The Supreme Court’s alleged pro-business stance and Kopf’s analytical skills test

I wrote a post some time ago entitled “Football is dead and so are civil jury trials in the federal courts.”  That prompted my extremely bright antagonist, Vince, to refer me to a New York Times article entitled “Corporations Find a Friend in the Supreme Court.”

In that article, the following was written:

Published last month in The Minnesota Law Review, the study ranked the 36 justices who served on the court over those 65 years by the proportion of their pro-business votes; all five of the current court’s more conservative members were in the top 10. But the study’s most striking finding was that the two justices most likely to vote in favor of business interests since 1946 are the most recent conservative additions to the court, Chief Justice Roberts and Justice Samuel A. Alito Jr., both appointed by President George W. Bush.

The study was prepared by Lee Epstein, who teaches law and political science at the University of Southern California; William M. Landes, an economist at the University of Chicago; and Judge Richard A. Posner, of the federal appeals court in Chicago, who teaches law at the University of Chicago.

I thought it might be fun if I actually read the law review piece rather than taking the paper’s summary at face value.   I was not entirely surprised to find that the Times cheery-picked the findings.

So, I decided to test my (few) gentle readers.  Figure 3, in yesterday’s post proposing a test of the analytical skills of the readers, comes from the Epstein, Landes and Posner law review article cited (rather breathlessly) by the NYT and relied upon by Vince.  (Let me hasten to say that this post is not a criticism of Vince.)

I will report some (but not all) of Epstein, Landes and Posner’s findings that were not reported in the Times piece.  I did my own “cherry-picking” to illustrate that care must be taken when generalizing about statistical data and the Supreme Court’s alleged pro-business stance.  For example:

 *The number of business cases the Supreme Court is taking has declined over time. “[T]he number of business cases in the Business Litigant Dataset averaged 36.6 per Term in the 1946 to 1952 Terms, compared to only 11 per Term since 2005; in the broader business dataset the decline is from an average of 46.1 to an average of 22.1.”  Lee Epstein, William M. Landes, & Richard A. Posner, How Business Fares in the Supreme Court, 97 Minnesota Law Review 1431, 1436 (2013).

*Comparing the Warren Court to the Roberts Court, there is a definite trend toward business support but there are anomalies. Referring to figure 3 (my skills test figure), the authors write: “Both [methods, judicial votes or outcomes] indicate a large drop in support for business during the 1960s, the era of the Warren Court, and a large rise in the Roberts Court. [However,] [t]he plunge in the early 2000s is a puzzle.”  Id. at 1448.

*While Alito and Roberts ranked 1 and 2 as being the most “pro-business,” Justice Jackson ranked third.  This is a startling. “As FDR’s Attorney General, he prosecuted businesses vigorously, in sync with the New Deal’s general hostility to business interests.” Id. at 1451-1452.

*“Even more surprising is that Ginsburg jumps to 11 (above Scalia and just below Powell and Stewart) in [one data set].”  Id. at 1452.

*Justices Kagan was not included in the rankings just discussed because the authors believed that she had cast too few votes.  Justice Sotomayor was excluded from one data set that was used to compute the ranking just discussed because the authors believed she had cast too few votes.  Id. at 1451.

*“Over the span covered by our study, business litigants  have generally fared worse in the Supreme Court than their nonbusiness opponents, receiving only 40% of the Justices’ votes and winning only 38% of the cases.”  Id. at 1470.

*“We find that decisions in favor of a business litigant over a non-business litigant are not uniformly conservative or the opposite decisions uniformly liberal; only 67.5% of the decisions (and 67.4% of the Justices’ votes) fit the pattern of business wins-conservative and business loses-liberal. And in subsets consisting for example of business cases involving civil liberties (such as a suit against a business for libel), the percentage is substantially lower.” Id.

*There is a trend toward appointing “pro-business” Justices whether the President is a Republican or Democrat. “Over time, Justices appointed by Democratic Presidents—not only those appointed by Republican Presidents—have become more favorable to business, consistent with the general growth in the public’s favorable attitude toward business.”  Id. at 1471.

*The Court tends to follow the “Tenth Justice” in business cases. “We used regression analysis to isolate additional factors that are correlated with (and might influence) Justices’ votes in business cases.  . . . [T]he Solicitor General’s position in a business case is highly correlated with the Justices’ votes in the case . . . .”

What do I think?  Well, what I think is not important.  You read the law review article, and decide for yourself what figure 3 and all the rest really means.  Incidentally, I am constantly annoyed by these three authors and their consistent failure in this and other papers to more specifically describe their statistical methods in understandable ways.  But, that’s a quibble because you can puzzle it out if you try hard enough.

