The Behavior of Federal Judges

Chapter 5
The District Courts and the Selection Effect

INTRODUCTION

Federal district judges have received less attention in academic studies of judicial behavior than Supreme Court Justices and court of appeals judges. There are several reasons. There are no datasets comparable to those that have been created for Supreme Court and court of appeals decisions (of course, this is a result of lack of academic interest as well as a cause). District court decisions do not create precedents, and precedents are a particular focus of legal training and scholarly interest; also, non-precedential decisions have less visible policy impact than decisions of higher courts. A substantial fraction of cases filed in the district courts (many by persons who do not have legal representation) have no possible merit and so really are just noise in the data. And information about district court decisions is hard to come by, because most are decided without written opinions, often on procedural grounds and rarely after a trial.

Yet these courts are the largest component of the federal judiciary, deciding 6.9 times as many cases as the courts of appeal and about 3000 times as many cases as the Supreme Court.1 There are 3.8 times as many authorized district judgeships as court of appeals judgeships (678/179). And district judges have a great deal of discretionary authority, the exercise of which can be reviewed by the courts of appeals to only a limited extent—for example, authority to decide whether a witness at a bench trial, or at some other evidentiary hearing in which the judge is the trier of fact, is credible, authority to admit or exclude evidence at a trial or other proceeding, and authority to pick a criminal sentence within the sentencing limits set by Congress.

Our analysis makes use of five datasets. One is the Sunstein database that we used extensively in the preceding chapter. Its 4498 decisions are appeals from 962 district judges,2 and for this chapter we have added information about the cases when they were at the district court level: whether a case was decided with or without a trial; whether if there was a trial it was a jury trial or a bench trial; whether if it was a jury trial the judge took the case from the jury before the jury began its deliberations and decided it himself; the ideological direction of the district court decision; whether the plaintiff or the defendant prevailed; who the district judge was and whether a Republican or a Democratic President had appointed him; the judge's own party affiliation if known; his senatorial courtesy score;3 and his personal characteristics, such as race, sex, age, law school attended, professional experience, and date of appointment to the district court.

One use we make of these data, along with data from earlier chapters, is to test the hypothesis that as cases move up the judicial hierarchy, ideology plays an increasingly important role in decision-making.4 We discuss that selection effect in a later section of this chapter, along with a related issue, which we call the paradox of discretion: the more scrupulous a court of appeals is in adhering to standards of judicial review that command deference toward certain kinds of district court ruling, such as rulings on admission of evidence and on the credibility of witnesses, the more freedom district judges have to let their own ideological preferences influence their decisions. We test this proposition with the aid of a second dataset, compiled by Corey Yung (a variant of his dataset that we used in chapter 4), which ranks the courts of appeals by "activism," which he defines as the inverse of deference to district court decisions when the applicable standard of appellate review is deferential.

We also use the Sunstein dataset to explore reversals of district court decisions. A study of reversals in one circuit, the Eighth, found that the court of appeals for that circuit— a court dominated by judges appointed by Republican Presidents—reversed decisions by district judges appointed by Democratic Presidents at a significantly higher rate than those by district judges appointed by Republican Presidents.5

A third dataset that we use to explore the exercise of discretion by district judges consists of data compiled and published by the Sentencing Commission on federal criminal sentences. District judges traditionally were given almost unlimited discretion in picking the defendant's sentence within the normally broad range between the congressionally prescribed minimum and maximum sentences for a given crime. The federal sentencing guidelines promulgated by the Sentencing Commission pursuant to the Sentencing Reform Act of 1986 greatly curtailed district judges' sentencing discretion, but in 2005 the Supreme Court in the Booker decision restored much of it in the name of the Constitution.6 The exercise of sentencing discretion is an area in which ideology might be expected to influence the behavior of district judges, and in the final section of the chapter we use the Commission's post-Booker sentencing data to explore that possibility. We use still another dataset, of individual federal district judges' sentences, compiled by the Transactional Records Access Clearinghouse of Syracuse University (TRAC), to explore further the exercise of sentencing discretion by federal judges.

Our fifth dataset consists of 3893 cases in which motions were filed in district courts to dismiss a suit for failure to state a claim, pursuant to Rule 12(b)(6) of the Federal Rules of Civil Procedure.7 The data cover the period from May 2006 to June 2010, a year before Bell Atlantic Corp. v. Twombly8 and a year after Ashcroft v. Iqbal, two Supreme Court decisions that made it easier for district courts to dismiss cases at the outset, before any pretrial discovery. We use these data to determine the impact of ideology, district court caseload, the prospect of reversal on appeal, and the changes in pleading standards brought about by Twombly and Iqbal on the propensity of district judges to grant motions to dismiss.9

1 Calculated from U.S. Courts, "Federal Judicial Caseload Statistics 2009," www.uscourts.gov/Statistics/ FederalJudicialCaseloadStatistics.aspx (visited Feb. 8, 2012) and Chief Justice John Roberts, "2009 Year-End Report on the Federal Judiciary," www.supremecourt.gov/publicinfo/year-end/ 2009year-endreport.pdf (visited Jan. 17, 2012).

