Selective Enforcement Discovery in Prison Contraband Cases: Small Samples, Self‑Selection, and Mandatory Referral Policies Can Justify Denial

Introduction

In United States v. Brian Foster (3d Cir. Apr. 2, 2025) (nonprecedential), the Third Circuit affirmed a district court’s denial of pretrial discovery sought to support a race‑based selective enforcement claim in a prison contraband prosecution. While incarcerated at FCI Schuylkill, Brian Foster was found with a sharpened piece of plexiglass with a tape handle in his pocket, admitted it was a weapon, and was indicted under 18 U.S.C. § 1791(a)(2), (b)(3). Rather than contest the merits, Foster sought discovery to develop a claim that contraband investigations and referrals at the institution were selectively enforced by race.

The appeal centers on the threshold for obtaining discovery on a selective enforcement theory under United States v. Washington, 869 F.3d 193 (3d Cir. 2017). The court held that the district court did not abuse its discretion in concluding Foster’s proffer—built on a small, methodologically fragile dataset and weakened by evidence that the institution had a mandatory referral policy and a demographically small white population—did not justify further discovery.

Summary of the Opinion

The panel (Judge Matey writing; Judges Shwartz and McKee joining, with Judge McKee concurring) affirmed. Applying Washington’s discovery framework for selective enforcement claims and reviewing for abuse of discretion, the court concluded it was reasonable for the district court to deny Foster’s discovery request. Key points:

The defendant’s proffer showed a disparity in prosecutions for contraband-only offenses but rested on a small sample (approximately 39 cases) and did not account for the institution’s demographics (only 18% of FCI Schuylkill’s 1,027 inmates were white).
The crime at issue involved “self-selection” (inmates chose to possess contraband), reducing the opportunity for biased targeting often implicated in selective enforcement claims.
The government voluntarily produced information that undermined Foster’s inference, including a policy to refer all contraband offenses for prosecution—suggesting diminished investigative discretion.
Given the limited and weakened proffer, and mindful of judicial economy, the district court acted within its discretion in denying additional discovery.

Judge McKee concurred in the judgment but emphasized that small samples can sometimes support reliable inferences. He agreed, however, that Foster’s particular dataset was methodologically flawed and thus not “reliable statistical evidence, or its equivalent” under Washington.

Analysis

Precedents Cited and Their Role

United States v. Armstrong, 517 U.S. 456 (1996) and Oyler v. Boles, 368 U.S. 448 (1962): Armstrong articulates that prosecutions may not be based on impermissible classifications such as race and requires “clear evidence” of discriminatory effect and intent to prevail on a selective prosecution claim. The panel cites Armstrong and Oyler to restate the constitutional baseline and the demanding merits standard.
United States v. Washington, 869 F.3d 193 (3d Cir. 2017): Washington distinguishes selective prosecution (prosecutorial decisions) from selective enforcement (law enforcement decisions). For discovery on selective enforcement claims, Washington relaxes Armstrong’s discovery threshold: defendants need a proffer with “reliable statistical evidence, or its equivalent,” strong enough to support a reasonable inference of discriminatory intent—while courts remain attentive to judicial economy. The panel applied Washington’s framework and emphasized that the “end goal” of discovery is a substantively viable selective enforcement claim; where preliminary material undermines that prospect (e.g., minimal discretion), courts may deny discovery.
Blunt v. Lower Merion Sch. Dist., 767 F.3d 247 (3d Cir. 2014): Cited for the caution that small sample sizes can undercut the probative value of statistical evidence. This principle supported the district court’s skepticism of Foster’s 39‑case dataset.
United States v. Chemical Foundation, Inc., 272 U.S. 1 (1926): Quoted in Armstrong for the “clear evidence” standard on the merits. The panel invokes it to underscore that discovery is a gateway to a demanding substantive claim.
United States v. Collins, 36 F.4th 487 (3d Cir. 2022) and Democratic Nat’l Comm. v. Republican Nat’l Comm., 673 F.3d 192 (3d Cir. 2012): Provide the abuse‑of‑discretion review standard—reversal is warranted only if the district court’s decision is “arbitrary, fanciful, or clearly unreasonable.” This deferential lens heavily influences the outcome.