What I will say is that the commentators to my earlier post have far better analytical skills than the New York Times.  (“Duh,” you say!)  It is all well and good to use punchy headlines like “Corporations Find a Friend in the Supreme Court” as the NYT did in order to further your ideological perspective or to sell newspapers.  It is an entirely different thing to seriously examine the data in an intellectually honest fashion.  That is hard work and it frequently leaves you with more questions than answers.



Genghis Khan and sentencing

Photo credit:  frans 16611"s photostream per Creative Commons license.

Photo credit: frans 16611’s photostream per Creative Commons license.

The authors of The Behavior of Federal Judges  A Theoretical and Empirical Study of Rational Choice, discussed in the last post, looked at the important question of the behavior of federal district judges when they sentence people.  Among other things, the authors concluded that the political party of the President who nominated the judge was a good proxy for judicial ideology.  The authors then looked to the Transactional Records Access Clearinghouse (TRAC) at Syracuse University to provide data about the sentencing decisions of individual judges.  After that, the authors matched up sentencing decisions by individual judges and the party affiliation of the President who nominated the judge.  The authors then ran a statistical analysis (once again using TRAC resources) to determine if there were differences in sentence lengths between judges nominated by Republican Presidents and judges nominated by Democratic Presidents.  Among other  things, the authors found that “there is evidence of ideological influence, with judges appointed by Republican Presidents generally imposing heavier sentences when other influences are corrected for.  The ideological influence is modest, however, . . . .”

For now, I intend only to highlight four things.  I hope to stay out of the weeds.  Here goes.

Point One

Federal courts have a lot of sentencing data, but there is a big problem collecting the data and putting into a useable form.  Because the TRAC data contains identifying information for individual judges, it is a unique and valuable resource.  At present, there is no other such information for the entire federal judiciary that is readily available.  The effort it took to accumulate this data and then put it into a useable form was truly gargantuan.  I laud TRAC for its efforts.

Point Two

There are problems with the TRAC information.  See Tracking TRAC’s New Sentencing Data, 25 Federal Sentencing Reporter No. 1 (October 2012).  Many of those problems are, however, vastly overblown as evidenced by the fact that attacks on TRAC are generated largely by defense lawyers.  They have a motive to trash anything that suggests that judicial discretion at sentencing is a bad thing.

The foregoing said, real problems persist.  In my view, the most significant problems with TRAC reports are these:  (1)  one has to pay for the TRAC data and the cost is not insubstantial;  (2) a user agreement must be signed that limits how the user may use the TRAC reports; (3) the TRAC reports do not separately address departure and variance motions that frequently drive sentencing outcomes.

There is a better way.  The United Sentencing Commission has data identifying each federal district judge by name and it has specific data for each sentence imposed by that judge.  In short, the Sentencing Commission data is more refined than the TRAC data.  Unfortunately, the Sentencing Commission will not normally release this data.  The only way to obtain the data is for the chief judge of the district to request the release of the data or for each identified judge to consent to the release of the data.  Since that data was generated by a judge publicly sentencing a defendant, there is no good reason not to make that data freely available to anyone who wants it.

To date, only the District of Nebraska (my district) has made that data freely accessible to the public.  See  United States District Court District of Nebraska, Judges’ Sentencing Data.  Among other things, the Sentencing Commission data for the judges of the District of Nebraska shows that for “drug cases, there appears to be a marked difference between sentences imposed in Lincoln (higher) and sentences imposed in Omaha (lower).”   Richard G. Kopf, Judge-Specific Sentencing Data for the District of Nebraska, 25 Federal Sentencing Reporter No. 1, p. 51 (2012).*  Why an analysis of the sentencing practices of judges in two courthouses no more than 60 miles apart reflects big differences in sentence outcomes is a grave cause for concern.

Point Three

It is critically important to remember that the authors of The Behavior of Federal Judges were looking at the data from a national perspective.  Thus, whether ideological differences matter more than modestly in individual districts does not appear to have been clearly addressed.

If you are a defendant, the existence of only modest ideological influences, when derived from an average of all the district judges in the nation, is not comforting.  In the district where the sentence will be imposed, a defendant wants to know whether he has drawn a Republican Genghis Khan.

The following chart (showing five years of data ending in September 2012) obtained from TRAC (for all sentence types) illustrates just how widely divergent sentence outcomes can be in the same district:


(Although I have long ago shed my affinity for the Republican party, it appears that I am, nevertheless, the District of Nebraska’s Genghis Khan.)

