Since 2006, the Bureau of Alcohol, Tobacco, Firearms, and Explosives (the "ATF") has engaged in sting operations wherein undercover agents present individuals in this District with an opportunity to rob a fictitious drug stash house. See generally United States v. Mayfield ,
No one feels stronger than this Court about the problems this District has continuously had with firearm violence. During this Court's three decades of public service, it has consistently pointed out and stressed the deadly toll firearms have taken in Chicago and throughout the country. In fact, during this Court's 11 years of service on the United States Sentencing Commission, it strongly advocated for and accomplished the strengthening of penalties for firearm trafficking.
This Court understands that dedicated federal agents, who often place themselves in life-risking situations, sincerely believed that they would recover deadly firearms from the violent streets of our District in these stash house sting operations. However, to paraphrase Supreme Court Justice Louis Brandeis, to declare that in the administration of the criminal law "the end justifies the means" is to declare that the government may violate fundamental principles of common fairness to secure the conviction of an alleged criminal. See Olmstead v. United States ,
Our nation's current tragic pattern of weapons violence does not justify the problematic consequences presented by the government's use of false stash house cases as an investigative technique. The answer to our nation's current tragic pattern of weapons violence lies in stricter firearms regulations, especially with respect to automatic, multi-round weapons, and traditional law enforcement investigative techniques. The Court is mindful that this city just passed the 89th anniversary of the infamous St. Valentine's Day massacre, which killed seven Chicagoans with automatic weapon fire. Yet, even during the low points of the great violence caused by the alcohol wars of Prohibition, the ATF did not seek to use "false alcohol warehouse" tactics against any ethnic organized crime groups to promote public safety. Instead, the ATF used solid investigative work to garner the great public respect of the Elliot Ness era that still lives today as the gold standard of law enforcement. This type of work inspires great public cooperation with law enforcement, unlike the false drug stash house cases before the Court.
The problems with the false stash house cases start at the beginning and do not get any better at the end. The typical false stash house case commences with the recruitment of a confidential informant (or "CI"). Most of the time these are individuals with criminal justice problems. In today's world, they tend to be black or Hispanic-two groups that unfortunately dominate both our federal and local American criminal justice systems. These principally minority confidential informants are told by federal agents to tell their friends and associates that there are places with a large quantity of drugs that can be robbed. The hope is that the informants will organize robbery crews that will result in significant arrests of violent criminals and seizure of weapons. Sometimes these cases work out in that fashion.
The inherent problems of this District's false stash house cases must be seen through the lens of our country's sad history of racism. Every time our country's law enforcement system can be perceived as contributing to that sad history, our justice system suffers and needlessly alienates minority communities who no longer wish to come forward as witnesses or victims. This Court spent more than ten years working as a member of the United States Sentencing Commission before it could reduce the needless disparity between the penalties for crack and powder cocaine, which were correctly revised because they were perceived as discriminatory toward black defendants. This country cannot afford such self-inflicted wounds in light of its sad history of racism. Viewed in this context, the false stash house cases are no different than the extraordinary powder/crack sentencing disparities that previously existed.
In these two cases, Defendants move to dismiss the indictments filed against them on grounds of racial profiling, arguing that ATF wrongfully targeted black and Hispanic individuals for participation in the stings.
Like our colleague in the U.S. Court of Appeals for the Sixth Circuit, this Court finds "the concept of these stash house sting operations at odds with the pride we take in presenting American criminal justice as a system that treats defendants fairly and equally under the law." United States v. Flowers ,
Fortunately for the government, the question before this Court is not whether the practices used in these sting operations are honorable or fair. Instead, the sole question presented to the Court in these cases is whether law enforcement agents violated the Due Process Clause of the Fifth Amendment in pursuing these investigations. At first blush, the evidence submitted by Defendants-including the sheer number of non-white individuals prosecuted as a result of the sting operations-presents troublesome indicia of discrimination. But this Court is not permitted to simply accept matters at first blush. The Court must dig deeper to test the reliability and weight of the evidence that has been submitted. In doing so, the Court concludes that Defendants have failed to meet the heavy burden that currently must be satisfied to obtain dismissal of a federal criminal indictment. Therefore, the motions to dismiss must be reluctantly denied.
BACKGROUND
"According to the standard ATF script for the [stash house] sting, a fictional drug courier was disgruntled with his organization and wanted to rob one of its stash houses. [The target of the investigation] was to meet with the courier to discuss logistics, then recruit a crew, gather weapons, and help the courier carry out the robbery." Mayfield ,
These sting operations have withstood a number of challenges in the courts, including
But at the same time, the stings have "drawn criticism in news reporting, scholarly writing, and from the judiciary." Flowers ,
I. The Brown investigation
The events underlying the Brown case began in April 2012, when Chicago ATF agents began a firearms investigation in Chicago's Englewood neighborhood. (Brown , R. 555-1, Report of Investig. at 144.)
In July 2012, the confidential informant reported to ATF agents that, during an unrecorded conversation, Jones raised the possibility of committing an armed robbery and asked the informant whether he knew about any good targets. (Id. , R. 1, Compl. ¶ 5.) During a controlled buy with Jones in Englewood the following day in which he sold the informant another firearm, the informant, working at the instruction of ATF agents, told Jones about a fictitious stash house that could be robbed. (Id. ¶ 6;
Later that same month, the informant introduced Jones to the undercover agent, Dave Gomez, who was posing as a disgruntled courier for a drug cartel. (Id. , R. 1, Compl. ¶ 8.) During a recorded conversation, Agent Gomez told Jones that he was looking for a "professional crew" to rob twenty kilograms or more of cocaine from a stash house that was guarded by three men with guns. (Id. , R. 1, Compl. ¶ 8;
The next day, Jones brought Abraham Brown and Kenneth Taylor to meet with Agent Gomez. (Id. , R. 1, Compl. ¶ 10; Gov't's Ex. 4E,
On August 1, 2012, Brown, Taylor, and Jones met Agent Gomez again, this time with another crew member Brown had recruited, later identified as Alfred Washington. (Id. , R. 1, Compl. ¶ 15.) Law enforcement records show that Washington was also affiliated with the Gangster Disciples street gang, and that he had a prior drug conviction. (Id. , R. 555-1, Records at 230;
On August 14, 2012, Brown, Taylor, Washington, and Jones met Agent Gomez at a prearranged location so that they could commit the robbery. (Id. , R. 1, Compl. ¶¶ 24, 26.) Brown brought a fifth crew member, Davis, who was Brown's cousin. (Id. ¶ 26;
As a result of these events, Brown, Taylor, Washington, Jones, and Davis were charged with conspiracy, attempt to commit Hobbs Act robbery in violation of
II. The Williams Investigation
The events underlying the Williams case began in the fall of 2012, when ATF agents from around the country came to Chicago as part of a law enforcement "surge" targeting Chicago neighborhoods that were experiencing a rash of violent crime. (Williams , R. 354-1, Gov't's Disc. Resps. at 43-44, 52; Brown , R. 511-5, Takedown Mem. at 111, 130.) The Williams investigation focused on an area near East 63rd Street and South Martin Luther King Drive on Chicago's south side.
On November 7, 2012, ATF Agent Carl Valles, posing as a disgruntled drug courier, met with Williams at 63rd and King Drive. (Id. , R. 354-1, Tr. at 464-71.) At the beginning of the conversation, Agent Valles asked Williams, "you ready to do this lick?" Williams replied, "Hell yea[h]." (Id. at 465.) Agent Valles asked, "You gotta crew," and Williams responded, "Yea, they right there, right now," as they all lived in the immediate area. (Id. at 466.) After Agent Valles explained that the robbery would target a cartel stash house containing eight to ten kilograms of cocaine and guarded by two men with guns, Williams responded that he had "guns like a motherfucker," including "AK's" and "AR-15's" and that if necessary, he would "come to the door shoo[t]in'." (Id. ) Williams said he had a group available to do the robbery, which included his uncle and others he knew from the area. (Id. at 466-68.) Williams described the robbery opportunity as "beautiful" and told Agent Valles that "it [was] an honor" to be involved. (Id. at 467.) At the end of the conversation, the two agreed to meet soon with other members of the robbery crew.
On November 12, 2012, Agent Valles met with Williams and his uncle, who agents later identified as John Hummons.
On November 14, 2012, Lee drove Williams and Hummons to meet Agent Valles and a second undercover ATF agent so that the men could commit the robbery. (R. 1, Compl. ¶¶ 52-54;
Following their arrest, all three Defendants made statements implicating themselves in the planned robbery.
