Bufkin v. Collins

604 U.S. 369

• Court: Supreme Court of the United States
• Decided: March 5, 2025
• Docket: 23-713
• Status: Published

Opinion

PRELIMINARY PRINT

Volume 604 U. S. Part 1 Pages 369–407

OFFICIAL REPORTS OF

THE SUPREME COURT March 5, 2025

REBECCA A. WOMELDORF reporter of decisions

NOTICE: This preliminary print is subject to formal revision before the bound volume is published. Users are requested to notify the Reporter of Decisions, Supreme Court of the United States, Washington, D. C. 20543, pio@supremecourt.gov, of any typographical or other formal errors. OCTOBER TERM, 2024 369

Syllabus

BUFKIN v. COLLINS, SECRETARY OF VETERANS AFFAIRS

certiorari to the united states court of appeals for the federal circuit No. 23–713. Argued October 16, 2024—Decided March 5, 2025* The Department of Veterans Affairs (VA) applies a “beneft-of-the-doubt rule” that tips the scales in a veteran's favor when evidence regarding any issue material to a service-related disability claim is in “approxi- mate balance.” 38 U. S. C. § 5107(b). Petitioners are veterans who ap- plied for service-connected post-traumatic stress disorder (PTSD) dis- ability benefts and were dissatisfed with the VA's resolution of their claims. Petitioner Joshua Bufkin claimed that his PTSD stemmed from his military service, but the VA found no clear link. Petitioner Norman Thornton obtained service-connected PTSD disability benefts, but the VA denied his most recent request to increase his disability rating. These adverse determinations were reviewed de novo by the Board of Veterans' Appeals, which rendered fnal decisions on behalf of the VA denying the claims. Petitioners then challenged the adverse determi- nations before the U. S. Court of Appeals for Veterans Claims (Veterans Court). Under § 7261(a), the Veterans Court reviews legal issues de novo and factual issues for clear error. And under § 7261(b)(1), the Vet- erans Court must “take due account” of the VA's application of the beneft-of-the-doubt rule. Applying those standards, the Veterans Court affrmed the VA's adverse beneft determinations, fnding that the Board's approximate-balance determinations were not clearly errone- ous. Petitioners then appealed to the Federal Circuit, challenging the Veterans Court's legal interpretation of § 7261(b)(1), and arguing that the statutory command to “take due account” of the VA's application of the beneft-of-the-doubt rule requires the Veterans Court to review the entire record de novo and decide for itself whether the evidence is in approximate balance. The Federal Circuit rejected this argument and affrmed. Held: The VA's determination that the evidence regarding a service- related disability claim is in “approximate balance” is a predominantly factual determination reviewed only for clear error. Pp. 379–388.

*Together with Thornton v. Collins, Secretary of Veterans Affairs (see this Court's Rule 12.4), also on certiorari to the same court. 370 BUFKIN v. COLLINS

(a) Under § 7261(b)(1)'s plain text, the Veterans Court must “take due account” of the VA's application of the beneft-of-the-doubt rule. This requirement directs the Veterans Court to give appropriate attention to the VA's work. The Veterans Court must review the VA's application of the beneft-of-the-doubt rule “[i]n making the determinations under subsection (a).” § 7261(b)(1). Accordingly, the standards of review provided in subsection (a) also govern the Veterans Court's review of beneft-of-the-doubt issues. Section 7261(b)(1) makes explicit the Vet- erans Court's previously implicit duty to review the VA's application of the beneft-of-the-doubt rule, pursuant to the standards set forth in subsection (a). Pp. 379–380. (b) The appropriate standard of review for any given challenge de- pends on whether the challenge is factual or legal in nature. The par- ties contest whether a veteran's challenge to the VA's determination that the evidence on a particular material issue is not in approximate balance involves a legal inquiry subject to de novo review, or a factual fnding, or at least a predominantly factual, mixed question of law and fact, subject to clear-error review. The approximate-balance determination involves two steps. First, the VA reviews each item of evidence and assigns weight to it—a fact- fnding inquiry reviewed only for clear error. Second, the VA deter- mines whether the evidence is in approximate balance. See § 5107(b). This second step includes both legal and factual components: factual because it involves marshaling and weighing evidence, and legal because the “approximate balance” determination involves whether the evidence satisfes a legal standard. The VA's approximate-balance determination is thus at most a mixed question. And the appropriate standard of review for a mixed question depends “on whether answering it entails primarily legal or factual work.” U. S. Bank N. A. v. Village at Lake- ridge, LLC, 583 U. S. 387, 396. Reviewing a determination whether record evidence is approximately balanced is “about as factual sounding” as any question gets. Id., at 397. In Bufkin's case, the Board weighed medical opinions and family testimony to assess his PTSD claims. Similarly, the Board in Thorn- ton's case analyzed symptom severity and medical evidence to assess his disability rating. Both cases demonstrate that approximate-balance determinations require case-specifc factual review warranting clear- error review. Pp. 380–383. (c) Petitioners' counterarguments are unpersuasive. First, peti- tioners urge that by amending § 7261(b)(1) to include the modest phrase, “take due account,” Congress imposed a new standard of review for challenges to the VA's application of the beneft-of the-doubt rule. But had Congress intended to do so, it would have identifed a standard, just as it did in § 7261(a). Petitioners next argue that, even if § 7261(b)(1) Cite as: 604 U. S. 369 (2025) 371

incorporates § 7261(a)'s standards of review, the VA's approximate- balance determination is much like a court's probable-cause determina- tion, which involves a mixed-question inquiry that appellate courts re- view de novo. Ornelas v. United States, 517 U. S. 690, 699. Two features distinguish the probable-cause determination from the VA's determination here. First, probable cause is a constitutional standard, creating a strong presumption that determinations under that standard are subject to de novo review. By contrast, the approximate- balance determination is a creature of statute, not the Constitution. Second, probable cause poses a question that requires substantial “legal work,” U. S. Bank, 583 U. S., at 398, but the VA's approximate-balance determination lacks a comparable legal component. Petitioners contend that the Federal Circuit's reading of § 7261(b)(1) renders the provision superfuous. While this Court's reading of § 7261(b)(1) might involve some redundancy, the canon against surplus- age does not apply here because petitioners have not identifed a com- peting interpretation that would avoid redundancy. See Marx v. Gen- eral Revenue Corp., 568 U. S. 371, 385. Pp. 383–387. 75 F. 4th 1368, affrmed.

Thomas, J., delivered the opinion of the Court, in which Roberts, C. J., Page ProofKagan, and Alito, Sotomayor, Pending Kavanaugh, and Publication Barrett, JJ., joined. Jackson, J., fled a dissenting opinion, in which Gorsuch, J., joined, post, p. 388.