Point Four

Whether one uses the TRAC data or data from the Sentencing Commission, when it comes to sending people to prison it is important to concentrate on what is going on in each district and in each courthouse within each district.  Given the broad and national scope of their endeavor, it is perfectly understandable why the authors of The Behavior of Federal Judges did not drill down to the district level.  For federal sentencing judges, we lack any similar excuse.


*The Federal Public Defender lobby suggests that the Nebraska disparity can be explained by differences in the use of Rule 35(b) sentence reduction motions that come after initial sentencing–Lincoln prosecutors use them more and Omaha prosecutors use them less, so say the FPDs.  However, my analysis of the data (cases numbers, section 5K1.1 motions, and Rule 35(b) motions), which I have shared with our FPD, suggests that is highly unlikely that any such difference is substantial enough to explain away the disparity.   So far, I have heard no rebuttal to my analysis.  As I said earlier, for this post, I want to stay out of the weeds.  So, I will go no further on this point.

Does ideology matter when district judges sentence?

Anyone who is keen to study what judges actually do must read a new empirical analysis by giants in their respective fields of political science, economics and law. See Lee Epstein, William M. Landes and Richard A. Posner, The Behavior of Federal Judges  A Theoretical and Empirical Study of Rational Choice, Harvard University Press (January 2013).* The study looks at the behavior of federal district judges, circuit judges and Justices of the Supreme Court.

As pertinent to this blog, The Behavior of Federal Judges contains a systematic statistical analysis of much of the existing data that has been collected regarding the activities of district judges. The effort is groundbreaking. As the authors point out, the behavior of district judges has largely been ignored.  In particular, there has been little empirical (statistical) analysis of what district judges actually do.

I will have more about this book in later posts as it regards district judges, but for now I want to highlight the authors’ analysis of sentencing in the district courts.  Here is the essence of the authors’ conclusions regarding sentencing:

To summarize our analysis of sentencing, there is evidence of ideological influence, with judges appointed by Republican Presidents generally imposing heavier sentences when other influences are corrected for. The ideological influence is modest, however, consistent with the overall result of the analysis in this chapter that ideology plays only a small role at the district court level, even though district judges have considerable discretionary authority.

Derived from their statistical analysis, here are the numbers that the authors provide to support the foregoing conclusion:

Differences between Rs and Ds persist when district effects and the other independent variables are held constant, although the differences are small. On average Rs can be expected to sentence defendants to 6.5 more months in prison than Ds in organized-crime drug cases (a 6.6 percent increase), 4.4 months in other drug cases (4.8 percent), 3.5 months in weapons cases (6.4 percent), and 1.1 months in immigration cases (17.7 percent).

While the ideology of the district judge may matter only modestly when viewed from a national perspective, if a Court of Appeals is “politically” homogeneous, then one can expect large differences in how district judges sentence in the circuit where the members of the court of appeals are simpatico as compared with how district judges sentence in a circuit where the members of the court of appeals lacks political homogeneity.  Thus, consider this fascinating and striking comparison between the Eighth Circuit (where I hang out) and the Ninth Circuit:

The results are consistent with our previous findings. In three of the four categories, the fraction of Rs in the court of appeals significantly lengthens prison sentences. In organized-crime drug cases, we predict that a district judge in the Eighth Circuit, where more than 80 percent of the [appellate] judges were Rs in the relevant period, would sentence a defendant to 25 more months (a 26 percent increase) than a district judge in the Ninth Circuit, where only 40 percent of the [appellate] judges were Rs. The difference in other drug cases is 19 months (a 22 percent increase), in weapons offenses 8 months (15 percent), and in immigration cases 4 months (65 percent).**

In the coming days, I will offer a critique of the data set used in the authors’ analysis of sentencing at the district court level. I will also suggest a way to address that data problem.  Additionally, I will focus on empirical questions pertaining to how district judges in the same district sentence–the question of intra-district disparity.  For now, however, it is enough to state that this book should be read by anyone who is serious about studying judicial behavior at the district court level.


*I have earlier written in this blog about my career law clerks and their incalculable value to me.  One of those clerks (Jim) tipped me off about this book after reading an excellent review. See Cass R. Sunstein, Moneyball for Judges, The Statistics of judicial behavior, New Republic (April 9, 2013).

** While it is not entirely clear, this analysis appears to be based upon sentences imposed after a trial rather than after a guilty plea.  I need to look into that question in more detail because it is very important.

%d bloggers like this: