Selective State Remands Still Count as Merits Adjudications Under AEDPA—and Off‑Bench Misconduct Alone Does Not Prove Judicial Bias

Commentary on Jimenez v. Guerrero, No. 23-50223 (5th Cir. Apr. 4, 2025)

Introduction

In Jimenez v. Guerrero, the Fifth Circuit denied federal habeas relief to a Texas prisoner who alleged his trial was tainted by judicial bias. The case posed two interlocking questions at the core of modern habeas review: (1) When state courts handle some claims expressly and remain silent about others, does the silent claim receive deference under the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA)? (2) Do off-the-bench ethical violations by a trial judge—here, receipt of misappropriated travel funds from the prosecutor—establish actual or presumptive judicial bias under the Due Process Clause?

The panel (Judge Oldham writing; Judges King and Jones joining) answers both questions against the petitioner. First, it holds that Texas’s summary treatment of the judicial-bias claim is presumed to be an adjudication on the merits, so AEDPA’s relitigation bar applies. Second, even reviewing the merits through AEDPA’s lens, the court finds no clearly established Supreme Court law that would compel a bias finding on the facts presented. The court also denies a motion to expand the certificate of appealability (COA) to reach ineffective-assistance and Brady claims and declines to consider a late-raised, unexhausted cumulative-prejudice claim.

Summary of the Opinion

• AEDPA deference applies: The state courts’ handling of Jimenez’s judicial-bias claim constitutes an adjudication on the merits under the Supreme Court’s Richter/Williams presumption, even though the Texas Court of Criminal Appeals (TCCA) remanded for findings only on other claims and later denied relief without a written opinion.
• No constitutional bias shown: The trial judge’s off-the-bench receipt of travel reimbursements from the district attorney—however “troubling” and “unacceptable”—did not establish either actual bias (pecuniary interest in the case) or presumptive bias (a constitutionally intolerable probability of bias) under clearly established Supreme Court precedent (Tumey and Caperton).
• COA expansion denied: The petitioner failed to make a “substantial showing of the denial of a constitutional right” on the ineffective-assistance and Brady claims; the district court’s merits rulings on those issues were not debatable among jurists of reason.
• Procedural bar for cumulative prejudice: The cumulative-prejudice theory was not raised in state court and surfaced only in a reply brief in federal court; it is procedurally barred for failure to exhaust.
• Ancillary motions: Appointment of counsel was denied as moot (no oral argument).

Analysis

AEDPA’s Merits-Adjudication Presumption Controls

The linchpin of the decision is AEDPA’s “relitigation bar,” which precludes federal courts from relitigating claims “adjudicated on the merits” in state court unless the petitioner meets the exacting standards in 28 U.S.C. § 2254(d)(1)–(2). The court frames the question through the Supreme Court’s seminal holdings in Harrington v. Richter and Johnson v. Williams:

• Even a one-line summary denial is presumed to be a merits adjudication (Richter).
• The presumption applies when a state court addresses some claims but not others in a reasoned decision (Williams). Silent claims are presumed rejected on the merits unless the petitioner shows they were inadvertently overlooked—a showing the Supreme Court describes as possible only in “unusual circumstances.”

Applying that framework, the Fifth Circuit concludes that Jimenez’s judicial-bias claim was adjudicated on the merits:

• Jimenez’s state habeas application raised judicial bias alongside ineffective assistance and Brady claims.
• The TCCA remanded for findings on ineffective assistance and Brady (and “any other” relevant findings), but the trial court limited the hearing to the two specified claims and entered findings rejecting them.
• On return from remand, the TCCA summarily denied relief “without written order on findings of trial court after hearing.”

From this sequence, the court infers the Texas courts were aware of the bias claim and rejected it on the merits (twice): first implicitly when the TCCA remanded only on other issues and the trial court construed that remand to exclude bias; and again when the TCCA summarily denied relief following the remand proceedings. The panel stresses that state courts need not “address every single claim,” and can treat some claims as too insubstantial to warrant discussion. In short, the district court’s de novo review “out of an abundance of caution” was unnecessary because AEDPA plainly applies.

Precedents Cited and Their Influence

Richter and Johnson v. Williams: The Presumption of Merits Adjudication

Harrington v. Richter (562 U.S. 86) and Johnson v. Williams (568 U.S. 289) anchor the opinion. They establish that summary denials and silent treatment of some claims still count as merits adjudications for AEDPA purposes, absent rare, clearly demonstrated oversight. The Fifth Circuit applies these cases to hold that the TCCA’s silence on bias does not strip AEDPA deference; rather, it triggers it.

Dunn v. Reeves and State-Court Workload

Dunn v. Reeves (594 U.S. 731) reinforces that state courts need not produce reasoned opinions addressing each claim; that institutional reality supports the presumption that silence equals adjudication on the merits.