2 Of the 962 judges, 443 were active in 2008 and 519 had either retired or taken senior status before 2008. We eliminated cases decided by an administrative agency, a magistrate judge, or a bankruptcy judge rather than a district judge, and we eliminated 11 cases in which we could not identify the judge.

3 These other ideological proxies (judge's party affiliation and senatorial courtesy score) turn out not to alter significantly the results based on the cruder R/D classification, even though Presidents sometimes do appoint district judges who are of a different party or no party. So we generally don't report the results of the other measures.

4 For previous tests of this hypothesis, see Jeffrey A. Segal, Harold J. Spaeth, and Sara C. Benesh, The Supreme Court in the American Legal System 201 (2005); Daniel R. Pinello, "Linking Party to Judicial Ideology in American Courts: A Meta-Analysis," 20 Justice System Journal 219 (1999). As noted in chapter 2, Pinello's survey finds that the effect of ideology on judicial votes is about twice as strong in the Supreme Court as in the courts of appeals; we'll see that it's even weaker in the district courts than in the courts of appeals.

5 Robert Steinbuch, "An Empirical Analysis of Reversal Rates in the Eighth Circuit during 2008," 43 Loyola of Los Angeles Law Review 51 (2009).

6 United States v. Booker, 543 U.S. 220 (2005).

7 We began with the data that Kendall W. Hannon used for his article "Much Ado about Twombly? A Study of the Impact of Bell Atlantic Corp. v. Twombly on 12(b)(6) Motions," 83 Notre Dame Law Review 1811 (2008). Being limited to cases decided between June 2006 and December 2007 that cited Twombly or the decision that it overruled, Conley v. Gibson, 355 U.S. 41 (1957), those data do not include decisions subsequent to the Supreme Court's decision in Ashcroft v. Iqbal, 129 S. Ct. 1937 (2009). We supplemented the data by searching Westlaw for all rulings between May 2006 and June 2010 on Rule 12(b)(6) motions, regardless of whether they cited Twombly, Conley, or Iqbal, and drawing a random sample of 25 percent of these rulings. We excluded cases that were not decided by a district judge.

8 550 U.S. 544 (2007).

9 Hannon, "Much Ado about Twombly? A Study of the Impact of Bell Atlantic Corp. v. Twombly on 12(b)(6) Motions," finding little effect of Twombly; Patricia W. Hatamyar, "The Tao of Pleading: Do Twombly and Iqbal Matter Empirically?" 59 American University Law Review 553 (2010), finding that Iqbal but not Twombly affected dismissal rates (but her study is limited to rulings in the three months after Iqbal); and Andrea Kupperman, "Review of Case Law Applying Bell Atlantic v. Twombly and Ashcroft v. Iqbal: Memo to the Civil and Standing Rules Committees of the Judicial Conference Advisory Committee," www.uscourts.gov/uscourts/RulesAndPolicies/rules/Iqbal_memo_072610.pdf (visited Dec. 9, 2011), finding, from a survey of a large number of cases, that Iqbal has not "dramatically changed the application" of pleading standards."

Chapter 5 Datasets

CSV Comma Delimited Text (ASCII)

DTA Stata (Version 10 or later)

POR SPSS Portable File (Version 13 or later)

RDATA R Data Format (Version 2.0.0 or later)

SAV PSS Data File (Version 13 or later)

XLSX Microsoft Excel Worksheet (Version 2007 or later)

XPT SAS Transport Format
Ch5SunsteinDCt
Ch5SunsteinDCt.csv.zip
Ch5SunsteinDCt.dta.zip
Ch5SunsteinDCt.por.zip
Ch5SunsteinDCt.Rdata.zip
Ch5SunsteinDCt.sav.zip
Ch5SunsteinDCt.xlsx.zip
Ch5SunsteinDCt.xpt.zip

Ch5TwomblyIqbal
Ch5TwomblyIqbal.csv.zip
Ch5TwomblyIqbal.dta.zip
Ch5TwomblyIqbal.por.zip
Ch5TwomblyIqbal.Rdata.zip
Ch5TwomblyIqbal.sav.zip
Ch5TwomblyIqbal.xlsx.zip
Ch5TwomblyIqbal.xpt.zip


For a full listing of available datasets, view the Dataset Inventory.