III. Racial Profiling Defense
Defendants first raised the issue of racial profiling in a discovery motion filed more than four years ago. (See Williams , R. 50.) In July 2013, this Court granted Defendants discovery relevant to this potential defense over the government's objection. (Id. , R. 69, Order.) Over the course of the next two years, this Court ordered significant additional discovery from the government, including information pertaining
In 2014, the Court also granted Defendants' motion for the appointment of an expert, Professor Jeffrey Fagan, to conduct a statistical analysis of the stash house cases in this District to determine whether there were statistically significant racial anomalies. (See Brown , R. 269, Order.) To assist Professor Fagan in obtaining the data necessary to conduct his analysis, the Court presided over protracted third-party subpoena proceedings involving the Illinois State Police ("ISP"). (See, e.g., Williams , R. 186, Order;
In May 2016, Professor Fagan completed his report, and he opines that the racial makeup of the stash house defendants in all cases prosecuted in this District shows a strong statistical anomaly when compared to a control group he constructed from the ISP data. (See
In January 2017, the government's expert, Professor Max Schanzenbach, completed his rebuttal report. (Id. , R. 354-1, Schanzenbach's Report.) Professor Schanzenbach opines that there are numerous
LEGAL STANDARD
Under Federal Rule of Criminal Procedure 12(b), "[a] party may raise by pretrial motion any defense, objection, or request that the court can determine without a trial on the merits." FED. R. CRIM. P. 12(b)(1). This includes motions alleging "a defect in instituting the prosecution" due to "selective or vindictive prosecution." FED. R. CRIM. P. 12(b)(3)(A)(iv). "Challenging an indictment is not a means of testing the strength or weakness of the government's case, or the sufficiency of the government's evidence." United States v. Moore ,
ANALYSIS
"The Attorney General and United States Attorneys retain broad discretion to enforce the Nation's criminal laws." United States v. Armstrong ,
Nevertheless, this discretion is "subject to applicable constitutional and statutory limitations and to judicially fashioned rules to enforce those limitations." Russell ,
A claim of selective enforcement is generally governed by the same standards. United States v. Barlow ,
In an appeal of a discovery order in another stash house case, the Seventh Circuit explained another important distinction between selective prosecution and selective enforcement claims: Selective enforcement
I. Applicable Standard of Proof
The first issue the Court must resolve is what standard of proof Defendants must satisfy to obtain dismissal of the indictment. The government argues that the "clear and convincing evidence" standard adopted by the U.S. Supreme Court in Armstrong is applicable, whereas Defendants believe that an ordinary preponderance of the evidence standard applies. (See Williams , R. 338, Defs.' Mot. at 18-19; Brown , R. 555, Gov't's Resp. at 15-16;
It is true that this case does not directly implicate prosecutorial discretion, as Defendants' argument is that the ATF, rather than the U.S. Attorneys' Office, discriminated against them. Armstrong is thus not directly on point. But at the same time, the Court does not find it appropriate to adopt a mere "preponderance of the evidence" standard in deciding Defendants' motion. Applying such a low standard would be inconsistent with the substantial discretion afforded to the executive branch in enforcing the laws of this nation. See Russell ,
One of society's most basic tasks is that of protecting the lives of its citizens and one of the most basic ways in which it achieves the task is through criminal laws .... Implementation of these laws necessarily requires discretionary judgments. Because discretion is essential to the criminal justice process, we would demand exceptionally clear proof beforewe would infer that the discretion has been abused.
The same reasoning applies here. Defendants are challenging decisions "at the heart" of the criminal justice system: the decision of law enforcement agents to pursue an investigation pertaining to the violation of federal law. As Defendants themselves acknowledge, such decisions inherently involve discretionary judgments that take into account a variety of factors. (See Williams , R. 378, Defs.' Reply at 64.) Most of these factors, including enforcement priorities and how best to achieve them, are beyond the scope of judicial review. See Armstrong ,
Substantive claims of selective prosecution and selective enforcement are generally evaluated under the same two-part test, which is derived from a line of seminal Supreme Court cases about the collision between equal protection principles and the criminal justice system. A defendant challenging a criminal prosecution at either the law enforcement or prosecution inflection points must provide "clear evidence" of discriminatory effect and discriminatory intent[.]
This Court reads Davis the same way: it recognized a lower standard for granting discovery on a selective enforcement claim, but it did not address the standard for resolving a motion to dismiss on grounds of selective enforcement on the merits. In the absence of clear direction from the Supreme Court or the Seventh Circuit that a lower evidentiary burden applies to a substantive claim of selective enforcement in these unique and troublesome stash house cases, this Court must reluctantly follow the Supreme Court's admonition in McCleskey and will require clear evidence before concluding that federal law enforcement agents abused their discretion in pursuing these investigations.
II. Daubert
Before turning to the merits, the Court must next address the threshold issues of the parties' expert reports, which are a key component of their respective positions in this case. The admissibility of expert testimony is governed by Federal Rules of Evidence 702 and 703 and the U.S. Supreme Court's decision in Daubert v. Merrell Dow Pharmaceuticals., Inc. ,
In the end, the inquiry is a flexible one that provides wide latitude for the Court to serve its role as "gatekeeper." C.W. v. Textron, Inc. ,
When the Court rather than a jury is the factfinder on a particular issue, "the judge need not conduct a Daubert (or Rule 702 ) analysis before presentation of the evidence," but still "must determine admissibility at some point." Kansas City S. Ry. Co. v. Sny Island Levee Drainage Dist. ,
A. Professor Fagan's Report
1. Professor Fagan's qualifications as an expert
In assessing an expert's qualifications, the Court should consider the expert's full range of education, experience, and training. FED. R. EVID. 702 ; LG Elec. USA v. Whirlpool ,
2. Reliability of Professor Fagan's methodology
Rule 702 requires that the expert explain the methodologies and principles that support his or her opinions; in other words, the expert cannot simply assert a "bottom line." Minix v. Canarecci ,
To reach his opinions in this case, Professor Fagan obtained criminal history records for each of the 94 stash house defendants and coded them for race, criminal history, and other factors. (Williams , R. 338-2, Fagan's Report at 5.) He then conducted a detailed statistical analysis of the stash house defendants with respect to race, criminal history, and other factors. (Id. at 15-20.) He also obtained extensive crime data from the ISP meeting certain parameters. (Id. at 6.) Once individuals were identified from these criteria, he constructed a complete criminal history for each individual using a search of ISP databases, which included variables such as race, gender, age, number of prison sentences, and other criminal history factors. (Id. at 7, 9.) He conducted a comparison of the stash house defendants and the pool of other individuals, who he calls "potential eligibles," with respect to race, criminal history, and other factors. (Id. at 21-23.) This included conducting a series of empirical tests for assessing Defendants' selective enforcement claims in this case: a disparate treatment test, a second test incorporating a principle known as augmented inverse probability weighting, and a third test using propensity score matching.
In the government's view, "The Fagan Report is flawed from top to bottom. It starts with the wrong comparison group, then performs elaborate statistical analyses based on core assumptions that are simply mistaken." (Brown , R. 555, Gov't's Resp. at 37.) The government argues that Professor Fagan's report "fail[ed] to ... make the most elementary comparisons," and "has no value as causal explanation and is therefore inadmissible in a federal court." (Id. at 27 (citations omitted).) But the Court finds that these criticisms go more to the weight of Professor Fagan's opinions than to their admissibility. For purposes of Daubert , Professor Fagan's opinions are properly grounded in established statistical principles and his extensive knowledge of the field of criminology. Whether there are any weaknesses in his opinions and how much weight they should be given are matters for this Court to decide separately. See Kawasaki Kisen Kaisha, Ltd. ,
3. Whether Professor Fagan's report will assist the trier of fact
Next, the Court must decide whether Professor Fagan's testimony will assist the trier of fact in determining a relevant fact at issue in the case. The Court finds that this inquiry is satisfied. Professor Fagan provides a statistical analysis of the stash house Defendants and a comparison to what he deems to be a "similarly situated" group of individuals. Making this comparison is a key component of Defendants' racial profiling claim, and this Court does not have the ability to gather its own extensive crime data or conduct detailed statistical analyses based on that data. Accordingly, the Court finds that Professor Fagan's report and testimony will assist the Court in resolving Defendants' racial profiling defense.
B. Professor Schanzenbach's Report
1. Professor Schanzenbach's qualifications as an expert
Professor Schanzenbach is a professor of law at Northwestern University School of Law. (Brown , R. 555-1, Schanzenbach's Report at 33.) He holds a law degree from Yale Law School and a Ph.D. in economics, also from Yale. (Id. ) Among his many credentials, he previously served as the director of the Searle Center on Law, Regulation, and Economic Growth, and served as a board member on the American Law and Economics Association and other scholarly organizations. (Id. at 37-38.) He has published books and articles in peer-reviewed journals, law reviews, and other scholarly publications involving empirical studies of sentencing disparities and other topics. (Id. at 33-35.) The Court concludes that Professor Schanzenbach is qualified to offer opinions about the statistical analysis conducted by Professor Fagan.
2. Reliability of Professor Schanzenbach's methodology
To reach his opinions in this case, Professor Schanzenbach reviewed a variety of materials, including detailed information from all 24 sting operations, Professor Fagan's report, Defendants' motion to dismiss, and the thousands of pages of discovery materials turned over by the government in this case. (Brown , R. 555-1, Schanzenbach's Report at 2.) From there he conducted a detailed analysis of Professor Fagan's report, including testing various assumptions on which Professor Fagan's
3. Whether Professor Schanzenbach's report will assist the trier of fact
The Court also finds that Professor Schanzenbach's testimony will assist the trier of fact. Defendants' motion to dismiss relies heavily on the opinions offered by Professor Fagan. Testing the validity and strength of Professor Fagan's conclusions is a necessary component of resolving the motion to dismiss. Through his extensive knowledge of statistical principles, Professor Schanzenbach can provide the necessary adversarial testing of Professor Fagan's report. Therefore, the Court finds that the third prong for admissibility is satisfied. The Court will consider the weight and persuasiveness of both experts' reports in connection with its analysis of the merits below.