Melanie L. Bostwick argued the cause for petitioners. With her on the briefs were Thomas M. Bondy, Jonas Q. Wang, Melanie R. Hallums, Katherine M. Kopp, Kenneth M. Carpenter, and Geoffrey C. Shaw. Sopan Joshi argued the cause for respondent. With him on the brief were Solicitor General Prelogar, Principal Deputy Assistant Attorney General Boynton, Deputy Solic- itor General Stewart, Evan Wisser, Meredyth Cohen Ha- vasy, and Christopher O. Adeloye.† †Briefs of amici curiae urging reversal were fled for Disabled Ameri- can Veterans by Amy F. Odom, Robert V. Chisholm, and Zachary Stolz; for the Federal Circuit Bar Association by Brian T. Burgess; for Military- Veterans Advocacy, Inc., by Michael E. Joffre, Anna G. Phillips, and John B. Wells; for the National Law School Veterans Clinic Consortium by Katie M. Becker and Amy B. Kretkowski; and for the National Veterans Legal Services Program et al. by D. Shayon Ghosh, Liam J. Montgomery, Barton F. Stichman, and Diane Boyd Rauber. 372 BUFKIN v. COLLINS

Opinion of the Court

Justice Thomas delivered the opinion of the Court. When evaluating a veteran's claim for service-related dis- ability benefts, the Department of Veterans Affairs (VA) applies a unique standard of proof known as the “beneft-of- the-doubt rule.” This rule requires the VA to “give the ben- eft of the doubt to the claimant” whenever “there is an ap- proximate balance of positive and negative evidence” on any issue material to the claim. 38 U. S. C. § 5107(b). When re- viewing the VA's benefts decisions, the United States Court of Appeals for Veterans Claims (Veterans Court) must “take due account” of the VA's application of the beneft-of-the- doubt rule. § 7261(b)(1). The question before us is what the Veterans Court must do to comply with that statutory command. We hold that the Veterans Court must review the VA's application of the rule the same way it would any other determination—by reviewing legal issues de novo and factual issues for clear error. See § 7261(a). And, we hold that the VA's determination that the evidence is in approxi- mate balance is a predominantly factual determination reviewed only for clear error.

I A Congress provides veterans a wide range of benefts, in- cluding compensation for disabilities caused or aggravated by active-duty military service. §§ 1110, 1131. The VA, led by the Secretary of Veterans Affairs, administers the laws that provide for those benefts. §§ 301, 303. Veterans who suffer from service-connected post- traumatic stress disorder (PTSD) are among those entitled to disability benefts. To establish service-connected PTSD, the VA requires medical evidence diagnosing the condition and linking the veteran's symptoms with an “in-service stressor,” as well as credible evidence that the in-service stressor occurred. 38 CFR § 3.304(f) (2023). When the VA Cite as: 604 U. S. 369 (2025) 373

fnds those conditions satisfed, it assigns a disability rating that refects reductions in earning capacity. 38 U. S. C. § 1155; 38 CFR § 4.130. This rating determines the amount of compensation a veteran receives. A veteran begins the claims process by submitting a re- quest for benefts to the VA. A VA regional offce then ad- judicates the claim. If the regional offce issues an adverse decision, the veteran may seek de novo review from the Board of Veterans' Appeals. The Board is an administrative body within the VA that renders fnal decisions for the agency. 38 U. S. C. §§ 7101, 7104(a). Congress designed the VA's adjudicatory process to func- tion “ `with a high degree of informality and solicitude for the claimant.' ” Henderson v. Shinseki, 562 U. S. 428, 431 (2011) (quoting Walters v. National Assn. of Radiation Sur- vivors, 473 U. S. 305, 311 (1985)). For example, Congress imposes no statute of limitations on a veteran's claim for benefts. The process at both the regional offces and the Board is ex parte and nonadversarial. 38 CFR §§ 3.103(a), 20.700(c). And, the VA must assist veterans in developing the evidence necessary to substantiate their claims. 38 U. S. C. § 5103A(a)(1). Congress also requires the regional offces and the Board to give veterans the beneft of the doubt on close issues. Specifcally, the VA must “consider all information and lay and medical evidence of record,” and, “[w]hen there is an approximate balance of positive and negative evidence re- garding any issue material to the determination of a matter, the [VA] shall give the beneft of the doubt to the claimant.” § 5107(b). This rule refects a pro-veteran policy choice that dates back to the post-Civil War era. See 50 Fed. Reg. 34454 (1985). A veteran who receives an adverse decision from the Board may appeal to the Veterans Court, an Article I tribu- nal. The Veterans Court has exclusive jurisdiction to re- view fnal decisions of the Board. § 7252(a). 374 BUFKIN v. COLLINS

Section 7261 defnes the scope of the Veterans Court's re- view. Subsection (a), which sets forth the court's adjudica- tory powers, is modeled after the scope-of-review provision in the Administrative Procedure Act (APA). See 5 U. S. C. § 706. It empowers the Veterans Court to decide relevant questions of law; to compel action unlawfully withheld or un- reasonably delayed; and to hold unlawful and set aside VA fndings, conclusions, rules, and regulations that are arbi- trary, capricious, an abuse of discretion, or otherwise not in accordance with law. 38 U. S. C. §§ 7261(a)(1)–(3). Unlike the APA, however, § 7261(a) imposes a slightly different standard of review for factual fndings. The Veterans Court may set aside or reverse a factual fnding only if “the fnding is clearly erroneous,” in contrast to the APA's “substantial evidence” standard. § 7261(a)(4); see Dickinson v. Zurko, 527 U. S. 150, 153, 162–163 (1999) (examining the “subtle” difference between the two standards). Subsection (b) sets forth additional requirements that the Veterans Court must follow “[i]n making the determinations under subsection (a).” § 7261(b). As originally enacted, the provision required only that the Veterans Court “ `take due account of the rule of prejudicial error.' ” Veterans' Ju- dicial Review Act, § 301(a), 102 Stat. 4115 (now codifed at 38 U. S. C. § 7261(b)(2)). We have held that this statutory command requires the Veterans Court to apply the same kind of harmless-error rule that courts ordinarily apply in civil cases. Shinseki v. Sanders, 556 U. S. 396, 406 (2009). In 2002, Congress amended subsection (b) to add the provi- sion at issue here. Veterans Benefts Act, § 401(b), 116 Stat. 2832. Codifed at § 7261(b)(1), it directs the Veterans Court to “ `take due account of the [VA]'s application of section 5107(b) of this title.' ” In other words, when conducting ju- dicial review under subsection (a), the Veterans Court must “take due account” of the VA's application of the beneft-of- the-doubt rule. Cite as: 604 U. S. 369 (2025) 375

In most cases, an appeal to the Veterans Court is a veter- an's fnal opportunity for judicial review of a benefts deci- sion. Although veterans may appeal certain legal issues to the United States Court of Appeals for the Federal Circuit, see § 7292, that court has held that it lacks jurisdiction to review most factual determinations, including whether the VA properly applied the beneft-of-the-doubt rule in individ- ual cases, see Ferguson v. Principi, 273 F. 3d 1072, 1076 (2001). B Petitioners are veterans who applied for service-connected PTSD disability benefts and were dissatisfed with the VA's resolution of their claims. 1 Petitioner Joshua Bufkin served in the Air Force from late 2005 to early 2006. Although Bufkin intended to join the military police, he never received a job assignment because he could not pass the required training classes. Bufkin blamed his poor performance on marital stress. He re- ported that his wife opposed his service and had threatened to commit suicide if he stayed in the military. Bufkin ulti- mately requested and obtained a nonprejudicial hardship discharge. Seven years later, Bufkin sought disability benefts from the VA. He claimed that his military service caused several psychiatric conditions, including PTSD. To support his PTSD claim, Bufkin submitted a letter from a VA physician who had recently treated him. The physician explained that Bufkin's wife suffered from depression, and that Bufkin felt forced to make a choice between divorcing his wife or leaving the military. The physician diagnosed Bufkin with PTSD and opined that the primary stressor was the perceived threat to his wife's life. The VA regional offce denied Bufkin's claim on the ground that the medical evidence was insuffcient to establish a link 376 BUFKIN v. COLLINS

between his symptoms and his military service. Bufkin sought reconsideration, and the VA ordered a second exami- nation by a second VA physician. The new physician deter- mined that Bufkin did not meet the criteria for PTSD. He explained that the frst physician was unable to review Buf- kin's military or medical records, a limitation that may have had a “signifcant impact” on Bufkin's initial diagnosis. App. in No. 2022–1089 (CA Fed.), pp. 26–27. Bufkin also submitted letters from his wife and mother describing changes they observed in his personality after his military service. The VA regional offce reviewed the new information and reaffrmed its denial of Bufkin's claim. Bufkin fled a notice of disagreement. He argued that the evidence for and against his claim was roughly equal and that he was there- fore entitled to “the beneft of the doubt.” § 5107(b). Buf- kin then underwent a third VA examination by yet another physician. Like the second physician, this physician con- cluded that Bufkin's symptoms did not meet the criteria for PTSD. She also noted that Bufkin's symptoms started one year after his discharge, and that no evidence indicated that his military service caused a disability. The VA regional of- fce again reaffrmed its decision to deny benefts. Bufkin appealed to the Board. While the appeal was pending, he submitted a short statement from a fourth physi- cian. This physician stated that, in his opinion, Bufkin suf- fered from PTSD, but he noted other doctors might disagree with that diagnosis. He suggested that, at a minimum, Buf- kin suffered from severe anxiety disorder. The Board concluded that Bufkin did not suffer from PTSD and denied his claim. In a detailed opinion, the Board recounted each physician's fndings and medical conclusions. The Board observed that one physician diagnosed Bufkin with PTSD, two physicians determined he did not have PTSD, and one physician thought the issue was debatable. Cite as: 604 U. S. 369 (2025) 377

After carefully reviewing each medical opinion, the Board explained that it found the second physician's report “espe- cially persuasive.” App. to Pet. for Cert. 61a. The Board noted that the one physician to diagnose Bufkin with PTSD had not reviewed his military or medical records. The Board also reviewed the letters from Bufkin's wife and mother but found that they lacked probative value because neither individual had relevant medical expertise. When taken as a whole, the Board explained, the evidence was not in approximate balance. Accordingly, the Board determined that Bufkin was not entitled to the beneft of the doubt.

2 Petitioner Norman Thornton served in the Army from 1988 to 1991. After being honorably discharged, Thornton applied for and received benefts for an “ `undiagnosed ill- ness.' ” Id., at 32a. Years later, he sought additional bene- fts for service-connected PTSD. The VA granted benefts for the PTSD claim, initially rating his condition as 10% disa- bling and later increasing the rating to 30%. In 2015, Thornton applied for another increase to his PTSD disability rating and underwent a new VA medical ex- amination. The physician concluded that Thornton suffered some common PTSD symptoms, including depressed mood, anxiety, memory loss, and sleep impairment, but questioned whether PTSD was the cause of those symptoms. The phy- sician did not fnd several of the more severe symptoms that earlier medical examinations had revealed, such as suicidal ideation, impaired impulse control, and panic attacks. Nev- ertheless, the regional offce increased Thornton's PTSD dis- ability rating to 50%. Thornton then sought an additional increase to 70% and underwent another medical evaluation, but the regional offce maintained the 50% rating. Thornton appealed to the Board, which reviewed the evi- dence de novo. As in Bufkin's case, the Board thoroughly 378 BUFKIN v. COLLINS

recounted each medical report and assessed the credibility of each physician's fndings. The Board ultimately declined to increase Thornton's 50% disability rating.1

3 Both petitioners appealed the Board's resolution of their PTSD claims to the Veterans Court. They argued that the evidence concerning their diagnoses was in “approximate balance,” so they were entitled to the beneft of the doubt. See § 5107(b). In both cases, the Veterans Court upheld the Board's decisions on the ground that the Board's approximate- balance determinations were not clearly erroneous. Petitioners then appealed to the Federal Circuit, chal- lenging the Veterans Court's legal interpretation of § 7261(b)(1). They argued that the statutory command to “take due account” of the VA's application of the beneft-of- the-doubt rule requires the Veterans Court to review the entire record de novo and decide for itself whether the evi- dence is in approximate balance. The Federal Circuit rejected this argument and affrmed the Veterans Court in both cases. Bufkin v. McDonough, 75 F. 4th 1368 (2023); Thornton v. McDonough, 2023 WL 5091653 (Aug. 9, 2023). The court held that “the statutory command that the Veterans Court `take due account' of the beneft of the doubt rule does not require the Veterans Court to conduct any review of the beneft of the doubt issue be- yond the clear error review required by” § 7261(a). Id., at *2. We granted certiorari to decide what it means to “take due account” of the VA's application of the beneft-of-the- doubt rule. 601 U. S. 1167 (2024). 1 At the same time, the Board granted Thornton a “total disability” rat- ing for his separate claim of unemployability, based on the combined effect of his service-connected PTSD and the undiagnosed illness. App. to Pet. for Cert. 86a–89a. Thornton currently receives full benefts, but those benefts are subject to continual review. He appeals only the denial of the increase to his PTSD disability rating. Cite as: 604 U. S. 369 (2025) 379

II A We start, as always, with the text. Section 7261(b)(1) provides that “[i]n making the determinations under subsec- tion (a),” the Veterans Court “shall” “take due account of the [VA]'s application of [the beneft-of-the-doubt rule].” The plain meaning of the text reveals the scope of the Vet- erans Court's task. It is undisputed that the word “shall” imposes a mandatory command. Shapiro v. McManus, 577 U. S. 39, 43 (2015). “Shall” means “must.” Kingdomware Technologies, Inc. v. United States, 579 U. S. 162, 171–172 (2016). It is likewise undisputed that to “take account of ” something means to give it attention or consideration. See American Heritage Dictionary 1763 (4th ed. 2000) (defning “take account of ” as “[t]o take into consideration”). The word “due” qualifes the attention that the VA must give. “Due” means “appropriate” or “proper.” See 4 Oxford Eng- lish Dictionary 1105 (2d ed. 1989) (“Merited, appropriate: proper, right”); Black's Law Dictionary 515 (7th ed. 1999) (“Just, proper, regular, and reasonable”). Taken together, the statutory command to “take due ac- count” of the VA's application of the beneft-of-the-doubt rule requires the Veterans Court to give appropriate attention to the VA's work, at least when the issue is properly presented. See id., at 516 (defning “due consideration” as “[t]he degree of attention properly paid to something, as the circum- stances merit”).2 The text makes clear that the appropriate attention due is that which is required under subsection (a). Sec- tion 7261(b)(1) states that the Veterans Court must “take

2 Petitioners invite us to decide whether the Veterans Court must ad- dress the VA's application of the beneft-of-the-doubt rule when a veteran does not raise the issue as a ground for reversal. Because both petition- ers adequately preserved and raised their challenges below, we leave that question for another day. 380 BUFKIN v. COLLINS

due account” of the VA's application of the beneft-of-the- doubt rule only “[i]n making the determinations under subsection (a).” This language highlights that taking due account is not a freestanding task but rather an aspect of judicial review under subsection (a). The upshot is straightforward. Review of the VA's beneft-of-the-doubt decision is just another determination made “under subsection (a).” § 7261(b)(1). Accordingly, the standards of review provided in subsection (a) also gov- ern the Veterans Court's review of beneft-of-the-doubt is- sues. In particular, subsection (a) requires the Veterans Court to review the VA's conclusions of law de novo and its fndings of fact for clear error. See §§ 7261(a)(1) and (4). Which standard applies will depend on whether a veteran raises a legal or factual objection. Section 7261(b)(1) makes explicit the Veterans Court's pre- viously implicit duty to review the VA's application of the beneft-of-the-doubt rule, pursuant to the standards set forth in subsection (a). Of course, this duty predated Congress's enactment of § 7261(b)(1); subsection (a) always required the Veterans Court to review all relevant fndings of fact and conclusions of law. See §§ 7261(a)(1) and (4). But, Con- gress's enactment underscores the beneft-of-the-doubt rule's status as a paramount consideration.

B The next question, then, is the appropriate standard of re- view for challenges to the VA's application of the beneft-of- the-doubt rule. The applicable standard will vary with the challenge, as the application of the rule has both factual and legal components. A veteran may challenge any aspect of the VA's application of the beneft-of-the-doubt rule on appeal. No one disputes that challenges to the VA's legal conclu- sions are subject to de novo review. For example, if the veteran argues that the VA misunderstood the defnition of Cite as: 604 U. S. 369 (2025) 381

“approximate balance,” the Veterans Court would construe the challenge as a legal one and review it de novo. So too if the veteran argues that the VA gave the beneft of the doubt to the wrong party. These arguments, however, are not the typical arguments a veteran raises on appeal. In mine-run cases, including the two before us, a veteran challenges the VA's determination that the evidence on a particular material issue is not in ap- proximate balance. Petitioners assert that the approximate-balance determi- nation is a legal inquiry subject to de novo review. The Gov- ernment counters that the determination is a factual fnding, or at least a mixed question that is predominantly factual and reviewed for clear error. After closely examining the way in which the VA conducts the approximate-balance in- quiry, we conclude it is a predominantly factual question and thus subject to clear-error review. The approximate-balance determination involves two steps. First, the VA reviews each item of evidence in the record and assigns weight to it. Both sides agree that this aspect of the VA's analysis is factfnding reviewed only for clear error. Second, the VA assesses the weight of the evi- dence as a whole, deciding whether “there is an approximate balance of positive and negative evidence” on any material issue. § 5107(b). The VA gives the beneft of the doubt to the veteran only after determining that the positive evidence and negative evidence on a material issue are in approximate balance. Otherwise, the VA simply resolves the issue in favor of the party with the more persuasive evidence. The second step—deciding whether the evidence as a whole is approximately balanced—has both legal and factual components. The task is partly factual because it involves marshaling and weighing evidence. See U. S. Bank N. A. v. Village at Lakeridge, LLC, 583 U. S. 387, 396 (2018). The VA must categorize the evidence based on whether it sup- ports or undermines the veteran's claim. After doing so, 382 BUFKIN v. COLLINS

the VA compares the relative strength and persuasiveness of the evidence on each side. Assigning weight to evidence— whether individual pieces of evidence or collections of it—is an inherently factual task. See ibid. But, the VA's decision also involves a legal determination. After assessing the weight of the assorted evidence, the VA must decide whether the evidence is in “approximate balance.” See § 5107(b). In other words, the VA must determine whether the evi- dence satisfies a legal standard. Because the VA's approximate-balance determination involves both legal and factual work, it is at most a mixed question, not a purely legal one.3 The appropriate standard of review for a mixed question depends “on whether answering it entails primarily legal or factual work.” U. S. Bank, 583 U. S., at 396. Some mixed questions require a court to “expound on the law” by “ampli- fying or elaborating on a broad legal standard.” Ibid. When applying the law involves developing legal principles for use in future cases, appellate courts typically review the decision de novo. Ibid. Other mixed questions require courts to resolve “case-specific factual issues. ” Ibid. When the tribunal below is “immerse[d]” in facts and com- pelled to “marshal and weigh evidence” and “make credibil- ity judgments,” the appellate court “should usually review a decision with deference.” Ibid. Reviewing a determination whether record evidence is ap- proximately balanced is “about as factual sounding” as any question gets. Id., at 397. Consider Bufkin's case. To de-

3 We need not decide whether the VA's approximate-balance determina- tion is best characterized as a mixed question or a factual one. See Alex- ander v. South Carolina State Conference of the NAACP, 602 U. S. 1, 18– 19 (2024) (observing that “a fnding of fact” may involve application of a “standard of proof ” with a “substantial legal component”). Whether the determination is mixed or factual, clear-error review is appropriate be- cause it is at least predominantly factual. Cite as: 604 U. S. 369 (2025) 383

cide whether Bufkin was entitled to the beneft of the doubt, the Board reviewed four medical opinions and two letters from Bufkin's family members. After assigning weight to each item, the Board assessed whether the evidence was roughly equal. To make that determination, the Board had to consider evidence of Bufkin's symptoms and their causes, while assessing the credibility of the physicians who treated him. Thornton's case is no different. The Board assessed the weight of his symptoms, the degree to which they af- fected his day-to-day life, and the credibility of each physi- cian. These cases illustrate that the approximate-balance determination necessarily immerses the Board “in case- specifc factual issues.” Id., at 396. When that is so—that is, when the initial decisionmaker is “marshal[ing] and weigh[ing] evidence” and “mak[ing] credibility judgments”— its work is fact intensive, and its determinations should be reviewed with deference. Ibid.

Page Proof Pending C Publication Petitioners and the dissent offer various paths to a differ- ent result. But, their arguments are not persuasive. First, petitioners urge us to hold that with one modest phrase, “take due account,” Congress imposed a new stand- ard of review for challenges to the VA's application of the beneft-of-the-doubt rule. Before the Federal Circuit, peti- tioners argued that § 7261(b)(1) established a de novo stand- ard of review. They hesitate to call the standard “de novo review” before this Court, instead describing it as “APA- plus.” Brief for Petitioners 25, 41–42. However labeled, the directive to “take due account” does not do the work that petitioners envision. Had Congress intended to impose a new standard of review for challenges to the VA's application of the beneft-of-the-doubt rule, it would have clearly named a standard—just as it did in § 7261(a). But, as we have explained, Congress chose not to 384 BUFKIN v. COLLINS

announce any new standard in § 7261(b)(1). It instead incor- porated the pre-existing standards of review set forth in sub- section (a). Petitioners contend, in the alternative, that even if § 7261(b)(1) incorporates the standards of review in § 7261(a), the VA's approximate-balance determination is a legal ques- tion subject to de novo review under § 7261(a). The dissent adopts a similar view, arguing that the approximate-balance determination is a predominantly legal mixed question. Post, at 398–402 (opinion of Jackson, J.). The de novo standard applies, they say, because the approximate-balance determination resembles other legal or predominantly legal questions that courts review de novo. They point to judicial review of a court's probable-cause de- termination as one such analogue. Post, at 401. Probable cause is the standard by which a court assesses whether a police offcer's search or seizure was permissible under the Fourth Amendment. For example, “probable cause to search . . . exist[s] where the known facts and circumstances are suffcient to warrant a man of reasonable prudence in the belief that contraband or evidence of a crime will be found.” Ornelas v. United States, 517 U. S. 690, 696 (1996). We have held that this inquiry is a mixed question that courts review de novo. See id., at 699. Two features of the probable-cause determination distin- guish it from the approximate-balance determination, how- ever, and underscore why courts review it de novo. First, because probable cause is a constitutional standard, we start with a strong presumption that determinations under that standard are subject to de novo review, even if they require courts to “plung[e] into a factual record.” U. S. Bank, 583 U. S., at 396, n. 4. Second, probable cause at bottom poses a question that requires substantial “legal work.” Id., at 398. Because probable cause asks whether the “offcer's un- derstanding of the facts and his understanding of the rele- vant law” was “reasonable,” Heien v. North Carolina, 574 Cite as: 604 U. S. 369 (2025) 385

U. S. 54, 61–64 (2014), it requires an objective, legally grounded inquiry as to what a hypothetical person could have found, see Kentucky v. King, 563 U. S. 452, 464 (2011). The answer to how a hypothetical person would act is, by its nature, one that courts refne over time, building out princi- ples that “acquire content only through application.” Or- nelas, 517 U. S., at 697. De novo review is therefore essen- tial so that courts can ensure “unif[orm] precedent” that will “provid[e] law enforcement offcers with a defned set of rules which, in most instances, makes it possible to reach a correct determination beforehand as to whether an invasion of pri- vacy is justifed.” Ibid. (internal quotation marks omitted). The VA's approximate-balance determination differs on both counts. First, the standard is a creature of statute. Because it does not dwell “[i]n the constitutional realm,” it is not entitled to the presumption of de novo review that applies to probable cause. U. S. Bank, 583 U. S., at 396, n. 4. Second, the approximate-balance inquiry does not have a comparable legal component. Instead, as these cases exem- plify, that inquiry tends to be case specifc and fact intensive, with courts evaluating the weight of the evidence for and against a particular veteran's claim. To decide Bufkin's claim, for example, the VA reviewed a panoply of medical and lay evidence, and Bufkin's central argument before the Veterans Court was that the overall balance of this evidence was closer than the VA thought. Thornton's claim similarly presented case-specifc, fact-bound issues about the existence and severity of symptoms. Resolution of these claims is un- likely to generate guidance for the VA or future courts in the way that probable-cause precedents do for police and the courts. The VA's approximate-balance determination is likewise distinct from a court's conclusion that evidence is suffcient to support a criminal conviction, a conclusion that courts review de novo. See Jackson v. Virginia, 443 U. S. 307, 318–319 (1979). While the suffciency-of-the-evidence inquiry neces- 386 BUFKIN v. COLLINS

sarily turns on facts, the reviewing court does not decide for itself whether the evidence establishes guilt beyond a reasonable doubt. Instead, it construes all evidence and makes all reasonable inferences in favor of the prosecution, and asks whether “any rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt.” Id., at 319. Courts therefore conduct their review through a legal lens, applying a hypothetical, objective standard and putting a thumb on the scale in favor of the prevailing party—the prosecution. No one—petitioners and the dissent included—wants the Veterans Court to do that. Instead, petitioners and the dissent would have the Veterans Court “weigh the evidence” in the frst instance and then make “inferences from basic facts to ultimate facts”— precisely what a court may not do on suffciency review. Ibid. Petitioners and the dissent also rely heavily on the asser- tion that our reading of § 7261(b)(1) renders the provision superfuous. Post, at 395–397. That objection is a serious one. “We are reluctant to treat statutory terms as surplus- age in any setting.” TRW Inc. v. Andrews, 534 U. S. 19, 31 (2001) (internal quotation marks omitted). Our reluctance increases when Congress amends a statute, as we ordinarily presume that when Congress does so, “ `it intends its amend- ment to have real and substantial effect.' ” Intel Corp. v. Advanced Micro Devices, Inc., 542 U. S. 241, 258–259 (2004). But, our reading of § 7261(b)(1) gives that provision a func- tion, even if it might involve some redundancy. As we have explained, Congress's amendment makes express the Veter- ans Court's obligation to review challenges to the VA's appli- cation of the beneft-of-the-doubt rule according to the stand- ards set forth in § 7261(a). Moreover, as the dissent notes, we have left open the possibility that Congress's enactment requires the Veterans Court to review the VA's application of the beneft-of-the-doubt rule even if the veteran fails to present that issue on appeal. See n. 2, supra; post, at 396. Cite as: 604 U. S. 369 (2025) 387

If the Court were to decide that Congress's enactment changed the party-presentation requirement, then the enact- ment of § 7261(b)(1) would serve an additional independent function. Notably, the dissent fails to offer a reading of the text that avoids redundancy. According to the dissent, the VA's approximate-balance determination is a predominantly legal question subject to de novo review. Post, at 399–402. But, if the dissent were correct, then the approximate-balance determination would have been subject to de novo review even before the enactment of § 7261(b)(1). All agree that under § 7261(a), which predated Congress's enactment of § 7261(b)(1), the Veterans Court is required to review ques- tions of law de novo. Thus, under the dissent's interpre- tation—no less than ours—the enactment of § 7261(b)(1) merely underscores a pre-existing duty. The canon against surplusage can be meaningful when a competing interpreta- tion would avoid superfuity. Microsoft Corp. v. i4i L. P., 564 U. S. 91, 106 (2011). But, when both interpretations in- volve the same redundancy, the canon against surplusage simply does not apply. See Marx v. General Revenue Corp., 568 U. S. 371, 385 (2013) (declining to apply the canon against surplusage when no interpretation would give effect to every word). In any event, “[s]ometimes the better overall reading of the statute contains some redundancy.” Rimini Street, Inc. v. Oracle USA, Inc., 586 U. S. 334, 346 (2019). Here, the text leaves no doubt that the Federal Circuit's reading of the statute is the better one.

* * * Section 7261(b)(1) does not establish a new standard of re- view for challenges to the VA's application of the beneft-of- the-doubt rule. Instead, it requires the Veterans Court to apply the appropriate standard of review under § 7261(a). Because the VA's approximate-balance determination is a 388 BUFKIN v. COLLINS

Jackson, J., dissenting

predominantly factual question, the Veterans Court reviews it for clear error. § 7261(a)(4).

III The judgment of the Court of Appeals is affrmed.

It is so ordered. Justice Jackson, with whom Justice Gorsuch joins, dissenting. The Court draws two conclusions from today's evaluation of Congress's “take due account” admonition. See 38 U. S. C. § 7261(b)(1). First, it holds that when the United States Court of Appeals for Veterans Claims (Veterans Court) reviews the Department of Veterans Affairs' (VA's) application of the “beneft-of-the-doubt” rule, that appellate tribunal must use the “same” standards of review that apply to its assessment of any other VA claims determination. Ante, at 372. Second, the Court concludes that whether evi- dence is in “approximate balance” for purposes of the beneft- of-the-doubt rule is a “predominantly factual determination” to be reviewed only for clear error. Ibid. The majority is wrong in both respects. Nothing about the text, context, or drafting history of subsection (b)(1) demonstrates that “take due account” actually means “pro- ceed as normal.” Reading the provision in that fashion, as the majority does, makes little sense. That interpretation is also inconsistent with how we have treated identical language elsewhere in this same statute and renders meaningless the “take due account” command that Congress specifcally amended § 7261(b) to insert. The majority's clear-error conclusion fares no better, inso- far as its reasoning ignores what appellate courts do and what we have consistently said about substantially similar circumstances. That is, even if the majority were correct Cite as: 604 U. S. 369 (2025) 389

that subsection (b)(1) instructs the Veterans Court to carry on applying the usual standards, clear-error review would not be appropriate because whether the VA properly applied the beneft-of-the-doubt rule does not present a question of fact. The VA's beneft-of-the-doubt determination poses, at most, a mixed question of law and fact—and one that is sub- stantively indistinguishable from the kinds of mixed ques- tions that this Court has long said are subject to de novo review on appeal. In short, the Court today concludes that Congress meant nothing when it inserted subsection (b)(1) in response to con- cerns that the Veterans Court was improperly rubberstamp- ing the VA's beneft-of-the-doubt determinations, and also that the Veterans Court is not obliged to do anything more than defer to those agency decisions notwithstanding Con- gress's “take due account” direction. I respectfully dissent.

I Under the time-honored beneft-of-the-doubt rule, veter- ans asserting claims for service-connected disabilities are entitled to have any reasonable doubt on a material issue resolved in their favor. Congress codifed this rule in 1988: Section 5107(b) of Title 38 states unequivocally that the VA must “give the beneft of the doubt to the claimant” when- ever “there is an approximate balance of positive and nega- tive evidence regarding any issue material to the” veteran's benefts claim. This generous standard of proof honors the sacrifces of those who have served in the Armed Forces. Recognizing the weighty interests at stake in cases that involve benefts for veterans, Congress's 1988 legislation also established judicial review of the VA's benefts decisions by the Veterans Court, a specialized Article I tribunal. Section 7261(a) lays out in detail the authority that the Veterans Court has been given. As relevant here, the Veterans Court is authorized to (1) “decide all relevant questions of 390 BUFKIN v. COLLINS

law”; (2) “compel action of the Secretary unlawfully withheld or unreasonably delayed”; and (3) “hold unlawful and set aside” certain VA decisions. §§ 7261(a)(1)–(3). Notably, clause (4) of § 7261(a) states that “in the case of a fnding of material fact adverse to the claimant made in reaching a decision in a case before the [VA] with respect to benefts,” the Veterans Court shall “hold unlawful and set aside or reverse such fnding if the fnding is clearly erroneous.” § 7261(a)(4). For present purposes, all parties agree that § 7261(a) has always authorized the Veterans Court to review the VA's application of the beneft-of-the-doubt rule. But judicial re- view of that issue proved illusory from the start, because the Veterans Court historically applied a deferential standard of review for assessing the VA's beneft-of-the-doubt determi- nations. See, e. g., Gilbert v. Derwinski, 1 Vet. App. 49, 57– 58 (1990); Wensch v. Principi, 15 Vet. App. 362, 367 (Ct. App. Vet. Cl. 2001). As a result, veterans groups routinely com- plained to Congress that judicial review by the Veterans Court was inappropriately deferential. See, e. g., Hearing before the Senate Committee on Veterans' Affairs, 107th Cong., 2d Sess., 47 (2002) (statement of the legislative direc- tor of a veterans group); see also Brief for Federal Circuit Bar Association as Amicus Curiae 11–14. Congress responded in 2002, when it amended § 7261(b) to specifcally require the Veterans Court to “take due account of ” the VA's application of the beneft-of-the-doubt rule. § 7261(b)(1). In its entirety, the amended subsection (b) states: “(b) In making the determinations under subsection (a), the [Veterans] Court shall review the record of pro- ceedings before the Secretary and the Board of Veter- ans' Appeals . . . and shall— “(1) take due account of the Secretary's application of section 5107(b) of this title; and Cite as: 604 U. S. 369 (2025) 391

“(2) take due account of the rule of pre judicial error.” § 7261(b).1 The plain text of § 7261(b)(1) thus requires the Veterans Court to “take due account” of the VA's obligations under the beneft-of-the-doubt rule when it reviews VA benefts de- terminations under § 7261(a).

II The dispute before us concerns the meaning of subsection (b)(1)'s “take due account” clause. The majority and I agree that subsection (b)(1) imposes a “statutory command” that the Veterans Court “give appropriate attention to the VA's work” related to its application of the beneft-of-the-doubt rule. Ante, at 379. But the majority concludes that the “appropriate attention due is that which is required under subsection (a).” Ibid. In my view, that holding is un- moored from subsection (b)(1)'s text, ignores § 7261's overall structure and drafting history, and renders entirely super- fuous Congress's requirement that the Veterans Court “take due account” of the application of the beneft-of-the-doubt rule. The better reading of the statute recognizes that sub- section (b)(1) requires the Veterans Court to determine, without deference, whether the VA properly applied the beneft-of-the-doubt rule. A The primary textual pillar of the majority's conclusion that “[r]eview of the VA's beneft-of-the-doubt decision is just an- 1 Subsection (b)(1) cross references § 5107(b), which (as explained above and bears repeating) codifes the beneft-of-the-doubt rule: “When there is an approximate balance of positive and negative evidence regarding any issue material to the determination of a matter, the Secretary shall give the beneft of the doubt to the claimant.” § 5107(b). The reference to “the Secretary” in § 5107(b) and § 7261(b)(1) includes the VA's regional offces and the Board of Veterans' Appeals. See ante, at 373. 392 BUFKIN v. COLLINS

other determination made `under subsection (a)' ” is its ob- servation that subsection (b)(1)'s review applies only “ `[i]n making the determinations under subsection (a).' ” Ante, at 380 (quoting § 7261(b)(1)). Based on that prefatory lan- guage, the majority reasons that subsection (b)(1) is merely an “aspect of ” the review that the Veterans Court performs under subsection (a), such that the “standards of review pro- vided in subsection (a) also govern the Veterans Court's re- view of beneft-of-the-doubt issues.” Ante, at 380. But the fact that the Veterans Court must apply subsec- tion (b)(1) “[i]n making the determinations under subsection (a)” does not justify collapsing these two provisions. The word “in,” when paired with a gerund—here, “making”—is generally “equivalent in sense to a temporal clause intro- duced by when, while, if, [or] in the event of.” 7 Oxford English Dictionary 760 (2d ed. 1989) (emphasis in original). The “in making” clause in subsection (b) thus merely pro- vides that when the Veterans Court makes one of the deter- minations authorized by subsection (a), it must also satisfy its duty to “take due account of ” the VA's application of the beneft-of-the-doubt rule. In other words, the “in making” clause the majority seizes upon to justify its same-standards holding simply establishes that review under subsection (b)(1) occurs alongside review under subsection (a), not that they are the same thing. Interpreting subsection (b)(1) to require a separate, con- current review of the VA's compliance with the beneft- of-the-doubt rule is consistent with how this Court has in- terpreted subsection (b)(1)'s parallel provision—subsection (b)(2). Subsection (b)(2) directs that, “[i]n making the deter- minations under subsection (a),” the Veterans Court shall “ take due account of the rule of pre judicial error. ” § 7261(b)(2). In Shinseki v. Sanders, 556 U. S. 396 (2009), we held that subsection (b)(2) “requires the Veterans Court to apply the same kind of `harmless-error' rule that courts ordinarily apply in civil cases,” id., at 406. We did not Cite as: 604 U. S. 369 (2025) 393

suggest—much less hold—that subsection (a)'s standards of review governed subsection (b)(2)'s “take due account” clause. To the contrary, we explained that the Veterans Court is not prevented from “directly asking the harmless- error question” and “resting its conclusion on the facts and circumstances of the particular case.” Id., at 408. Thus, there is no dispute that, under subsection (b)(2), the Veterans Court analyzes whether any error was harmless de novo, without applying the standards of review prescribed by subsection (a). And because subsection (a)'s standards do not apply when the Veterans Court “take[s] due account of the rule of prejudicial error” under subsection (b)(2), it is oddly discordant for the majority to conclude that those standards do apply when the Veterans Court “take[s] due account of the Secretary's application of ” the beneft-of-the- doubt rule. See FCC v. AT&T Inc., 562 U. S. 397, 408 (2011) (“ `[I]dentical words and phrases within the same statute should normally be given the same meaning' ”). An understanding of how subsection (b)(2)'s “take due ac- count” language functions in practice further clarifes subsec- tion (b)(1), as subsection (b)(2) existed prior to subsection (b)(1) and plainly served as its model. The majority does not dispute that the VA's appellate tribunal, the Board of Veterans' Appeals, sometimes applies the rule of prejudicial error in the frst instance. See, e. g., Medrano v. Nicholson, 21 Vet. App. 165, 170 (Ct. App. Vet. Cl. 2007).2 If the Board determines that an error was harmless and denies a veteran's claim, then 2 This happens often in the context of notice errors. Consistent with the pro-claimant nature of the veterans-disability system, the VA has a statutory obligation to notify veterans of certain information. Shinseki v. Sanders, 556 U. S. 396, 400 (2009) (citing 38 U. S. C. § 5103(a)). If the VA breaches that duty, the Board may nonetheless deny a veteran's claim if it determines that the notice error was harmless. The Veterans Court reviews any such harmlessness determination de novo. See, e. g., Me- drano, 21 Vet. App., at 170, 173; Lowery v. McDonough, 2024 WL 3949253, *4, *5–*6 (Ct. App. Vet. Cl., Aug. 27, 2024) (unpublished); Weaver v. Shinseki, 2011 WL 2745773, *1 (Ct. App. Vet. Cl., July 13, 2011) (unpublished). 394 BUFKIN v. COLLINS

per subsection (b)(2), the Veterans Court reviews the record for prejudicial error and does so “de novo, in other words, without any deference to the Board,” before affrming. Id., at 171. Subsection (b)(1) works in precisely the same way: If the Veterans Court is inclined to affrm the VA's denial of benefts, subsection (b)(1) requires it to frst confrm de novo that the beneft-of-the-doubt rule was properly applied.

B Nor does the drafting history of the relevant provisions support the majority's contention that subsection (b)(1) is an “aspect” of subsection (a), or that subsection (a)(4) substan- tively limits the review that subsection (b)(1) requires. In fact, the historical account suggests the opposite. The ini- tial Senate bill amended subsection (a)(4)—rather than sub- section (b)—to command that the Veterans Court “tak[e] into account” the VA's application of the beneft-of-the-doubt rule. S. Rep. No. 107–234, p. 40 (2002) (italics deleted). That bill also proposed two other relevant changes: It authorized the Veterans Court to “reverse” the VA's factual fndings rather than remand them, and it eliminated the “clearly erroneous” standard of review. Ibid. (italics deleted). Accordingly, under the Senate's bill, subsection (a)(4) would have author- ized the Veterans Court to “set aside or reverse” any factual fnding of the VA that was “unsupported by substantial evi- dence of record, taking into account the Secretary's applica- tion of ” the beneft-of-the-doubt rule. Ibid. (italics deleted). Had Congress adopted this proposal, the majority's con- fation of subsections (a) and (b)(1) might stand on frmer ground. But Congress plainly rejected that approach. It instead amended subsection (a)(4) only to empower the Vet- erans Court to reverse certain factual fndings (i. e., those that are material and adverse to the claimant), and neither modifed subsection (a)(4)'s clear-error standard nor inserted a requirement that the Veterans Court review the VA's ap- plication of the beneft-of-the-doubt rule into that provision. Cite as: 604 U. S. 369 (2025) 395

§ 401, 116 Stat. 2832. At the end of the day, then, instead of incorporating a beneft-of-the-doubt-rule reminder into sub- section (a)(4), Congress crafted an entirely new statutory provision—subsection (b)(1)—to address judicial review of the VA's application of the beneft-of-the-doubt rule. Ibid. Recognizing that review under subsection (b)(1) is distinct from review under subsection (a)(4) respects Congress's choice to separate those two provisions. It also comports with subsection (b)(1)'s language and the statute's overall design. No one disputes that Congress sought to subject the VA's beneft-of-the-doubt decisions to increased judicial scrutiny while otherwise preserving the VA's not-clearly- erroneous factual fndings. Reading subsection (b)(1) to be such a congressional mandate furthers that objective, while the majority's conclusion that subsection (b)(1) implicitly in- corporates subsection (a)(4)'s limited clear-error standard undermines it. C The usual indicators of statutory meaning thus confrm that subsection (b)(1) was meant to have bite. But the ma- jority's reading renders it toothless. According to the ma- jority, subsection (b)(1) does not impose any new obligation on the Veterans Court; instead, that provision merely “un- derscores” the importance of § 5107(b)'s beneft-of-the-doubt rule. Ante, at 380. At oral argument, the Government put the matter even more bluntly, arguing that subsection (b)(1) is an “exclamation point” that does “duplicate work” and is ultimately “redundant.” Tr. of Oral Arg. 42, 44. This interpretation, which the majority adopts, violates the “cardinal principle of statutory construction” that courts or- dinarily “ ` “give effect, if possible, to every clause and word of a statute.” ' ” Williams v. Taylor, 529 U. S. 362, 404 (2000) (quoting United States v. Menasche, 348 U. S. 528, 538–539 (1955)). The majority acknowledges that the surplusage argument is “a serious one” and that its interpretation “might involve 396 BUFKIN v. COLLINS

some redundancy.” Ante, at 386. Still, the majority main- tains that subsection (b)(1) has a “function” even if it per- forms zero independent work. Ibid. That is because, as the majority sees it, subsection (b)(1) makes “express” the Veterans Court's duty to hear challenges to the VA's applica- tion of the beneft-of-the-doubt rule under subsection (a), and may even require the Veterans Court to do so sua sponte. Ibid. But the majority cannot overcome the surplusage problem simply by asserting that subsection (b)(1) empha- sizes or clarifes a pre-existing duty. To state the obvious, “most superfuous language” can be justifed as providing “clarity.” NLRB v. SW General, Inc., 580 U. S. 288, 304 (2017). Perhaps for this reason, the majority does not iden- tify any other case in which this Court held that Congress passed an amendment with no substantive effect. Moreover, the majority does not and cannot explain why any such expression of the Veterans Court's duties was needed—much less why Congress would go out of its way to amend the statute to underscore an existing obligation. Subsection (a) covers the waterfront of possible exercises of judicial authority and has always required the Veter- ans Court to consider any argument “presented” by a vet- eran. Subsection (a) cannot be reasonably read to exclude consideration of a veteran's claim that the VA misapplied the beneft-of-the-doubt rule. And no one actually thought this; indeed, prior to Congress's enactment of subsection (b)(1), the Veterans Court regularly considered challenges to the VA's application of the beneft-of-the-doubt rule—it just did so deferentially. See supra, at 390. Thus, the problem that Congress enacted subsection (b)(1) to address was not that the Veterans Court seemed to have limited authority under subsection (a). Nor was it that the Veterans Court was somehow unaware of its duty to review the VA's application of the beneft-of-the doubt rule upon re- quest. Rather, as I have explained, Congress enacted sub- section (b)(1) after veterans service organizations brought to Cite as: 604 U. S. 369 (2025) 397

its attention how the Veterans Court was performing that review—i. e., it was giving too much deference to the VA's beneft-of-the-doubt determinations. Ibid. Congress's in- disputable focus was on the improper degree of deference the Veterans Court was affording to the VA's determinations, a target that the majority's reading misses completely. Nor is the majority's surplusage problem solved by the “possibil- ity” that subsection (b)(1) requires sua sponte review of the VA's beneft-of-the-doubt determinations. Ante, at 386. Even assuming that subsection (b)(1) seeks to effect that change, sua sponte review would make no difference unless the Veterans Court proceeds nondeferentially. A rubber- stamp applied to every case remains a rubberstamp. Finally, the majority maintains that if Congress had in- tended for subsection (b)(1) to do anything other than under- score the pre-existing duty to consider challenges to the VA's beneft-of-the-doubt determinations, it would have more clearly said so. Ante, at 383–384. That gets things exactly backwards. Ordinarily, when Congress amends a statute, “ ` “we presume it intends its amendment to have real and substantial effect.” ' ” Husky Int'l Electronics, Inc. v. Ritz, 578 U. S. 355, 359 (2016) (quoting United States v. Quality Stores, Inc., 572 U. S. 141, 148 (2014)). The majority instead presumes that Congress intended to enact a do-nothing amendment. But if Congress had wanted to maintain the status quo, why enact subsection (b)(1) at all? The pre- sumption is, and should be, that Congress expects a change when it goes through the trouble of amending a statute. And, here, the majority's reasoning has plainly failed to over- come that presumption. D The veterans canon resolves whatever lingering doubt might remain about the proper interpretation of subsection (b)(1). We have “long applied `the canon that provisions for benefts to members of the Armed Services are to be con- strued in the benefciaries' favor.' ” Henderson v. Shinseki, 398 BUFKIN v. COLLINS