Wilson v. Sellers: Look-Through Inapplicable

Wilson v. Sellers (584 U.S. 122) allows federal courts to “look through” an unexplained decision to the last reasoned state decision. Here, no state court ever issued a reasoned opinion on the bias claim, so Wilson’s look-through doctrine does not apply. This matters because AEDPA analysis proceeds without the benefit of articulated state-court reasoning.

Terry Williams v. Taylor and the AEDPA Standards

Terry Williams v. Taylor (529 U.S. 362) provides the backbone for AEDPA: clearly established law means Supreme Court holdings as of the time of the state-court decision; a decision is “contrary to” such law only if it reaches an opposite conclusion on a question of law or on materially indistinguishable facts; “unreasonable application” requires an error beyond fairminded disagreement. The panel repeatedly returns to these constraints.

White v. Woodall, Shinn v. Kayer, Shoop v. Hill, Woods v. Etherton, Sexton v. Beaudreaux

These cases collectively underscore AEDPA’s stringency: no extension of precedent to new contexts; relief only when a state court’s error is beyond any possibility for fairminded disagreement; mere error or clear error does not suffice. The court uses this cluster to reject any attempt to broaden judicial-bias doctrine beyond established Supreme Court holdings.

The Bias Canon: Tumey, Caperton, Bracy (and Withrow, Balisok distinguished)

• Tumey v. Ohio (273 U.S. 510): Due process violation when a judge has a direct, personal, substantial pecuniary interest in conviction (or a structural scheme that effectively creates such an interest). The court distinguishes Tumey because there was no statute or structure giving Judge Prohl a personal stake in Jimenez’s conviction; there was a jury; and no fines or compensation turned on conviction.
• Caperton v. A.T. Massey Coal Co. (556 U.S. 868): Due process violation where disproportionate and temporally pertinent campaign expenditures created a constitutionally intolerable probability of bias. The court finds Jimenez’s case “nothing like Caperton” because the travel reimbursements were not a party’s expenditures in the case, had no timing tied to Jimenez’s trial or appeal, and did not have a significant or disproportionate influence comparable to Caperton.
• Bracy v. Gramley (520 U.S. 899): The Supreme Court permitted discovery where a judge’s bribery in other cases supported a plausible theory of bias in the petitioner’s case (to “deflect suspicion”). The Fifth Circuit highlights that Bracy was a discovery ruling, not a merits grant, and involved case-specific indicators (including counsel’s links to the judge). Those indicia are absent here.
• Withrow v. Larkin (421 U.S. 35) and Edwards v. Balisok (520 U.S. 641) are inapposite to establishing bias here (Withrow concerned combined investigative and adjudicative roles; Balisok addressed the non-cognizability of certain § 1983 claims).

Fifth Circuit Parameters: Buntion, Richardson, Langley, Senn, Lucio

• Buntion v. Quarterman and Richardson v. Quarterman emphasize that constitutional bias is “not lightly established” and requires “extreme facts.”
• Langley v. Prince (en banc) and Senn v. Lumpkin reiterate AEDPA’s demanding “no fairminded jurist could agree” threshold.
• Lucio v. Lumpkin (en banc plurality) and the court’s footnote explicitly caution that pre-Richter Fifth Circuit tests (e.g., the Woodfox/Mercadel three-part framework for distinguishing merits adjudications from procedural denials) cannot displace the Supreme Court’s Richter/Williams approach.

Legal Reasoning

1) AEDPA Applies: The Silent Bias Claim Was Adjudicated on the Merits

The court rejects de novo review. It holds that Jimenez could not meet the heavy burden to “very clearly” show that the TCCA “inadvertently overlooked” his judicial-bias claim. The record—including the remand colloquy in which the trial court limited issues to ineffective assistance and Brady—confirms that the state bench knew about the bias claim but did not treat it as meriting separate treatment. When the TCCA later denied relief without written order on the remand findings, the presumption of a merits adjudication attached again. Under Richter/Williams, that is enough to trigger AEDPA.

2) Merits Under AEDPA: No Clearly Established Supreme Court Law Compels a Bias Finding

Proceeding under § 2254(d)(1), the panel reasons that neither the “contrary to” nor “unreasonable application” prongs is satisfied:

• No materially indistinguishable Supreme Court case: The judge’s receipt of travel reimbursements from the prosecutor—a wrongdoing for which the judge resigned and pleaded guilty—does not map onto Tumey (no direct pecuniary interest in this case or contingent compensation scheme) or Caperton (no disproportionate, case-tied campaign spending by a party with a stake in the outcome). Bracy did not decide a merits bias question and rested on case-specific discovery triggers absent here.
• No unreasonableness beyond fairminded disagreement: Given the Supreme Court’s caution that recusal standards cannot be defined with precision and set only a constitutional floor, reasonable jurists could conclude that the described misconduct—however “troubling”—did not meet the “extreme facts” required for a constitutional violation.