III. Merits
A. Discriminatory Effect
The first element of a selective enforcement claim is discriminatory effect. Armstrong ,
The statistics compiled by Professor Fagan show that between 2006 and 2013, the defendants prosecuted in the ATF false stash house cases in this District were 78.7 percent black, 9.6 percent Hispanic, and 11.7 percent white. (Williams , R. 338-2, Fagan's Report at 15.) To create a comparison group, Professor Fagan gathered the names of individuals in an eight-county area (Cook, Lake, Will, DuPage, Kane, Kendall, LaSalle, and
Professor Fagan conducted a series of tests "to determine whether race predicts selection into the Stash House defendant pool." (Id. at 23.) This included conducting several regression analyses to "see if there are particular types of information, such as demographic factors or criminal history, that explain whether and why the selection of Stash House defendants is based on race or ethnicity." (Id. at 25.) For each model, "ail defendants were pooled for the analyses," (Id. at 24.) He explains that "[i]n each instance, the outcome of interest is selection as a defendant." (Id. ) Professor Fagan also used Augmented Inverse Probability Weighting "to estimate a first predictor of race (the treatment) adjusted for the covariates, and then the effects of the adjusted treatment variable on the outcome (selection into the Stash House Program)." (Id. at 33.) He also used Propensity Score Matching, which "takes into account all background characteristics (i.e. covariates) other than race that might be correlated with race." (Id. at 35.) In this test, the 94 stash house defendants were
It must be recognized at the outset that constructing a comparison group in a selective enforcement case is an extremely difficult task. See Hare ,
1. Assumptions about Defendants
Critical to Professor Fagan's entire analysis in an assumption that the ATF selected each of the 94 stash house defendants for participation in the stings. Professor Schanzenbach convincingly explains that "[t]reating all 94 individuals as a 'choice' or 'random draw' made by the ATF wrongly increases the sample size and statistically attributes choices to the ATF that it did not make." (Id. at 16.) Defendants believe that Professor Fagan's assumption was appropriate because the ATF "owned" the entire stash house operation, such that, in their view, the agency is legally responsible for the race of all participants and not just the initial targets.
As a factual matter, the bulk of the 94 defendants were not recruited by ATF. Instead, the record reflects that the stash house investigations began with the ATF's selection of one or more initial targets, based on information provided to them by confidential informants. The remaining defendants were recruited by the targets themselves, or by other members of the conspiracy who were, in turn, recruited by the targets. In fact, the record reflects that the ATF never even met 32 of the 94 defendants-nearly one-third of the total defendant group-until the day of their arrest. (Evid. Hr'g Tr. at 100-01; Brown , R. 630, Defs.' Reply at 27.) Additionally, two of the stash house cases, involving a total of 14 defendants, were not initiated by the ATF at all. Instead, those investigations were initiated by the Federal Bureau of Investigation ("FBI") based on information provided by its own confidential informants.
Defendants make arguments about "apparent agency" and the ATF "owning" the stash house stings, but these are legal arguments, and legal arguments do not affect probabilities. Professor Fagan's report is about probabilities. (See, e.g., Brown , R. 603-3, Fagan's Reply Report at 21-22 (opining on the "probability of randomly selecting defendants with this racial composition"); Evid. Hr'g Tr. at 54 ("We found it extremely unlikely to select a pool of defendants who were non-white, meaning black and Hispanic, compared to a pool of whites.").) As he explained at the hearing, his report describes-in very simple terms-the odds of a person randomly choosing certain color jelly beans by reaching into a jar containing jelly beans of various colors.
With respect to the FBI cases, for example, Professor Fagan explained that he felt it was appropriate to consider the 14 defendants indicted in those cases, all of whom were black, as a "selection" made by ATF because "once the ATF agents took over the case, they could easily have rejected those persons if they didn't meet the ATF criteria." (Id. at 50.) This ignores the realities of these cases, which is that ATF did not meet with up to a third of the defendant group until the day of the arrest. (Evid. Hr'g Tr. at 100-01; Brown , R. 630, Defs.' Reply at 27.) But putting that aside, Professor Fagan is essentially stating that ATF could have put certain jelly beans back in the jar had it chosen to do so; even assuming this was the case, it would not reveal anything about the probability of selecting particular jelly beans from the jar in the first place.
On a different record, it might be appropriate to attribute the race of other members of the conspiracy to the ATF. Suppose, for example, that Williams brought two potential recruits to a meeting with
At the hearing, Defendants suggested that it was "foreseeable" to ATF that the conspiracies would form along racial lines. (Evid. Hr'g Tr. at 108, 112.) It is not clear that the concept of foreseeability, which arises from negligence jurisprudence, would apply to a claim alleging purposeful race discrimination under the Fifth Amendment. See Pers. Adm'r of Mass. v. Feeney ,
On the question of attributing the race of other members of the conspiracy to the ATF, a recent opinion from the Seventh Circuit is instructive. In Conley , which also involved a fictitious stash house sting conducted by the ATF, the defendant raised a defense of entrapment, arguing that he was not predisposed to commit a stash house robbery and was unfairly lured into committing the offense by the conduct of ATF agents.
Similarly, the Fourth Circuit recently rejected a claim of outrageous government conduct in a fictitious stash house case because the defendants raising the challenge "were not targeted by ATF but recruited by [the initial target], whom ATF targeted based on information that he was an active, armed drug trafficker." Hare ,
The Court finds this reasoning persuasive. In Brown and Williams , the ATF targeted Jones and Williams, both of whom came to the ATF's attention based on information provided by its confidential informants. Jones was known to affiliate with the Gangster Disciples, and the ATF had good reason to believe he was actively involved in dangerous criminal activity, as he had recently sold more than 20 firearms to a confidential informant working with the ATF. (Brown , R. 555-1, Report of Investig. at 144;
For these reasons, the Court finds an insufficient basis in the record to attribute the race of defendants other than the initial target to the ATF. Professor Fagan's entire report relies heavily on the race of defendants who were not actually selected by ATF agents, and it does not provide any analysis specific to just the initial targets. As Professor Schanzenbach explained, considering only the initial targets would "significantly reduce the sample size" and would also change the racial composition, making the defendant group somewhat whiter,
Another of Professor Fagan's critical assumptions is that the stash house defendants were drawn independently of each other, in other words, that "the selection of one defendant did not influence the selection of any other defendant[.]" (Brown , R. 555-1, Schanzenbach's Report at 19.) Professor Schanzenbach convincingly explains that "[i]f independence is assumed when in fact the data are grouped and correlated, a naive regression assuming independence can lead to an erroneous conclusion that the results are statistically significant." (Id. at 19-20.) Professor Fagan believes that his assumption of independence was appropriate because "the [stash house] cases are often made up of individuals with weak ties to each other." (Id. , R. 603-3, Fagan's Reply Report at 5; see also Evid. Hr'g Tr. at 119 ("[T]here really wasn't evidence that these are tightly clustered groups of like-minded or behaviorally similar individuals.").) But the record does not bear this out.
Taking Brown as an example, the ATF presented the stash house robbery idea to Jones, who in turn recruited Brown and Taylor; they in turn recruited Washington and Davis. (Id. , R. 1, Compl. ¶¶ 8-10.) Law enforcement records show that Brown, Taylor, and Washington were all affiliated with the Gangster Disciples, and Brown and Washington admitted on the undercover recordings to having committed past home invasion robberies with Davis, who was Brown's cousin. (Id. ¶¶ 8-26;
Professor Fagan offers various conclusions about the "probability of randomly selecting defendants with this racial composition," (id. , R. 603-3, Fagan's Reply Report at 20-21 (emphasis added) ), but the record does not show that each of the
2. Construction of the Comparison Group
The Court also has concerns about the comparison group constructed by Professor Fagan. The sheer number of people in the group, as he defined it, demonstrates the breadth of his analysis. (Brown , R. 510-2, Fagan's Report at 4-10.) He opines that ATF could have selected anyone among a group of 292,000 individuals living in an eight-county area to participate in the stings. (Id. at 8, 17.) But it seems implausible to assume that nearly 300,000 people-which is approximately 10 percent of the population of men aged 14 to 49 living in the relevant counties during the relevant period-would have been willing to commit a stash house robbery had they been given the opportunity. (See R. 555-1, Schanzenbach's Report at 3.) By its very nature, a stash house robbery is a highly dangerous crime that requires access to firearms, the potential use of deadly force, and a significant risk to one's own safety.
Professor Fagan also took a very broad view of the relevant years for constructing the comparison group. Although no stash house stings were conducted until 2006, he included in the comparison group anyone with a conviction for the relevant offenses in 2000 or later. He testified at the evidentiary hearing that he "backed it up to 2000 so that we would have some opportunity to observe and measure [the potential eligibles'] criminal activity prior to the period of the stash house recruitments." (Id. at 21.) This may have been a valid decision in terms of criminology, but using data from as far back as 2000 led to a potential overinflation of the comparison pool. By way of example, someone with one criminal conviction from 2000 was counted as available for selection each year of the stash house sting operations, even though, by Professor Fagan's own admission, at the time the stash house investigations were underway they may have been deceased, in prison, or no longer involved in criminal activity. (See id. at 174, 250-52.)
Controlling for the recency of criminal activity by those in the comparison group would have offered some predictor of whether they were willing and able to commit a stash house robbery in the years the stings were being conducted. It also would have correlated more closely with ATF criteria, which required that targets be "currently involved in criminal activity." (Id. at 235; Brown , R. 511-2, ATF Order at 5.) This information was contained in the data Professor Fagan was provided, but he chose not to use it in his analysis. (Evid. Hr'g Tr. at 224, 290.) Without an adequate basis for concluding that the 290,000 individuals in the comparison group were actually available for selection by the ATF and willing to commit a stash house sting between 2006 and 2013, it is difficult for this Court to rely on Dr. Fagan's findings to meet the required evidentiary threshold. See Hare ,
Professor Fagan also took a very broad view of the proper geographic scope for constructing the comparison group. For instance, he included all of LaSalle County in his analysis because one defendant in one case brought in 2007 lived in LaSalle County. (Evid. Hr'g Tr. at 206-07.) Despite this rather tenuous link to the stash house sting operations, he assumed that for each sting, including those involving a predominately black gang in Chicago's Englewood neighborhood, the ATF was just as likely to have drawn a participant from rural LaSalle County, located nearly 100 miles from Chicago. (See id. at 206-09.) This would appear to conflict with his opinion that "most people don't travel very far from their homes or from wherever their current residence is to commit a crime." (Id. at 221.) The inclusion of LaSalle County alone dramatically increased the number of whites in the comparison pool, as that county has a population that is 95 percent white. (See Defs.' Hr'g Ex. 4.)
Professor Fagan conducted a number of detailed regression analyses to create a closer comparison between members of the eligibles group and the stash house defendants with respect to various factors. (See id. at 54-69.) But again, he did not account for geographic differences when making his comparisons.
As a "counterfactual" to Professor Fagan's findings, Professor Schanzenbach conducted his own basic analyses of the defendant group and various subsets of the eligibles group. (Brown , R. 555-1, Schanzenbach Report at 5-8.) He found that black, white, and Hispanic defendants in the stash house cases had statistically similar criminal histories, and that on average, black defendants had more convictions, arrests, and sentences, and were more likely to have been convicted or arrested for a weapons offense than white defendants. (Id. at 5.) In the eligibles group, he found that those with an arrest for a violent offense (which he and Professor Fagan defined as murder, rape, robbery, assault, and battery) were 71 percent black and 15 percent Hispanic.
When litigants asserting an equal protection claim fail to make a proper comparison to a similarly situated group, courts have declined to find sufficient statistical evidence of discrimination, even where-as here-the raw data suggests startling racial anomalies. See, e.g., Hare ,
That is not to say that the statistics presented in this case are not deeply troubling. They are. The Court should not have to look to p-values, clustering, or other complex statistical principles to determine whether agents of the federal government engaged in race discrimination. Their conduct should be above reproach under any basic standard. At the same time, the Court has fundamental concerns about the reliability of Professor Fagan's findings-concerns that cannot be brushed aside simply because of an underlying discomfort with these type of sting operations. For the reasons outlined above, the Court reluctantly concludes that Defendants have fallen short of proving discriminatory effect. The Court credits the valiant efforts of defense counsel to show discriminatory effect. These efforts, however, substantially fail to meet the exacting standards set by the current applicable law.
B. Discriminatory Intent
Assuming Defendants satisfied the discriminatory effect prong, they "must still prove discriminatory intent in order to establish a violation of the Equal Protection Clause." Chavez ,
Under the applicable law, then, it is not enough for Defendants to show that the ATF was " 'willfully blind' to the racially disparate impact of its stash house sting operations[.]" Hare ,
To prove discriminatory intent, Defendants point to two factors. First, they claim that comments made by the undercover agents involved in the stings-Agent Gomez in Brown and Agent Valles in Williams -prove that these agents "expressly targeted" them because of their race, (Williams , R. 338, Def's Mot. at 62-65.) Second, they argue that ATF agents substantially deviated from ATF guidelines in these investigations, and that these deviations prove that agents targeted Defendants because of their race. (Id. at 40-62.) The Court addresses these points separately below.
1. Comments Made by ATF Agents
a. Agent Gomez's Comments in Brown
Defendants first point to certain comments made by Agent Gomez, who is Hispanic, during the recorded conversations. (Williams , R. 338, Defs.' Mot. at 64-65.) In the context of explaining that he
They believe that the clear import of Agent Gomez's comments, taken together, was that Agent Gomez "targeted Mr. Jones because he was Black," and that he "wanted Jones to recruit Black people." (Id. at 64-65 (emphasis in original).) The Court disagrees. There is no evidence to suggest that Agent Gomez was involved in the underlying investigation of Jones's sale of multiple guns to a confidential informant, or that he was involved in the initial decision to target Jones for investigation. (See Brown , R. 555-1, Report at 144;
Additionally, when viewed in context, it is apparent that Agent Gomez's comments were part of a ruse to make his story plausible rather than evidence of race-based animus. Agent Gomez was essentially trying to offer an explanation for why he would offer the robbery to a group of strangers whom he had no apparent reason to trust. He explained that he wanted to commit the robbery with Defendants, in essence, because they had no readily discernible connection to him.
Agent Gomez worked on other stash house investigations in this District, and he followed the same basic script for his work in those cases. For instance, in United States v. Corson , No. 06 CR 930, which involved two white defendants and one Hispanic defendant, he said, "I need a crew that ain't gonna come back to me. I don't know you guys. You don't know me. Nobody can put us together." (Brown , R. 555-1, Tr. at 445; see also Williams , R. 338-2, Fagan's Report at 13.) Later in the conversation, he said, "I want [the crew members] to see my face. 'Cause I don't wanna take a round ... when I get inside." (Id. at 447.) He made similar comments to a group of defendants in United States v. DeJesus , No. 12 CR 511, who, like him, were Hispanic. (See Williams , R. 338-2, Fagan's Report at 14.) Specifically, he said to them, "Now what I like about you...nobody can put me and you together." (Id. , R. 354-1, Tr. at 638.) He reiterated the same thing several times during the conversations. (Id. at 643 ("[W]hat I like about you, bro, is like nobody can put me and you together."); see also id. at 661 ("[T]hat's what I like about you. Nobody can put me and you together."); id. at 678 ("[N]obody can put me and you together, and that's great.").) During another conversation he said he needed to meet the remaining member of the robbery crew because, in his words, "I want him to see my face, 'cause the other guys ... they're ... Mexican just like me ... [and] I don't want [to] accidentally take a round." (Id. at 746.) The fact that Agent Gomez made basically the same comments to defendants of other races undercuts Defendants' argument that his comments prove he targeted them because they are black. That is not to say that the Court condones the ATF's tactics. But considered in context, the Court finds that Agent Gomez's comments fall short of establishing that the decisionmakers in Brown acted with a racially discriminatory purpose. See McCleskey ,
2. Agent Valles's Comments in Williams
Agent Valles is not normally assigned to the Chicago area and instead was here on temporary assignment as part of the fall 2012 law enforcement "surge." (Williams , R. 354-1, Gov't's Disc. Resps. at 43-44, 52.) Williams was the only stash house case he worked on in this District. (See
Defendants believe that Agent Valles's comments demonstrate "in no uncertain terms" that he presented the stash house robbery opportunity to Williams "because Mr. Williams is black," and further, that "the race of the robbers was important, and they couldn't be Mexican[.]" (Id. , R. 338, Defs.' Mot. at 63 (emphasis in original).) Again, the record does not show that it was Agent Valles who actually selected Williams as a target for the investigation. Instead, ATF records reflect that on November 7, 2012, a confidential informant working with the ATF identified Williams to a different ATF agent as someone who had expressed an interest in committing a robbery. (Id. , R. 1, Compl. ¶ 8;
Agent Valles's repeated mention of race is very regrettable. But the Court considers that it was Individual B, not Agent Valles, who first broached the subject of race. The Court also considers that when Agent Valles made these comments, he was in a rapidly evolving, potentially dangerous situation that required him to think quickly and adapt his story based on statements of the crew members. Agent Valles was obviously aware that all of these conversations were being recorded, and the Court finds it rather implausible that he would simply announce during a recorded conversation that he had a racially discriminatory motive for pursuing the investigation.
Additionally, when considered in context, it is apparent that Agent Valles's comments were part of an attempt (however inartful) to provide a plausible explanation for why he was offering the robbery to a group of strangers. In effect, he was explaining that it was helpful to him that the crew members (at least those whom had met thus far) were not of the same race as him, because there would be no readily discernable basis to connect them. As far as his comment went, there also would have been no basis to connect him to crew members who were white, Asian, or of some other non-Mexican ethnicity had Williams decided to recruit such individuals. Obviously his comment had no basis in fact, because there was no actual Mexican cartel and no actual stash house. Agent Valles was simply trying to assure the potential crew members that everything was in order, and that he was choosing to offer them the robbery opportunity for legitimate reasons. Agent Valles stated several times during the recorded conversations that it was up to Williams to decide how the robbery should be conducted, and
B. Departures From Protocol
Defendants' remaining argument is that ATF agents' significant departures from ATF protocols demonstrate that they were motivated by racial animus in pursuing these investigations. (Williams , R. 338, Defs.' Mot. at 45-61.) The Supreme Court has recognized that a significant failure by decisionmakers to follow established protocols can give rise to an inference of discriminatory purpose. Vill. of Arlington Heights v. Metro. Hous. Dev. Corp. ,
ATF guidelines provide that the stash house stings are intended to "concentrate on identifying persons and infiltrating groups that collectively and/or as a community focus their criminal activities on executing robberies, by means of force, for personal gain." (Brown , R. 511-2, ATF Order at 3-4.) Agents conducting the stings are directed to "target persons who show a propensity of doing harm to the public through violent behavior/armed robberies and whose activities have been documented either through criminal history, criminal reputation, or self-incrimination." (Id. ) ATF guidelines define "violent crime" as "offenses that involve force or threat of force," including murder, forcible rape, robbery, aggravated assault, and arson. (Id. ) ATF guidelines further provide that at least two members of the group must be "identified as violent offenders," at least one target must have a past violent crime arrest or conviction, and all targets must be "currently involved in criminal activity." (Id. at 4-5.) The operations manual used to conduct the stash house stings provides that ATF agents must ensure that targets "are a viable robbery crew or violent criminals." (Id. , R. 511-1, ATF Manual at 33.) Training materials used to prepare ATF agents to conduct the stings provide that targets "must have the ability to commit a home invasion by ... having possession of, or access, to firearms." (Id. , R. 511-3, Zayas Training Materials at 7.) ATF guidelines instruct that "[c]riminal histories alone are not indicative of the ability, commitment, and experience of the subjects being investigated to execute the robbery." (Id. , R. 511-2, ATF Order at 7.)
These criteria, on their face, "do not suggest any discriminatory motive" on the part of the ATF. Hare ,
First, Defendants make much of the fact that there is no evidence showing that they were working as "established" robbery crews prior to the stings.
As to Brown , the record reflects that Jones had no prior felony convictions, but he had arrests for two violent offenses, assault and aggravated battery. (Id. , R. 511-5, Takedown Mem. at 95.) ATF guidelines specifically permit consideration of a
ATF policy specifically permitted agents to consider a target's own statements when deciding whether he or she met the targeting criteria, (id. , R. 511-2, ATF Order at 3), and these men made many damaging statements in the undercover recordings. Jones made inculpatory statements describing his involvement in a recent retaliatory shooting near his home that had left an innocent bystander paralyzed. (Id. , R. 555-1, Tr. at 177-78.) A check of Chicago police department records confirmed that there had been a shooting that matched the details provided by Jones. (Id. , R. 511-5, Takedown Mem. at 94.) The other men described prior home invasion robberies they had committed, and also gave indications that they were active in the illegal drug trade: They knew the current market price for a kilogram of cocaine, and they were able to line up a buyer to purchase between 10 and 15 kilograms of cocaine after the robbery. (Brown , R. 555-1, Tr. at 259-80; Williams , R. 354-1, Tr. at 248.) Their ability to facilitate a sale of that magnitude reasonably suggested that this was not their first time selling illegal drugs.
ATF agents also had a basis to believe that the group had access to firearms, as Jones had recently sold multiple guns to the confidential informant. ATF records reflect that the informant continued to purchase guns from Jones while the stash house sting was underway, including purchasing an automatic rifle and multiple firearms with obliterated serial numbers. (Williams , R. 354-1, ATF Operational Plan at 197.) On the day of the arrest, Brown told Agent Valles that the group had brought three guns with them, and when they were arrested, agents found two loaded 9-millimeter handguns and a sawed-off shotgun in Jones's vehicle. (Brown , R. 1, Compl. ¶ 29.)
Defendants criticize ATF agents for not including Washington and Davis in the "takedown memo" that was prepared prior to the day of their arrest. (Brown , R. 518, Defs. Mot. at 57 n.35; see also
Based on the record, the Court finds that ATF had an adequate basis to conclude that the Brown Defendants constituted a viable robbery crew, that at least two members of the group were "violent offenders," at least one member of the group had a past violent crime arrest or conviction, and the group had access to firearms. (Id. , R. 511-2, ATF Order at 5.) In short, the Court does not find that the targeting of this group, or the manner in which the sting was conducted, so deviated from ATF policies as to suggest a racially discriminatory purpose by the decisionmakers in the case.
As to the Williams investigation, the record shows that the ATF initially targeted Williams, who had five felony convictions, including convictions for aggravated assault with a deadly weapon and unlawful use of a weapon following a felony conviction. (Williams , R. 332-5, Takedown Mem. at 126.) Williams recruited Hummons, who had previous convictions for home invasion, armed robbery, and other offenses. (Id. , R. 12, Report at 5-8.) They in turn recruited Lee to act as the driver, who they said was experienced in committing robberies. (Id. , R. 354-1, Tr. at 488-89.) Indeed, Lee gave a post-arrest statement admitting that he had previously acted as the driver in armed robberies of drug dealers for a mutual friend of Williams's. (Id. , R. 354-1, Lee's Statement at 513.) Lee also had multiple convictions for drug offenses and arrests for attempted aggravated robbery and aggravated assault. (Id. , R. 13, Report at 3-4.) There was also reason to believe that the men were engaged in criminal activity at that time. Williams told Agent Valles that he possessed numerous guns, which itself would be unlawful for someone previously convicted of a felony. See
Defendants complain that ATF agents did not identify Hummons or Lee before they were arrested. (Id. , R. 338, Defs.' Mot. at 17-18.) But at the time the takedown memo was written-November 9, 2012-Agent Valles had not yet met either Hummons or Lee. (See
ATF protocols provide that agents should try to identify all targets of the investigation, but should use discretion "to avoid an investigation being compromised." (Brown , R. 511-2, ATF Order at 7.) Agent Valles could hardly ask crew members for their first and last names when they appeared for a meeting without arising suspicion. Agents were also required to list all known targets in the takedown memorandum, (id. at 8), but they obviously could not list individuals they had not yet met or identified. Nothing about the fact that agents were unable to identify Hummons or Lee by name until the day of their arrest suggests that the agents targeted them because of their race. Again, ATF guidelines specifically provide that undercover agents may use the day of arrest as "an opportunity to speak with all members of the organization in the event that all subjects were not present at prior meetings." (Id. at 7.) Thus, the actions of agents in these cases did not deviate from ATF policy.
Based on the record, the Court finds that the ATF had an adequate basis to conclude that the group targeted in Williams -including individuals who did not show up on the date of the robbery for reasons beyond the ATF's control-constituted a viable robbery crew, that at least two members of the group were "violent offenders," at least one member of the group had a past violent crime arrest or conviction, and the group had access to firearms. The Court does not find that the targeting of this group, or the manner in which the sting was conducted, so deviated from ATF policies as to suggest a racially discriminatory purpose on the part of the decisionmakers in the case. Therefore, the Court finds that Defendants' evidence falls short of proving the discriminatory intent necessary to establish a constitutional violation. For these reasons, the motions to dismiss are denied.
IV. Post-Hearing Discovery Motion
As a final matter, following the evidentiary hearing held in December 2017, Defendants moved to reopen discovery.
The issue of discovery arose at the evidentiary hearing, when defense counsel made several references to a discovery request Defendants previously made in which they sought information about individuals who-as counsel described it-were "rejected" by ATF agents. (See, e.g. , Evid. Hr'g Tr. at 487.) Counsel suggested that in denying this particular discovery request, this Court had deprived Defendants of critical information they needed to prove their racial profiling claim. (Id. ) This is not an entirely accurate account of the record. To clarify, in February 2015, Defendants filed a 28-page motion seeking a vast amount of discovery materials from the government. (Williams , R. 177, Disc. Mot.) Within the motion, Defendants noted that the government had contended "that the proper comparison group consists of people who are in some way associated with stash house robberies and the ATF's recruiters, but who did not participate or were not arrested." (Id. at 13.) Defendants "strongly dispute[d] this definition of the comparison group" but nevertheless sought broad discovery on that issue. (Id. ) Among other things, they sought information about individuals who "demonstrated a willingness to conduct a robbery to a government informant or undercover officer, but were not offered the opportunity to conduct a robbery." (Id. at 13.) The Court granted a substantial portion of the requests contained in Defendants' discovery motion but denied this particular request as too broad.
Between March 2015 and June 2017, the parties proceeded based on the universe of documents that the Court had ordered to be disclosed. Defendants acknowledge that at no point during that extended period did they seek to narrow their request or advise the Court that Professor Fagan needed this information to conduct a proper analysis. (See Brown , R. 670, Tr. at 14-15.) Then, nearly a year after Professor Fagan's report was completed and several months after the government filed its response brief and rebuttal report by Professor Schanzenbach, Defendants made a broad request for information about "any and all individuals whom an ATF Confidential Informant or ATF agent approached about a Stash House Operation in this District during the years 2006 to 2013." (Id. , R. 359, Renewed Disc. Mot. at 7.) The government objected, pointing out that Defendants already had detailed information about the 24 stash house cases brought in this District and information about an additional three stash house investigations that did not result in prosecution. (See Williams , R. 364, Gov't's Resp. at 5.) The government represented that it had no ready means of compiling additional responsive information, given the sheer number of confidential informants and
In their post-hearing discovery motion, Defendants seek vast discovery on four issues: (1) the "[f]ull universe of individuals with whom the ATF or its CIs had contact in this district regarding a Stash House Operation, from 2006-2013 (regardless of whether the ATF ultimately arrested the individuals)"; (2) information about the "entire pool of CIs available to the ATF" nationwide (3) the "[f]ull universe of ATF targeting criteria, including any criteria regarding gangs and geography"; and (4) "[i]nternal ATF documents recording ATF compliance or lack of compliance with its criteria and procedures[.]" (Flowers , R. 706, Defs.' Post-Hearing Disc. Mot. at 1-15.) Upon careful review of the parties' filings and the transcript of the evidentiary hearing, the Court declines to order further discovery. The government also provided the Court with several new documents for an in camera inspection. The Court carefully considered whether these documents should be disclosed to the defense, but concluded that these documents revealed too many confidential, investigative strategies that would not meaningfully change the Court's decision in these cases. The documents make it clear that ATF targeted ethnic gangs that it considered violent. They do not change this Court's findings with respect to discriminatory effect or intent. In any event, these documents will be preserved under seal as part of the record for any necessary appellate review.
As is fully outlined in the government's response brief, Defendants received extensive discovery in this case, including detailed information about the 24 stash house cases charged in this District, detailed criminal histories of all 94 defendants charged in those cases (many of whom have closed cases and are not part of the present litigation), sensitive materials from the ATF detailing the guidelines, policies, and training materials used in connection with the stash house stings, and information about the confidential informants, including their race, and instructions they were given by the agents. (See R. 674, Gov't's Resp. to Defs.' Post-Hearing Disc. Mot. at 2-4.) Defendants' new discovery requests would not fill the gaps in Professor Fagan's report. The most significant problems this Court had with Professor Fagan's report, which are all fully outlined above, were his decision to include the cases initiated by the FBI, his decision to attribute the race of all 94 stash house defendants to the ATF, and his decision to treat the selection of each defendant as independent. There were also significant problems with his construction of the comparison group-as he chose to define it-and problems with the regression analysis he used to compare the defendant group to the comparison group. Providing Defendants with vast new discovery about ATF's nationwide supply of CIs or an entirely different population of individuals who may have had "contact" with an agent or CI about a stash house sting would not remedy any of these problems. In fact, some of these belated requests appear to be based entirely on questions posed by the judges during the evidentiary hearing, and not on statements made by the government or evidence suggesting that the government withheld relevant evidence. (See Flowers , R. 706, Defs.' Post-Hearing Disc. Mot. at 5 (as to request # 1, stating that one judge had asked, "Isn't the real analysis of what the cadre of CIs that the ATF has?", an issue not previously raised by Defendants or addressed in any manner in their motion
The government has also convincingly explained that the information sought by Defendants, at least as to certain requests, would be extremely burdensome and time-consuming to compile. (Id. at 10-12.) As to request # 1, assembling the information would require "asking every agent and task force officer who worked in the relevant ATF offices between 2006 to 2013 (some of whom are now retired), and every CI and 'associated' person who worked with them, to draw on their memories to compile a list of every person they approached about a home invasion." (Id. at 10.) Such an undertaking would be "massive,"
As to the two remaining requests, the government represents that it has no additional information about the ATF's targeting criteria responsive to request # 3 that has not already been disclosed. (Id. at 12-13.) As to request # 4, the government rightly points out that Defendants' request is very broad and is not limited in temporal scope or to the issue of race. (Id. at 14.) As drafted, the request is likely to encompass a vast amount of sensitive law enforcement materials that are not relevant to Defendants' racial profiling defense. Defendants have already received a substantial amount of pertinent law enforcement documents as a result of the Court's prior
The Court also finds no basis to draw any adverse inference against the government, as Defendants request. (See Flowers , R. 706, Defs.' Post-Hearing Disc. Mot. at 13.) Defendants have not come forward with any evidence to suggest that the government intentionally destroyed existing documents or evidence for the purpose of concealing adverse information. See Faas v. Sears, Roebuck & Co. ,
In short, this Court carefully and diligently presided over the difficult and time-consuming discovery terrain dictated by the Seventh Circuit in Davis that has been subsequently followed by the Third Circuit in its careful Washington opinion. This Court did not favor the government or the defense in its delicate treatment of these sensitive law enforcement issues, which is borne out by the elaborate discovery record and the numerous oral and written rulings issued in these consolidated cases.
CONCLUSION
These delayed criminal cases must come to a conclusion soon. Defendants' attorneys raised serious, difficult issues that relate to the continuous historic tension our country's criminal law enforcement procedures have had with our country's racially diverse community. The future of these cases is squarely within the discretion of our new U.S. Attorney. This Court can only point out its concerns in this opinion and proceed to set priority trial dates in these long delayed cases. This matter will be set for a status hearing on March 21, 2018, at 10:30 a.m. for the express purpose of setting both cases for jury trials in the near future.
Notes
These statistics were calculated from 2010 census data for the 18 counties that make up this District.
Counsel for the government represented during oral arguments in this case that, thankfully, the government is no longer charging individuals in false stash house cases in this District. (See Evid. Hr'g Tr. at 527-28.)
The defendants in Brown are Abraham Brown, Kenneth Taylor, Alfred Washington, Christopher Davis, and Dwaine Jones. The defendants in Williams are Antonio Williams, John Hummons, and Howard Lee. The defense of these cases has been handled in a coordinated fashion, and nearly identical sets of motions to dismiss and other documents have been filed in the two cases. The Court refers to the cases collectively throughout this opinion.
The "stash house sting" operation was the brainchild of ATF Agent Richard Zayas, who has "made a career of coordinating the stings, having worked on or been involved with over 100 similar scenarios across the country," Flowers ,
The Court notes that the page numbers listed for record cites throughout this opinion refer to the page numbers assigned at the time the document was electronically filed with the Clerk of Court, not to the internal page numbers contained within the documents.
It appears that this document (a transcript of one of the recorded conversations) was inadvertently not uploaded at the time the government filed its voluminous exhibits. The Court refers to it by the document number assigned by the government.
Defendants were initially charged with certain drug offenses in addition to the charges outlined above, but as to all Defendants other than Jones, the drug charges were later voluntarily dismissed by the government. (Brown , R. 288, Gov't's Mot. to Dismiss;
Two other stash houses cases pending in this District, United States v. Payne , 12 CR 854 (N.D. Ill. filed Nov. 5, 2012), and United States v. Cousins , 12 CR 865 (N.D. Ill. filed Nov. 6, 2012), also arose from the fall 2012 law enforcement surge. (See R. 354-1, Gov't's Disc. Resps. at 43 n.1.)
On November 8, 2012, Agent Valles met with Williams and two individuals introduced to him by Williams to discuss details of the planned robbery, but these two individuals never showed up on the date of the takedown and are not defendants in this case. (See Williams , R. 354-1, Tr. at 522-61;
Defendants moved to suppress inculpatory statements they made while being transported to the ATF office in a police vehicle, which had been bugged with an audio device. See United States v. Williams ,
Lee has indicated his intent to plead guilty, and a change-of-plea hearing has been tentatively scheduled for March 2018. (R. 416, Min. Entry.) Lee has not formally joined in the motion to dismiss, although the Court has deferred holding his plea hearing until the motion is resolved.
The Court notes that Defendants do not assert any claim of "outrageous government conduct." See United States v. Russell ,
This Court previously granted Defendants leave to file an amended reply brief and a document correcting portions of the reply report prepared by Professor Fagan. (Williams , R. 403, Reply;
Defendants seek to distinguish McCleskey , arguing that it was based solely on a desire to avoid interference with prosecutorial discretion. (See Brown , R. 630, Defs.' Reply at 48.) Unfortunately, McCleskey was broadly written and was not solely about prosecutorial discretion. Instead, the Supreme Court explained that raw statistical data would not properly account for the many discretionary judgments made in connection with a decision to impose the death penalty, including those made by each individual jury based on a variety of permissible factors.
[E]ach particular decision to impose the death penalty is made by a petit jury selected from a properly constituted venire. Each jury is unique in its composition, and the Constitution requires that its decision rest on consideration of innumerable factors that vary according to the characteristics of the individual defendant and the facts of the particular capital offense. Thus, the application of an inference drawn from the general statistics to a specific decision in a trial and sentencing simply is not comparable to the application of an inference drawn from general statistics to a specific venire-selection or Title VII case. In those cases, the statistics relate to fewer entities, and fewer variables are relevant to the challenged decisions.
Notably, McCleskey was a 5-4 opinion, and Justice Lewis Powell, who authored the majority opinion, later publicly stated that he came to regret his decision and would change his vote if he could. See https://www.washingtonpost.com/archive/politics/1994/06/10/retired-justice-changes-stand-on-death-penalty/9ccde42b-9de5-46bc-a32a-613ae29d55f3/?utm_term=.41373c2298c0.
Professor Fagan describes probability weighting as again using "a logistic regression equation. The distinction in this analysis is that the procedure first estimates one logistic regression model to predict the status-in this case, race-and then uses another logistic regression model to predict the outcomes given the results of the first model.... If a predictor is statistically significant, it is presumed to be not independent of the outcome, but instead a predictor of that outcome." (Williams , R. 338-2, Fagan's Report at 28.) He describes propensity score matching as follows: "For each person in the 'treatment' group-Black or non-White people-one or more persons is selected from the 'control' group-White people-that are matched to the first group on all characteristics except race. This simulates random assignment to a treatment group-race -by matching persons on numerous predictors of treatment assignment.... The average treatment effect (ATE) is computed by taking the average of the difference in probability of selection between the observed and potential outcomes (Stash House defendant v. potential eligible) for each subject." (Id. at 29.)
The Court did not grant Defendants access to information about every individual contained in the ISP data with only an arrest, given the massive scope of such a data set, but Professor Fagan testified that this did not impact his analysis. (See Evid. Hr'g Tr. at 38.)
Professor Fagan originally testified that he obtained the additional criteria from "memoranda encouraging ATF to expand their search-their eligibility criteria to search for people based on those particular offense categories in addition to what's in the manual." (Evid. Hr'g Tr. at 27.) At a later point in the hearing, he stated that he had been mistaken in his earlier testimony and that there were no such memoranda; instead, he had obtained the additional targeting criteria from statements made by the prosecutor during oral arguments in Davis. (Id. at 78-79.) It is axiomatic that arguments made by counsel are not evidence, United States v. Diaz ,
The ISP data did not contain information about Hispanic ethnicity, so Professor Fagan used a convention, employed by the U.S. Census Bureau and other entities, for estimating Hispanic ethnicity using last names. (Williams , R. 338-2, Fagan's Report at 8-9; see also Evid. Hr'g Tr. at 40-42.) In effect, he treated individuals as Hispanic if the probability that they were Hispanic exceeded a certain threshold based on data pertaining to common Hispanic surnames. (Williams , R. 338-2, Fagan's Report at 8.) "Hispanic (60%) means that, based on their last name, a person is more than 60% likely to be Hispanic." (Id. ) The experts agreed that there is some level of error inherent in using this convention, usually involving the misclassification of Hispanics as white. (See
Professor Fagan used this same terminology-that the ATF "owns" the stash house stings-to explain why he analyzed all 94 stash house defendants together rather than just analyzing the initial targets. (See Brown , R. 603-3, Fagan's Reply Report at 34 ("A close examination of the recruitment of the defendants shows, in fact, that the ATF 'owns' the entire case, and therefore the correct unit of analysis is each of the 94 defendants."); Evid. Hr'g Tr. at 101-02 ("[W]e would have to make a judgment call, and the judgment call is whether or not to cluster the defendants or by the cases. Our position was that the ATF agents ... owned all the defendants in the cases because they had had extensive contact-63 out of 94-with individuals prior to the stash house operation."). Professor Fagan's assumption that the ATF "owns" the stash house stings does not appear to be based on his expertise in criminology or statistics, but rather, on a legal argument advanced by Defendants. At one point during the evidentiary hearing he explained that "my position that we took in my report was that before one of these operations was launched, that the ATF undercover officer, agent ... knew of and approved each of the members of the conspiracy ... and so they were aware of the criminal histories or criminal activity of those defendants." (Evid. Hr'g Tr. at 94.) There is little evidence in the record to support this assumption, and in fact it is contradicted by Defendants' own brief, which recognizes that the ATF did not meet a full 31 of the defendants until the day of their arrest. (Brown , R. 630, Defs.' Reply at 27.) Defendants do not argue in their brief that ATF was actually aware of the criminal histories of these individuals, but rather, that agents "ratified" the choice of the initial target when they let these individuals participate in the stings. (Id. ) The Court notes that Professor Fagan acknowledged at the hearing that although he read "several" of the takedown memoranda, he did not read them for all 24 cases. (See Evid. Hr'g Tr. at 146.)
Those cases were United States v. Davis , 13 CR 63, and United States v. Flowers , 11 CR 779. (See Williams , R. 354-1, Gov't's Ex. 3 at 94-100.)
The Court does not equate human beings to jelly beans, nor is there an intent to make light of the very serious interests at stake in this case. However, Professor Fagan's jelly bean analogy provides a useful simplification of complex statistical principles.
Professor Fagan did not run any tests omitting the 14 defendants from the FBI cases but he acknowledged that omitting them would have made the defendant pool "whiter." (Evid. Hr'g Tr. at 50-51, 122.) Professor Fagan testified at the hearing that Professor Schanzenbach ran those tests and "still found evidence of racial discrimination against black defendants," (Id. at 51), but this is not an accurate summary of Professor Schanzenbach's findings. Professor Schanzenbach made clear in his testimony that he ran various tests solely to determine how "sensitive" Professor Fagan's findings were-and he found them to be extremely sensitive, meaning that any slight change in the assumptions led to a significant change in the results. (Id. at 409, 413.) He tested one assumption at a time, including omitting the 14 defendants from the FBI cases, but accepted all other assumptions applied by Professor Fagan, including that clustering at the case level was unnecessary. (Brown , R. 555-1, Schanzenbach's Report at 17.) As explained herein, a number of Professor Fagan's conventions are problematic, including the lack of clustering and the decision to analyze all 94 defendants rather than just the initial targets. Professor Schanzenbach did not find evidence of discrimination at any level when he clustered by case level or analyzed just the initial targets. (Id. ; Evid. Hr'g Tr. at 349.) There was a separate question raised at the hearing as to whether findings can ever be "statistically significant" at the 10 percent level, as Professor Schanzenbach found in some instances. (See Evid. Hr'g Tr. at 315-17, 421-23.) Although Defendants argued at the hearing that 10 percent is a sufficient marker of statistical significance (see id. at 535-36), Professor Fagan himself acknowledged that findings at the 10 percent level would only be significant if they were part of a "pattern of findings which include more stringent levels ... 1 percent and 5 percent." (Id. at 459-60.) He acknowledged that "[a]s a member of an editorial board where I've reviewed hundreds and hundreds of articles, ... if I saw an article that said we're going to base our finding on a finding of .10, I would say, no, I don't think that's reliable." (Tr. at 460.) This was one of the few points of agreement between the experts, as Professor Schanzenbach also opined that findings solely at the 10 percent level would be a "hard sell in the social science literature." (Id. at 426.)
The Court notes that despite a suggestion to the contrary by defense counsel at the hearing (see Evid. Hr'g Tr. at 476), Defendants had information about the race of the confidential informants in their possession since the early stages of this case. (See Brown , R. 261, Order at 8-10, 12-13; id. , R. 171, Order at 2; Williams , R. 354-1, Gov't's Disc. Resps. at 60-69.)
Professor Schanzenbach reports that the initial targets were 75 percent black, 16.6 percent white, and 8.3 percent Hispanic. (Jt. Hr'g Ex. 5.)
As one of the Defendants in DeJesus put it, "To go in a house full of guys that are strapped, that's suicide." (Williams , R. 354-1, Tr. at 639.)
The Court accepts Professor Fagan's criticism that home invasion with a firearm does not present an exact match with a stash house robbery, given that a portion-he estimates approximately 16 or 17 percent-involve domestic disputes rather than a financial motive. (Evid. Hr'g Tr. at 230.) But there is some amount of guesswork in all of these comparisons, since this case involves a very unusual offense; as Professor Schanzenbach aptly noted, there is no "stash house robbery" offense reported in the crime data. (Id. at 253, 256, 374.) The Court considers the data pertaining to home invasions only in the overall context of the other offenses identified by Professor Schanzenbach. (See Brown , R. 555-1, Schanzenbach's Report at 7; Jt. Hr'g Ex. 5.)
To the extent Defendants have relied on employment discrimination case law in their briefs, that body of law also holds that "interest" is an important factor that must be taken into account when constructing a comparison group. See EEOC v. Chicago Miniature Lamp Works ,
At the evidentiary hearing, a number of comments were made about the fact that the home addresses of the eligibles group were not available to Professor Fagan. (See, e.g. , Evid. Hr'g Tr. at 154-55, 219.) To clarify the record, the Court did order ISP to turn over this data; however, the information was apparently not tracked for all cases and in Professor Fagan's view the data provided to him was "just not usable," (Id. at 219; see also Williams , R. 239, Mot. at 12.) The Court had also ordered ISP to turn over data showing the location of each arrest, but it was later revealed that ISP did not collect this data. (See Williams , R. 239, Mot. at 12-13;
According to a document submitted by the defense summarizing census data, Cook County's population is 24 percent black, 58 percent white, and 21 percent Hispanic. (Defs.' Hr'g Ex. 4.) The remaining counties are overwhelmingly white. DuPage County's population is 4 percent black, 82 percent white, and 11 percent Hispanic. (Id. ) Kane County's population is 5 percent black, 77 percent white, and 26 percent Hispanic. (Id. ) Kendall County's population is 5 percent black, 87 percent white, and 13 percent Hispanic. (Id. ) Lake County's population is 6 percent black, 80 percent white, and 17 percent Hispanic. (Id. ) LaSalle County's population is 2 percent black, 95 percent white, and 6 percent Hispanic. (Id. ) Will County's population is 11 percent black, 84 percent white, and 8 percent Hispanic. (Id. ) Winnebago County's population is 11 percent black, 84 percent white, and 8 percent Hispanic. (Id. )
Professor Schanzenbach explained the problem with the lack of a geographic marker in the matching analysis as follows: "[I]f you do a matching analysis where you're saying, here's somebody with a certain propensity, I'm going to match them to the eligibles data, you could identify someone in Joliet, right, who was ... equal to somebody in Rockford and the Rockford person is the defendant, but you're actually going to match them and say that the only reason the person in Joliet wasn't chosen was because of race[.]" (Evid. Hr'g Tr. at 275.) Professor Fagan believed that criminal propensity was an adequate proxy for geography, but he acknowledged at the hearing that it was "possible" that he had matched, for example, "a Black Disciple from Englewood with a member of a white supremacist gang in Joliet" when conducting his matching analysis. (Id. at 466.)
In the matching model, Professor Fagan also did not account for the fact that certain individuals appeared more than once in the data. For example, if a person had a qualifying conviction in 2000, they appeared 13 times in the data he used to do his matching analysis. (See Evid. Hr'g Tr. at 319-22, 513; Brown , R. 555-1, Schanzenbach's Report at 6.) Professor Schanzenbach provided compelling testimony that the failure to account for this anomaly made the results of the matching test unreliable. (See Evid. Hr'g Tr. at 320 ("Both Professor Fagan and myself agree that you at least have to cluster at the individual level because individuals are observed multiple times and we know that they're correlated with themselves. And if you don't do that, you're going to bias yourself towards finding a statistically significant result. The AIPW and matching analyses presented here do not account for clustering by individual.").) Professor Schanzenbach understood Professor Fagan to agree that there was a need to cluster by individual "[b]ecause he clusters at the individual level when he performs his logit regressions." (Id. at 321.) There was no available method to cluster by individual in the AIPW and matching analyses, however, which in Professor Schanzenbach's view made these methods inappropriate to test for discrimination in this context.
The Court considers that these high percentages could themselves be attributable to racial bias within the criminal justice system as a whole. (See Evid. Hr'g Tr. at 256-57.) But this would not prove race discrimination by the ATF , which is the issue before this Court. As Professor Schanzenbach noted, ATF must "tak[e] the world as they find it and ... that world has these [racial] disparities." (Id. at 257.)
Given the problems with Professor Fagan's analysis discussed at length above, the Court finds it unnecessary to probe the myriad other criticisms of his report outlined by Professor Schanzenbach. (See generally Brown , R. 555-1, Schanzenbach's Report at 5-24.)
Defendants seem to argue that a statistical showing can prove both discriminatory effect and discriminatory intent. (Brown , R. 630, Defs.' Reply at 47.) In support, they cite to civil cases brought under Title VII of the Civil Rights Act of 1964. (Id. (citing Int'l Bhd. of Teamsters v. United States ,
Defendants try to draw some nefarious inference from the fact that Agent Gomez chose "Blanco," which is Spanish for "white," as his nickname during the investigation, which is puzzling, since he clearly identified himself as Mexican during the recorded conversations. (See Williams , R. 338, Defs.' Mot. at 64 n.41.)
Ironically, Agent Gomez's narrative was apparently convincing to Defendants. During the meeting on August 12, 2012, Brown explained to Agent Gomez why he felt Agent Gomez's story was legitimate. (Brown , R. 555-1, Tr. at 385.) He said, "[N]o racist shit, but you Mexican, Like you gotta be Mexican to ... get close to the Mexicans to get they shit.... 'Cause you would've been black, I would've been skeptical of you[.]" (Id. at 386.)
Individual B is identified by name in the takedown memorandum, but given that he was never arrested or charged, the Court declines to name him here. (See Williams , R. 332-5, Takedown Mem. at 126.) Individual B's criminal record shows four felony convictions, including murder, manufacture/delivery of cocaine, and unlawful use of a firearm following a felony conviction. (Id. ) He also had arrests for battery, aggravated kidnapping, aggravated battery, carjacking, and unlawful use of a weapon. (Id. )
Defendants provide a detailed "departure analysis" containing a description of what they deem to be procedural deviations in all 24 stash house cases. (See Williams , R. 338, Defs.' Mot. at 40-62.) As with their statistical analysis on discriminatory effect, this analysis does not make any distinction between initial targets and defendants recruited by other members of the conspiracy. For example, if a defendant was not known or identified by ATF agents until the day of the takedown, it would be inappropriate to fault ATF for failing to ensure that this individual had the requisite criminal record to participate in the conspiracy. (Whether the decision to ensnare such individuals in a sting operation is an appropriate law enforcement strategy is another matter.) It also would not be appropriate for this Court to accept evidence of procedural irregularities in other stash house cases as proof that the decisionmakers in Williams and Brown had a discriminatory purpose in pursuing these investigations. See McCleskey ,
Defendants also offer anecdotal evidence about the racial makeup of two groups of "real" stash house robbers arrested in the Chicago area in unrelated cases. (See Williams , R. 338, Defs.' Mot. at 44-45.) Defendants believe that there are statistical anomalies when these "real" stash house robbers are compared to the stash house defendants. (Id. ) The Court agrees with Professor Schanzenbach that such a small sampling does not provide an appropriate basis for comparison. See Robinson v. Sinclair & Valentine, L.P. , No.
"Viable" is defined as "capable of existence and development as an independent unit," or "having a reasonable chance of succeeding." See Viable , Merriam-Webster Collegiate Dictionary (10th ed. 1997).
At a status hearing held in January 2018, defense counsel stated that Defendants were "content with the record" and that, in their view, "what the Court has in front of the Court is sufficient to determine [discriminatory] effect and intent." (R. 670, Tr. at 7, 9.) But later at that same hearing, defense counsel expressed a need for post-hearing discovery, which appeared to conflict with her earlier statements. (Id. at 9-10.) When pressed by the Court as to whether Defendants wanted a ruling based on the present record or whether they wanted an opportunity to seek additional discovery, counsel asked for leave to adopt the post-hearing discovery motion filed in United States v. Flowers , 11 CR 779. (See id. at 11-12; Flowers , R. 706; Brown , R. 669, Min. Entry.)
Notably, even though no order has ever been entered consolidating discovery matters in the many stash house cases in this District before this Court, Defendants chose to raise the bulk of their racial profiling-related discovery requests in Brown and Williams. (See Brown , R. 670, Tr. at 7-8.) In a few of the other cases, the presiding judges chose to adopt the discovery orders entered by this Court. (See Brown , R. 674, Gov't's Resp. to Defs.' Post-Hearing Disc. Mot. at 1 n.2.) In other cases, the presiding judges determined that Defendants were not entitled to any discovery or that only very limited discovery was warranted. (Id. ) In still other cases, Defendants chose to make no discovery requests. (Id. ) Given the strategic decisions that Defendants made, they can hardly be heard to complain that they were denied the ability to present their defense by virtue of this Court's rulings.
Defendants argue that the government should have existing documentation pertaining to any stash house investigation that was initiated. (R. 679, Defs.' Reply in Supp. of Mot. for Post-Hearing Disc. at 6-7.) But Defendants' discovery request, as drafted, seeks a far larger subset of materials. Specifically, Defendants seek information about "the full universe of individuals with whom the ATF or its CIs had contact in this district regarding a Stash House Operation" during a seven-year period. (Flowers , R. 706, Defs.' Post-Hearing Disc. Mot. at 3 (emphasis added).) There is nothing before the Court to suggest that the government has ready access to documents that meet this broad criteria.
Defendants question whether this is the case, as they have obtained information showing that the ATF created an electronic database for its CIs in 2016. (See R. 680, Defs.' Sealed Reply in Supp. of Mot. for Post-Hearing Disc. at 8-9.) But that was well after the period for which Defendants' seek information, and the government has convincingly explained that it would still have to comb through paper documents to respond to Defendants' discovery request seeking data for 2006-13. (See R. 680-3, Sealed Letter from U.S. Attorneys' Office at 2-3.)