Legal Reasoning

The court’s analysis proceeds in three steps: (1) identify the discovery standard for selective enforcement claims, (2) evaluate the defendant’s proffer under that standard (including new facts voluntarily produced by the government), and (3) apply deferential appellate review to the district court’s judgment about whether discovery is warranted.

Discovery standard under Washington. A defendant seeking discovery on selective enforcement must present “some evidence” of discriminatory effect through “reliable statistical evidence, or its equivalent,” strong enough to support a reasonable inference of discriminatory intent. Courts must still weigh judicial economy and avoid protracted pretrial litigation.
Evaluation of Foster’s proffer. The district court found that Foster’s figures suggested a racial disparity in contraband-only prosecutions, but the inference was undermined by:
- Small sample size: Approximately 39 cases over three years is limited; small samples diminish inferential strength.
- Institutional demographics: Only 18% of inmates at FCI Schuylkill were white; a single white prosecution cannot be assessed without that context.
- Self-selection and completed conduct: In prison contraband cases, enforcement follows inmates’ choice to possess contraband. Unlike scenarios where officers choose targets in advance, the opportunity for pre‑contact bias is comparatively reduced.
- Policy eliminating discretion: The institution’s policy to refer all contraband offenses for prosecution materially weakens any theory that investigators selectively chose which cases to refer based on race.
The panel agreed that, taken together, these considerations made the proffer too weak to warrant discovery.
Abuse‑of‑discretion review. Applying the highly deferential standard, the panel held the district court’s decision was neither arbitrary nor unreasonable. The additional government disclosures (institutional demographics; mandatory referral policy) further depressed the probative value of Foster’s already thin statistics.

The Concurrence (Judge McKee): Methodology Matters

Judge McKee concurred in the judgment, emphasizing that small samples are not inherently unreliable; if representative and methodologically sound, they can support reasonable inferences. But he found Foster’s proof wanting because it lacked “reliable statistical evidence, or its equivalent,” highlighting several methodological defects:

Questionable data collection: Reliance on ECF coding that could miss sealed/miscoded cases; assumptions about defendants’ race based on surnames and an unverified website; missing race data for some defendants.
Improper comparator and failure to disaggregate: Initial reliance on nationwide BOP demographics instead of FCI Schuylkill’s; lumping together all contraband categories (weapons, drugs, cellphones, currency) rather than analyzing weapons‑only cases (the charge Foster faced). The weapons‑only subset was even smaller, further eroding statistical reliability.
Arbitrary time window: Unexplained three‑year span that extended past Foster’s indictment date, shrinking the relevant weapon‑only dataset even more when corrected.
Inaccuracy within a tiny dataset: A superseding indictment eliminated one case from the 39 but without clarifying which case, race, or contraband category—significant because small changes materially shift inferences in small samples.

Judge McKee suggested that a stronger proffer might include expert analysis validating the dataset, institution‑specific and offense‑specific breakdowns, prior investigative findings or lawsuits involving the same facility, and affidavits from witnesses—illustrating how litigants can meet Washington’s threshold.

Impact and Practical Implications

Although nonprecedential, Foster provides a concrete application of Washington’s discovery threshold in the prison contraband context and offers practical guidance for both defendants and the government.

For defendants seeking discovery on selective enforcement:

Build institution‑specific, offense‑specific datasets: Use the correct denominator (FCI Schuylkill’s population in this case) and focus on the specific contraband category charged (e.g., weapons), not mixed categories.
Demonstrate investigative discretion (and its exercise): If an institution truly has a mandatory referral policy, discovery must plausibly show exceptions, inconsistent application, or ways discretion actually exists (e.g., at the incident reporting or referral stages).
Ensure reliability: Avoid surname‑based race coding; seek official race data, incident reports, disciplinary records, and USAO referral forms; correct for sealed or miscoded cases; provide expert statistical analysis.
Define a principled timeframe: Justify the period and avoid post‑indictment cases unless relevant to the alleged pattern.
Augment statistics with qualitative evidence: Affidavits from witnesses or staff, prior findings or investigations at the same facility, and policy documents showing how discretion operates in practice can fortify the proffer.

For the government and correctional institutions:

Document neutral policies and compliance: Written policies (e.g., “refer all contraband offenses”) and consistent application records can blunt inferences of discriminatory selection.
Voluntary targeted disclosures can be dispositive: Producing key demographic and policy data early may undercut a defendant’s statistical inference and support denial of discovery on judicial economy grounds.
Maintain accurate, searchable records: Clear incident, referral, and disciplinary tracking reduces disputes about completeness and reliability.

For courts applying Washington:

Balance reliability and economy: The court can deny discovery where the proffer is small, methodologically infirm, and where institutional policies substantially eliminate discretion.
Context matters: “Self‑selection” in completed contraband offenses differs from contexts where law enforcement selects targets ex ante; the opportunity for discriminatory selection is narrower.
Deferential appellate review: The “arbitrary, fanciful, or clearly unreasonable” standard insulates careful gatekeeping decisions on discovery from reversal.

Complex Concepts Simplified

Selective prosecution vs. selective enforcement:
- Selective prosecution challenges decisions by prosecutors to bring charges; discovery requires a credible showing that similarly situated persons were not prosecuted (Armstrong/Bass).
- Selective enforcement challenges decisions by law enforcement (or those acting in that capacity) to investigate or refer suspects; under Washington, discovery may be ordered on a lower threshold—“reliable statistical evidence, or its equivalent” that is strong enough to support a reasonable inference of discriminatory intent—mindful of judicial economy.
“Reliable statistical evidence”: Not just numbers; it means methodologically sound data: correct comparator population, accurate race coding, disaggregated by the relevant offense category, sufficient size, and expert validation where helpful.
“Self‑selection” in prison contraband: Unlike operations where officers choose targets, contraband cases often begin because an inmate chooses to possess contraband and is discovered. This reduces the role of ex ante targeting discretion and, correspondingly, the traction of a selective enforcement theory absent additional evidence.
Mandatory referral policies: If a facility requires that all contraband incidents be referred for prosecution, there may be little or no operative discretion to apply unequally. Proving selective enforcement then requires showing deviations from, or discriminatory application of, that policy.
Abuse of discretion: An appellate standard asking whether the district court’s decision was “arbitrary, fanciful, or clearly unreasonable.” It is highly deferential and often decisive on appeal.

Conclusion

Foster reaffirms Washington’s pragmatic gatekeeping for discovery on selective enforcement claims. A defendant must offer reliable, institution‑specific statistics or their equivalent, sufficient to support a reasonable inference of discriminatory intent, while courts remain mindful of judicial economy. In the prison contraband setting, the combination of small and methodologically weak datasets, evidence of institutional demographics, the self‑selecting nature of the offense, and a documented policy that eliminates investigative discretion can justify denying discovery.

Judge McKee’s concurrence adds an important methodological gloss: small samples are not per se unreliable, but their probative value depends on sound collection, correct comparators, offense‑specific disaggregation, coherent timeframes, and accurate coding—preferably supported by expert analysis and corroborating qualitative evidence.

Though nonprecedential, the decision offers a practical roadmap. For defendants, it underscores the need for rigorous statistics and concrete evidence of discretionary, unequal enforcement. For the government and institutions, it highlights the value of clear, consistently applied policies and targeted disclosures that address and deflate speculative inferences. In the broader equal protection landscape, Foster illustrates how Washington’s discovery standard functions as an efficient filter—allowing robust claims to proceed while curtailing pretrial litigation where the proffer fails to clear the reliability threshold.

Citations: United States v. Foster, No. 24-1538 (3d Cir. Apr. 2, 2025) (nonprecedential); United States v. Washington, 869 F.3d 193 (3d Cir. 2017); United States v. Armstrong, 517 U.S. 456 (1996); Oyler v. Boles, 368 U.S. 448 (1962); United States v. Chemical Found., Inc., 272 U.S. 1 (1926); Blunt v. Lower Merion Sch. Dist., 767 F.3d 247 (3d Cir. 2014); United States v. Collins, 36 F.4th 487 (3d Cir. 2022); Democratic Nat’l Comm. v. Republican Nat’l Comm., 673 F.3d 192 (3d Cir. 2012).