The court also underscores timing and scope: “Clearly established” means Supreme Court holdings existing when the TCCA ruled in 2014. Williams v. Pennsylvania (2016) therefore cannot be considered.

3) COA Expansion and Procedural Bar

On the ineffective-assistance and Brady claims, the court declines to enlarge the COA because the district court’s rejection of those claims (and its refusal to issue a COA) was not debatable among jurists of reason. As to the cumulative-prejudice claim, the court finds it procedurally barred for lack of exhaustion and because it emerged only in a reply brief in federal court.

Impact

• Reinforced presumption of merits adjudication: The decision fortifies Richter/Williams in the Fifth Circuit. Even when the TCCA’s remand instructs findings on some claims and omits others, and even when the trial court then limits the hearing accordingly, the silent claim is still presumed denied on the merits. Federal habeas courts must apply § 2254(d) unless the petitioner makes the exceedingly rare showing of inadvertent state-court oversight.
• Constrained judicial-bias habeas claims: Off-the-bench ethical breaches—here, a judge’s receipt of misappropriated travel funds from the prosecutor—do not by themselves cross the constitutional line without a case-specific, direct, personal stake (Tumey) or a constitutionally intolerable probability of bias akin to Caperton’s extraordinary facts. Petitioners will need evidence tying the misconduct to their case or to a structural incentive that affects adjudication.
• Signal on prior Fifth Circuit “three-part” tests: The court’s explicit reliance on Williams and skepticism toward pre-Richter circuit tests (e.g., Woodfox/Mercadel) clarifies the governing framework in this circuit and reduces room for alternative, judge-made approaches to rebutting the presumption.
• Practical litigation consequences: Defense counsel must fully develop, plead, and exhaust bias theories in state court with case-specific facts. Raising cumulative-prejudice or new theories late (e.g., in a federal reply brief) risks procedural default. On COAs, Buck/Miller-El’s “debatability” threshold continues to screen out issues that federal courts find straightforward under AEDPA.
• Jury trials matter: The court repeatedly notes the presence of a jury in distinguishing Tumey and in assessing prejudice. Where juries decide guilt and punishment, petitioners alleging due process bias against a judge must show how the judge’s conduct implicated a protected, case-specific interest or critically affected the fairness of proceedings.

Complex Concepts Simplified

• AEDPA’s “relitigation bar” (28 U.S.C. § 2254(d)): A federal court cannot grant habeas relief on a claim already decided by a state court unless the state decision contradicts or unreasonably applies Supreme Court holdings, or rests on an unreasonable view of the facts based on the state-court record.
• “Adjudicated on the merits”: A state court is presumed to have decided a claim on its substance—even in a summary denial or when it writes about some claims and not others—unless the petitioner clearly shows the claim was inadvertently overlooked.
• “Clearly established Federal law”: Only holdings (not dicta) of the U.S. Supreme Court at the time of the last state-court merits decision count.
• Actual vs. presumptive bias:
  • Actual bias: The judge has a direct, personal, substantial pecuniary interest in the outcome (Tumey) or equivalent.
  • Presumptive bias: An objective, constitutionally intolerable probability of bias (Caperton), found only in “extreme” circumstances.
• Wilson “look-through”: When the final state decision is unexplained, a federal court ordinarily looks through to the last reasoned state decision for the rationale. If none exists, there is nothing to look through to.
• Certificate of Appealability (COA): A jurisdictional gateway to appeal in habeas cases. It issues only if reasonable jurists could debate the district court’s resolution or the claim deserves encouragement to proceed.
• Procedural default and exhaustion: Federal habeas relief generally requires that claims be presented to and decided by state courts first. Claims not exhausted (or raised too late in federal court) are usually barred.

Conclusion

Jimenez v. Guerrero provides a crisp reaffirmation of AEDPA’s core architecture in the Fifth Circuit. Two messages stand out. First, the Richter/Williams presumption is robust: state-court silence—particularly when paired with selective remand orders and summary denials—means the claim was adjudicated on the merits, triggering AEDPA deference. Rebutting that presumption will be “unusual” at best. Second, the constitutional floor for judicial recusal is very low. Off-the-bench ethical violations, even serious ones involving the prosecutor, do not automatically render a judge constitutionally biased. Absent a direct, case-specific pecuniary stake or Caperton-level “extreme facts,” due process is not violated.

Beyond its immediate disposition—affirming denial of habeas relief and declining to expand the COA—the opinion strengthens the doctrinal guardrails that cabin federal habeas review and narrows avenues for judicial-bias claims that rely on generalized misconduct rather than case-specific, constitutionally cognizable interests. For litigants and courts alike, Jimenez underscores that AEDPA deference is the rule, and only truly extraordinary malfunctions trigger habeas relief.