful prohibition or discouragement of almost all grounds for downward departure and its encouraged list of permissible grounds for upward ones tells us all we need to know about their priorities. whatever the answer is, it is not the commission and it is not in stricter guidelines or guideline floors.

Posted by: big bad wolf | Mar 12, 2010 12:24:47 AM

Nobody should be surprised by this. Senator Kennedy and the other liberals who helped forge the SRA back in the mid-80's weren't as stupid as you thought.

So have you got what you wanted pro-Booker apologists?

Disparity for all, special treatment for some. What a result.

Posted by: Ferris Bueller | Mar 12, 2010 8:34:30 AM

With the exception of the gender disparity, all the others can be filed under "duh." It is no secret that prosecutors and judges routinely show leniency to those in better socioeconomic conditions, and outright breaks to those who are privileged. Those who are given harsher sentences, on balance, have lower socioeconomic status than their counterparts.

In the state criminal system, the disparities are atrocious, as several stories on this blog have indicated. (Andrew Sullivan having all charges dropped for pot possession, the facebook diva getting to plead to negligent homicide and sentenced to a whole six months, et al.) The guidelines and the DOJ mercifully have cut down the disparity to some extent in the federal system, but as the study shows, they still exist with a passion.

Posted by: Res ipsa | Mar 12, 2010 9:03:06 AM

My guess, Ferris, is that only you and Supremecy Clause (and maybe a few others with fake names) would be quick to call Senator Kennedy and the other liberals "stupid." Are we to view your comment as a statement that you view Senator Kennedy and the other liberals who ran Congress in the mid-80s were smarter that the Republicans in control in 2005-2006 who failed to respond to Booker by making the guidelines mandatory again?

More fundamentally, Ferris, the hard question for someone who thinks that most of the federal guidelines are too harsh is whether it would be better for nearly all drug dealers and child porn downloaders to get 15 years (a simplified account of what harsh mandatory guidelines produce), or instead for the more white/female/educated of these folks to get 8 years and the more black/male/under-educated to get 12 years (a simplified account of Booker's impact). This is the essential question at the heart of any debate over whether Booker made federal sentencing better or worse. Obviously, if you think 15 years is/was the right/best outcome for offenders, all the reduced sentences are a concern and the evidence of increased disparity adds to that concern. But if you think that, say, 10 years is much closer to the right/best outcome in these cases, then the post-Booker world has enabled improved (but still not ideal) sentencing outcomes.

I consider excessive imprisonment to NOW be a MUCH bigger (and MUCH more expensive) problem in a nation "conceived in liberty" than increased disparity. Consequently, despite these troublesome data, I still think Booker amounted to a net improvement in federal sentencing practice and policy. And apparently so does most everyone else: there has been no serious movement in Congress for a statutory fix for now over 5 years.

You are right that Senator Kennedy and the other liberals who helped forge the SRA back in the mid-80's were quite bright because THEN disparity (along with a lack of honesty/transparency) --- not severity --- was likely the biggest problem in sentencing practice before the SRA. But now, circa 2010, when our prisons are bursting at the seems and our treasuries are int he red, I think we must now be worrying a lot more about severity than about disparity.

Posted by: Doug B. | Mar 12, 2010 10:31:19 AM

I love the fact that "big bad wolf" immediately calls into question the integrity of the sentencing commission. That is EXTREMELY unfair. I just finished reading the report, and one important part of it is, the Commission has put the actual data output in the appendix and also shows exactly how the data was created. I guess if you have access to the data, you can "check" the Commission's work if you believe they are pulling the wool over your eyes. It seems to me that if the Commission wanted to hide something, they wouldn't have done that.

That being said, anyone who follows federal sentencing shouldn't be surprised. All you have to do is look at the data since the advisor guidelines came to be and see that the type of offenses that are mostly populated with white people are the ones that judges are going outside the guidelines. Instead of accussing the Commission of being dishonest, maybe "big bad wolf" should use common sense. It isn't a matter of WHETHER disparity exists, it is now a matter of "is it worth discriminating against black men in order for judges to feel important?" If I was a black man (and our President is) I would be very disturbed by this report. When judges were "hemmed in" during the Protect period, there was no or little discrimination and now there is...doesn't take a rocket scientist to figure out the problem

Posted by: Kelly | Mar 12, 2010 11:16:19 AM

I am no friend of the present Sentencing Commission, viewing its majority as the cat's paw of the defense bar. However, the idea that they or federal district judges are racist, is preposterous.

Demographic statistics do not prove identity group animus. The prison population is overwhelmingly male. The composition of death row is even more overwhelmingly male. Does this mean the "system" has it in for males?

What it actually means is that males commit proportionately vastly more serious and violent crime. I am not an anthropologist so I am not going to go into the reasons for this fact. The point is simply that, as anyone knows who has actually seen the federal courts operate over time, federal sentencing is not based on racial or gender animus. It is based on defendant behavior.

Posted by: Bill Otis | Mar 12, 2010 5:05:32 PM

"kelly," i did not call the commission's statistical work dishonest. i said that i did not trust the commission to REMEDY disparity in any way but by pushing, as it has relentlessly in the past, for higher sentences set by it. as doug b. says above, severity and overimprisonment are far bigger problems now than disparity is. i don't expect the commission to do much to address those problems. disparity as the commission uses it is a stalking horse, i believe, for giving the commission and its guidelines more authority.

i represent a lot of non-white clients, all cja clients, and they are doing better---not great, but better (a few much better)---under the advisory guidelines regime.

Posted by: big bad wolf | Mar 12, 2010 5:54:27 PM

I haven't read the report, but I am assuming the USSC does not simply issue reports with the "duh" conclusions that the prison population is disproportionately male/non-white. I assume from terms such as "multivariate analysis," that the USSC's analysis purports to control for disproportionate rates of offending, etc., and determine if, holding all else equal, sentence lengths vary in a way that can be attributed to gender, race, etc.

Naturally, such analysis is probably as much art as science, and likely involves discretionary factors in designing the study -- in other words, the USSC may very well be bad at controlling for other factors and isolating the effect of gender/race/etc. But, that said, I don't think it is a very effective critique to imply that the whole report can be explained by facts such as "males commit proportionately vastly more serious and violent crime," as it seems obvious to me that the USSC would have at least *tried* to control for such underlying disparities.

Posted by: Observer | Mar 12, 2010 6:18:25 PM

Prof. Berman, I hope, attended the high school social studies class on the superior, not co-equal branch, the legislature. It can fire the executive. The judiciary is its retriever dog, plunging into bog and swamp to retrieve shot down ducks from between rolled logs. That is why it is sacrosanct, and above criticism. When powerful non-lawyer Representatives make threatening and critical remarks about the Supreme Court, the hierarchy find a pretext to get rid of the powerful representative. The number of collateral corruptions for pretextual expulsion is infinite for everyone on earth. No one missed that lesson. Even if the figurehead elected official is not a lawyer and decides to be candid about the Supreme Court, the lawyers on his staff will straighten him out by pointing to that example.

As to fake names, no. That is the real name of a fictional character. And Prof. Berman has all the personal information about its author that he wanted. There is an open invitation to lunch in New York. However, the Professor would likely be bored by the inferior intelligence and knowledge of this not a lawyer person.

Posted by: Supremacy Claus | Mar 12, 2010 8:48:55 PM

Prof. Berman knows that I have many times called the Guidelines and their subsequent suppression of crime a tremendous lawyer achievement. This is a sincere compliment and expression of admiration, not sarcasm.

A drop of 40% in the crime rate likely added $trillions in value to the economy, and this is a tremendous illustration of lawyer profit seeking, where the lawyers provided a million percent return on investment in their salaries. No doubt, this added value (profit) contributed massively to the boom times of that decade. That judgment is made from its outcomes, not from its authors. Nor have I ever called any lawyer stupid. The opposite is true.

The word, dumbass, is not an epithet. It a technical term of art. Modern people with IQ's of 300 are made into lawyer hierarchy zombies, and come to believe that minds can be read; the future accidents can be forecast; 12 stranger can detect the truth by using their gut feelings, and many other Medieval superstitions. I see the lawyer as a victim of cult indoctrination. If the public is oppressed by this criminal cult, the lawyer is doubly so, and the judge triply so. When this hierarchy is removed, the profession will be great, half its current size, making 4 times as much by adding value, not by plunder, and esteemed 10 times as much. The lawyer will thank the Supremacy, and put up statues to her.

Posted by: Supremacy Claus | Mar 13, 2010 2:20:33 AM

There is NOTHING the commission can do about this. This study controls for what place defendants start with when the judge pronounces his/her sentence. It says that FROM THAT POINT, Judges sentence black males more harshly than white males by a huge magnitude. It doesn't call judges racist, but maybe factors that are tied to race may be influencing judges decisions. I would advise all of you to actually READ the report and see what it says and doesn't say before making comments. From reading the comments, I don't think any of you (besides Kelly) have actually read it and are just reacting.

