Based on extensive and very current data (through July 2012), an analysis by the Transactional Records Access Clearinghouse (TRAC) has found wide variations in the criminal caseloads of individual federal district court judges, in some instances even among the judges serving in the same courthouse.

The surprising extent of the variations documented in TRAC's study raises a series of concerns surrounding a rarely examined question: are there serious overall problems in the way the government as a whole is managing the federal prosecution of criminal cases?

TRAC's latest judge-by-judge analysis — using newly updated sentencing records on over 400,000 defendants^[1] — followed the criminal caseloads of 909 federal district court judges for the past 70 months. See TRAC's Judge Info Center for a listing of the individual judges included in the study^[2], along with the number of defendants each sentenced.

Among the 909 district court judges whose records were examined, however, about half had not actively served during the entire 70-month study period. Thus, in the interest of better comparability, this report focuses on the 430 judges who had actively served for the entire 70 months^[3]. (For statistics on all 909 judges see Appendix Tables A and B.)

For these 430 active judges located in 179 courthouses across the country, this report discusses two particularly striking criminal caseload patterns that have emerged from TRAC's case-by-case data examination.

Surprising Variation in Criminal Caseloads Within Some Courthouses

As noted briefly above, even when the analysis was limited to the active judges who had served for the entire study period, the data disclosed surprising workload variations. In ninety separate courthouses there were two or more active judges, allowing for a comparison of the variation in caseloads within the courthouse. Among these ninety, there were four courthouses where the criminal caseload for one or more of the judges serving in them was twice the caseload of their colleague(s). The courthouses with this extreme judge-to-judge variation in caseload were located in very different regions of the country, mainly but not exclusively in major metropolitan areas:

• Los Angeles (California Central)
• Beaumont (Texas East)
• Camden (New Jersey)
• Manhattan (New York South)

For a broader perspective, Table 1 lists the 25 courthouses in the country with the largest judge-by-judge differences in their criminal caseloads. These ratios were calculated by comparing the judge who had sentenced the most defendants to the judge who had sentenced the fewest. Across the country, the data show that each judge sentenced an average of 615 defendants during the study period. (For a complete alphabetical listing of the number of criminal defendants sentenced by each district court judge, see this listing.)

As demonstrated by the difference ratios in Table 1, the surprising variations seen in judge-to-judge criminal caseloads were not limited to a handful of districts.

Across all 90 courthouses, the average difference in criminal caseloads among judges was 29 percent, and the median or typical difference was 18 percent. Given the large number of defendants usually handled by each judge, such differences were unlikely to have arisen from chance variations in case assignments.

By way of contrast, however, there were quite a few districts in which the differences in criminal caseloads for the judges serving in the same courthouse were minimal. Table 2 lists the 25 courthouses with the smallest variation among the criminal caseloads of the judges. Most of these caseloads were within five percent. The top three — Brownsville (Texas South), El Paso (Texas West) and Providence (Rhode Island) were within one percent. Again, it is striking that judges in these courthouse shouldered very similar caseloads despite differences in locale or the magnitude of the criminal caseloads at that courthouse.

In an effort to better understand the procedures by which criminal cases are assigned to judges, TRAC contacted court clerks in a sample of federal districts. While not all of the clerks contacted would speak with us, those that did uniformly indicated that while judges were randomly assigned by the court's computerized software system, adjustments were allowed in the "odds of selection" when directed by the chief judge in a district (or sometimes by an individual judge).

For example, if a retired (senior) judge wanted to handle about half the criminal caseload of active judges, the computer selection could be adjusted so that the odds that a case was randomly assigned to that judge would be set at one half^[4]. Cases were still randomly assigned, TRAC was told, but the senior judge's turn would come up half as often. Such a system could therefore be utilized to adjust the criminal caseloads for even active judges without apparently altering the random nature of case assignments.

While the results in Table 1 exclude senior judges, there well may be other special circumstances that explain some of these caseload variations. While acknowledging that these differences do not by themselves indicate that a random assignment system was not being used^[5], the relatively large number of offices with caseload variations, as well as the sheer size of some of these differences, were surprising and appear to raise complex and difficult management challenges for the federal court system.