Posted by: Chelsea | Mar 13, 2010 7:58:56 AM

I perused most but not all the report and have the following observations:

Longer sentences for black males can be explained by the following;

1)Minority defendants commit crimes with higher penalties-crack cocaine vs cocaine powder;
2)Minority defendants have criminal histories with more instances of violent behavior:
3)Minority defendants have less history of gainful employment.

Items two and three are factors that judges typically take into account when imposing a sentence.

Another factor militating against racial animus in federal sentencing is the fact that minority female offenders received shorter sentences than white female defendants.

Posted by: mjs | Mar 14, 2010 12:35:10 PM

mjs --

Please. Don't you know that actual facts are not allowed? The report is designed to imply that the system is racist, and now you have the bad taste to spoil the party. For shame.


Board of Political Correctness

Posted by: Bill Otis | Mar 14, 2010 2:51:34 PM

Mr. Otis:

Consider me chastised and chastened.

I should have taken a page from the Mike Nifong playbook-never let the facts get in the way of a politically correct story.

Posted by: mjs | Mar 14, 2010 4:44:52 PM

Again, I will ask mjs and Bill Otis, have you actually read the report? The report takes LOTS of space to say that there are unmeasured factors that have racial implications and may explain the differences in sentencing. It goes out its way to say the opposite of what you are saying it implies...so again...DID YOU ACTUALLY READ THE REPORT?

Posted by: Kelly | Mar 15, 2010 7:08:08 AM

Kelly --

No, I did not read the report. And if as you say, it explictly denies that the federal sentencing system is racist, good for it. The fact remains, though, that reports about racial disparities are frequently used to state or imply exactly that, as the Sentencing Commission full well knows.

And while I did not read the report, I spent 18 years as an AUSA and argued over 100 cases in the federal courts of appeals, dozens of them about sentencing. So if you're waiting for me to bow to your superior authority because you read one report, you'll be waiting a long time.

Posted by: Bill Otis | Mar 16, 2010 2:52:54 AM

"I spent 18 years as an AUSA and argued over 100 cases in the federal courts of appeals, dozens of them about sentencing."

The plural of anecdote...

Posted by: Michael Drake | Mar 17, 2010 12:51:09 PM

As a former "academician" (but not a reviewer of this report), I have read the Commission's report and conclude that the analyses reported therein are conceptually and statistically flawed, and sufficiently so that the results cannot be taken at face value. I suspect that most readers will not have the background to assess the statistical analysis, therefore I will focus on the conceptual flaws. Actually, many of the statistical issues are a result of the conceptual problems, so it is a good place to start the discussion.

The majority of conceptual issues flow from the grouping together of many disparate offense categories for a single analysis (how else to you get 178,000 observations?). Such a grouping assumes that the effects of the explanatory variables are constant across these categories. Would we expect that the effects of race and gender on sentencing, if any, are the same in immigration, white collar, drug trafficking, and child pornography cases? I would not expect that, but that is what the model specification posits. Similarly, the model treats all departures for substantial assistance as equal. This specification assumes that offenders whose sentence falls within offense level 34 receive the same percentage decrease in their sentence as those offenders whose presumptive guideline sentence falls within offense level 24, offense level 12, or offense level 2. It also assumes that the effect of such departures is the same regardless of offense type. Both assumptions may be correct, but I am skeptical and would like to see some supporting evidence. Besides, the coefficients for this variable in the estimated models (Appendix C) suggest that offenders receive a greater than 100% decrease in their sentence for this type of departure. That was an immediate red flag as I looked over the results.

One could go down the list of explanatory variables and ask similar questions about how their effects are or are not constant across offense categories. The larger point is that this analysis is an effort to do a "one size fits all" analysis. I think anyone experienced with sentencing data knows that doesn't work very well. The Commission publishes annual reports with breakouts by offense type. Why wasn't that done here? On the statistical side, the grouping of offense categories, with their very different ranges, led the analysts to use the log of the sentence received. Had they done separate analyses by offense category, the analysts might have found that there was no need to "linearize" the data by using the log of sentence. A statistical correction of this type cannot be used to overcome a theoretical flaw in the model.

In conclusion, I don't believe the reported results and would not believe them without a more thorough statement of the theoretical basis for the Commission's model and a more thorough justification of their statistical methods.

Posted by: YellowDog | Mar 17, 2010 1:54:51 PM

Post a comment

In the body of your email, please indicate if you are a professor, student, prosecutor, defense attorney, etc. so I can gain a sense of who is reading my blog. Thank you, DAB