The Sheer Volume of the Criminal Caseload in Some Federal Courthouses

There was also enormous variation in the volume of criminal cases a judge was required to handle that depended upon the courthouse at which the judge served. While on average active judges sentenced 615 defendants during the study period, this average caseload level varied from a low of 147 criminal defendants in the District of Columbia federal court to a high of 7,020 in the Las Cruces, New Mexico district courthouse.

Federal judges, however, handle both civil and criminal cases. Thus, having a low criminal caseload did not necessarily imply a judge's workload was lower since that courthouse might have a higher civil caseload^[6].

But varying civil caseloads do not appear to explain the differences observed. According to the latest published statistics of the Administrative Office of the United States Courts (AOUSC), the weighted combined civil and criminal caseload per judgeship for the District of Columbia court was less than half the level of judges serving on the New Mexico district courts^[7]. (It was true that the New Mexico court ranked third in the nation on criminal felony filings, while the D.C. court ranked 92nd. However, the New Mexico court ranked 84th on civil filings per judgeship, while the D.C. court ranked even lower — 86th.) Thus, at best only a portion of the differences observed in average criminal caseloads can be explained by differing civil caseloads at a given courthouse^[8].

Table 3 lists the 25 courthouses that had the highest average number of criminal defendants per active judge. Topping the list as mentioned earlier was Las Cruces, New Mexico. Judge Robert C. Brack — the sole district court judge in Las Cruces — sentenced a total of 7,020 defendants during the 70 month study period. The McAllen courthouse in the Southern District of Texas ranked second with an average of 5,467 defendants sentenced for each of the three active judges serving there. The Midland courthouse in the Western District of Texas was ranked third with 4,810 defendants sentenced by Judge Robert A. Junell (the sole U.S. district court judge serving at Midland).

Southwest border courthouses have such high criminal caseloads because of the government's sharply increased emphasis on the criminal enforcement of immigration matters. While apprehensions along the southwest border haven't grown — indeed they generally have been falling for a decade or more — the onslaught of criminal immigration prosecutions that have engulfed federal district courts has occurred as a result of a policy shift to increase criminal prosecutions for illegal entry rather than use civil and administration procedures to remove and sanction individuals found to have illegally entered the U.S.

In fact, the top eleven ranked courthouses listed in Table 3 were all courthouses along the southwest border. After Las Cruces, McAllen, and Midland previously mentioned, in order of their ranking came the federal courthouses in: El Paso, Del Rio, Brownsville, Tucson, San Diego, Austin, Waco and San Antonio. See Table 3.

An overwhelming proportion of criminal defendants plead guilty. In this study period, for example, less than four percent of convictions for the nation as a whole were decided by a judge or jury after a court trial. While the large number of guilty pleas no doubt reduces the time requirements for handling these criminal caseloads, the sheer volume of defendants dealt with by individual judges nonetheless is astounding.

However, southwest border courthouses weren't the only ones with unusually heavy criminal caseloads per active judge. Cedar Rapids in the Northern District of Iowa was ranked 12th out of the 90 courthouses compared. Judge Linda R. Reade — the sole district judge at that location — actually handled a larger caseload (1,383) than the average caseload for judges in Phoenix, Arizona (1,366).

A number of other courthouses not in districts along the southwest border made the list of courthouses with the heaviest average criminal caseloads per judge. These included Buffalo (New York West), Salt Lake City (Utah), New Bern (North Carolina East), Sioux City (Iowa North), and Greeneville (Tennessee East). Again, see Table 3.

Table 4 lists the 25 courthouses with the smallest average criminal caseloads. District Court judges in the District of Columbia registered the lowest criminal caseload with only a average of only 147 defendants sentenced per judge during the 70 month study period — a mere 2 percent of the criminal caseload handled by the judge in Las Cruces, New Mexico.

Going down the list of the 25 courthouses with the smallest criminal caseloads per judge were a number of courthouses in large metropolitan cities: Boston (8th), Philadelphia (9th), Chicago (16th), Detroit (17th), Los Angeles (21st), and San Francisco (23rd). But also on this list were many courthouses not located in major metropolitan areas. These included Rome, Georgia (5th), Worcester, Massachusetts (6th), Jackson, Mississippi (10th), and Toledo, Ohio (11th) just to mention a few.

Conclusion

Equal justice under the law is a phrase engraved on the front of the Supreme Court Building in Washington, D.C. But can this ambitious promise be realistically met by a national court system with extraordinary differences in the workloads imposed on the individual judges handling cases in different parts of the country?

The responsibility for managing criminal caseloads rests on many shoulders. Congress, of course, is obligated to provide the funding that allows the courts to operate in a fair and effective manner. On the other hand, executive branch policies adopted by such agencies as the FBI, DEA and DHS drive the volume and nature of criminal prosecutions and determine where they are filed. But within these varied constraints, it ultimately is up to the judicial branch to manage the day-to-day operations of the courts.

This clearly is a difficult challenge. With the availability of this new and very detailed courthouse-by-courthouse, judge-by-judge data, we hope that federal officials in all parts of the government can better recognize and openly discuss possible remedies.

Appendix

Footnotes

^[1] For more information about this TRAC database and how we developed it, see The Development of Judge-Specific Federal Sentencing Data at TRAC.

^[2] Caseload figures for each judge in this listing are updated monthly with the latest available data compiled by TRAC. The version of the listing used in this report incorporated sentencing records available through July 2012.

^[3] The study covered all district court judges who had sentenced at least 50 defendants during the period October 2006 through July 2012. The category of active judges covered those who had been appointed before October 2006 and had not retired at any time prior to July 2012. Judges who split their time between districts were excluded from district comparisons. In addition, because the study period was longer than the one used for TRAC's March report, slightly fewer active judges qualified for inclusion because more judges had retired.

^[4] The Constitution forbids Congress to diminish a federal judge's salary. Thus they continue to receive their full salary after they retire unless they resign from office. Despite retirement, however, many senior judges volunteer their time to shoulder some of the court's caseload without extra compensation for this work. Indeed, in our study 301 of the 909 district court judges had retired before or during the study period. They handled 27 percent of the criminal caseload during this same period. A total of 159 out of the 909 judges had retired before the study period began yet they handled fully 10 percent of the criminal caseload in the nation. See Appendix Tables A and B.

^[5] We were unable to identify any past empirical studies that actually tested districts to determine whether random assignment took place. The use of random assignment is not a constitutional or statutory requirement. The latest posted web version of the Deskbook for Chief Judges of U.S. District Courts (3rd edition, 2003), published by the Federal Judicial Center, outlines existing guidelines on random assignment as follows (p. 106-7): "Section 137 of Title 28 directs district courts to adopt rules or orders that specify how cases will be assigned to the individual district judges. [...] Most district courts use a random case-assignment system. There are various devices for randomly assigning cases, ranging from sealed envelopes to marbles in a bin. An automated system, available from the Administrative Office's Office of Information Technology, permits courts to use a variety of approaches to random assignment. For example, a court may decide simply to assign each new case randomly to the judges, or a court may decide to assign cases randomly within different divisions of the district or within categories of cases, such as civil and criminal or routine and complex."

^[6] Adding to this complexity when looking at individual courthouses, judges serving in one courthouse not infrequently are assigned civil cases filed at other courthouses within their district.

^[7] See Federal Court Management Statistics, September 2011. Unlike TRAC's comparisons at the federal courthouse level, the AOUSC rankings are not published by individual courthouse within each federal district. Several other caveats should be noted. For example, it is challenging to capture all relevant differences among cases in adjusting for the differences in time requirements from one case to the next, so even weighted averages are an imperfect index. AOUSC statistics are also based on a per judgeship basis, and do not reflect the actual judge time available in each district which could differ depending upon unfilled positions as well as the availability of senior, visiting and magistrate judges who shoulder part of the load.

^[8] The challenge of widely differing caseloads appears to be one of long standing. For example, a 1998 GAO analysis also revealed widely differing judge caseloads among districts, even after weighting cases by their complexity and adjusting for vacancies in judgeships. It is interesting that the D.C. court even then was ranked as having much lower caseloads while the New Mexico court was ranked towards the upper end on the caseload measures used. One contribution of TRAC's current study is that by examining actual caseloads of active judges, our measures have adjusted for the portion of the caseload handled by senior, visiting and magistrate judges.

Correction

An early pre-release version of this report included Wichita, Kansas in Table 1. Special circumstances explained differences in judge caseloads in that courthouse.