562 U. S. 428, 441 (2011) (quoting King v. St. Vincent's Hospi- tal, 502 U. S. 215, 220–221, n. 9 (1991)). Likewise, under the veterans canon, “interpretive doubt is to be resolved in the veteran's favor.” Brown v. Gardner, 513 U. S. 115, 118 (1994). The veterans canon strongly supports interpreting subsec- tion (b)(1) to represent more than Congress's grandiloquence. Requiring subsection (b)(1) to affect the judicial review the Veterans Court provides aligns with the signifcance of the beneft-of-the-doubt rule and mitigates the unique problems that arise in veterans' cases, including that veterans are “often unrepresented” before the VA. Sanders, 556 U. S., at 412. Veterans' claims also often involve a considerable time lag that produces problems of proof, since no statute of limi- tations applies to such claims. D. Nagin, The Credibility Trap: Notes on a VA Evidentiary Standard, 45 U. Memphis L. Rev. 887, 894, 898 (2015). An appeal to the Veterans Court is generally a veteran's “fnal opportunity” to correct the VA's mistakes. Ante, at 375. Nondeferential review of the VA's beneft-of-the-doubt determinations plainly mini- mizes the risk that veterans with borderline claims will be denied benefts to which they are entitled. Therefore, the veterans canon “garnish[es] an already solid argument” based on the statute's text and structure. Arellano v. McDonough, 598 U. S. 1, 14 (2023). Reading subsection (b)(1) as a standalone command that requires the Veterans Court to review the VA's application of the beneft- of-the-doubt rule nondeferentially obviously favors veterans more than an interpretation that reduces the provision to an exclamation point. III The majority does not stop at the mistaken conclusion that Congress enacted subsection (b)(1) for no reason other than to reiterate the Veterans Court's pre-existing statutory du- ties. It also proceeds to analyze how a challenge to the VA's application of the beneft-of-the-doubt rule fares when fl- Cite as: 604 U. S. 369 (2025) 399

tered through the authority conferred to the Veterans Court in subsection (a). In this regard, the majority reasons that part—but not all—of the VA's application of the beneft-of- the-doubt rule must be reviewed for clear error. Ante, at 380. But in my view, the Court mischaracterizes the “ap- proximate balance” assessment under § 5107(b), which is the part of the analysis that is at issue here. The majority maintains that that aspect of the beneft-of-the-doubt rule involves the type of predominately factual fnding that appel- late courts ordinarily review with deference. Ante, at 382– 383. Stated simply, such is not the case.

A To understand the majority's mistake, one must frst be clear eyed about the two steps that are necessary for the Veterans Court to review the VA's application of the beneft- of-the-doubt rule. First, the Veterans Court identifes and reviews the VA's relevant factual fndings concerning the evi- dence presented. Then, the Veterans Court determines whether, based on the VA's plausible factual fndings, the “positive and negative evidence” is in “approximate balance” (what I call the approximate-balance standard). All agree that the clear-error standard applies to the VA's factual fndings regarding the evidence that the Veterans Court examines at step one. So, for example, the Veterans Court must accept the VA's determination that a particular piece of evidence is reliable or that an individual medical examiner is qualifed, unless those fndings are clearly er- roneous. It is at the second step—when the Veterans Court determines whether the evidence is in approximate balance—that questions of law come in. At step two, the Veterans Court is essentially being asked to decide whether a legal standard (“approximate balance”) has been satisfed on the established facts. As a general matter, this Court has long treated the application of a legal standard to a given set of facts as an exercise that poses a 400 BUFKIN v. COLLINS

legal question. See, e. g., Guerrero-Lasprilla v. Barr, 589 U. S. 221, 227–228 (2020). What the majority misses in its effort to shoehorn § 5107(b)'s approximate-balance inquiry into the question-of-fact category is the true nature of that assessment, as well as the fact that making the approximate- balance determination is more nuanced than simply weighing the evidence on hand. For one thing, the court must have a standard for deciding when the evidence before it is in “approximate balance” as a matter of law. In our common-law legal system, devising that standard happens over time, on a case-by-case basis. Therefore, applying it might well involve examining Veter- ans Court precedents concerning the meaning of “approxi- mate balance,” as well as past cases that establish when and under what circumstances certain kinds of evidence will be found to qualify. One can easily imagine a body of case law developing to illuminate the substantive and procedural con- tours of the approximate-balance inquiry and the factors that are to be considered. Thus, applying § 5107(b)'s “approxi- mate balance” test is in no way akin to “fnding” a “material fact” that is ultimately subject to subsection (a)(4)'s clear- error review, as even the Government appeared to concede at oral argument. See Tr. of Oral Arg. 59. Rather, applica- tion of the “approximate balance” test presents a “classic mixed question of law” that requires applying a “legal stand- ard” to “historical facts.” Ibid. Where Congress has not prescribed a standard of review for evaluating a mixed question, the appropriate degree of deference refects “the nature of the mixed question” and “which kind of court . . . is better suited to resolve it.” U. S. Bank N. A. v. Village at Lakeridge, LLC, 583 U. S. 387, 395 (2018) (citing Miller v. Fenton, 474 U. S. 104, 114 (1985)). In general, trial courts' “expertise” as factfnders justifes def- erence to their factual determinations. Anderson v. Besse- mer City, 470 U. S. 564, 574–575 (1985). For this reason, appellate courts “usually” review with deference mixed Cite as: 604 U. S. 369 (2025) 401

questions that require resolving “case-specifc factual is- sues.” U. S. Bank, 583 U. S., at 396. But, as I have already explained, that is not what is happening here. Even if the majority is correct to assume that § 5107(b)'s “approximate balance” standard calls for a mere head-to- head weighing of the positive and negative evidence, that is still an insuffcient basis upon which to rest the majority's clear-error conclusion, because deference is “not always” given regarding such fact-specifc inquiries. Ibid., n. 4. There are circumstances where de novo review is appro- priate “even when answering a mixed question primarily in- volves plunging into a factual record.” Ibid. Consider one common example. Whether probable cause existed to support a search or seizure in the Fourth Amend- ment context is a mixed question of law and fact that appel- late courts review de novo. Ornelas v. United States, 517 U. S. 690, 696–699 (1996). And review of probable-cause de- terminations proceeds in two steps, just like the approximate-balance inquiry. At step one, the appellate court identifes the relevant historical facts and reviews them for clear error. Id., at 696–697, 699. At step two, the court decides whether an objectively reasonable police offcer would believe that those facts establish probable cause to search. Id., at 696–697. The second step is a legal question that asks whether the facts satisfy a legal standard. Ibid. Another highly fact-sensitive inquiry is whether evidence presented at trial is suffcient to support a verdict—and ap- pellate courts review that determination de novo as well. In a civil case, the trial court may grant judgment to a moving party as a matter of law after the nonmoving “party has been fully heard on an issue during a jury trial,” if it “fnds that a reasonable jury would not have a legally suffcient evidentiary basis to fnd for the party on that issue.” Fed. Rule Civ. Proc. 50(a)(1). When making this decision, the trial judge determines whether, reviewing the facts in the light most favorable to the nonmovant, the evidence is suff- 402 BUFKIN v. COLLINS

cient to meet the applicable evidentiary standard. Reeves v. Sanderson Plumbing Products, Inc., 530 U. S. 133, 150 (2000); see also 9B C. Wright & A. Miller, Federal Practice and Procedure § 2524 (3d ed. 2008) (explaining that the suffciency-of-the-evidence inquiry is a legal question deter- mined by the court). And, on appeal, the reviewing court undertakes the same suffciency analysis by evaluating the evidence without any deference to the trial court's determi- nation. Ibid.; see also Reeves, 530 U. S., at 150.3 These examples demonstrate that nondeferential appellate review of factbound legal issues is not unprecedented. And under § 7261(b)(1), the Veterans Court must follow suit. Section 7261(b)(1) directs the Veterans Court to evaluate the VA's approximate-balance decisions by frst accepting the agency's not-clearly-erroneous factual fndings about the evi- dence presented and then determining whether that evi- dence satisfes the approximate-balance standard, without deference to the VA's own conclusions. This analysis— which “has both legal and factual components,” ante, at 381—is plainly analogous to the suffciency-of-evidence and probable-cause questions that appellate courts regularly re- view de novo. B The majority dismisses these compelling parallels on the grounds that assessing the suffciency of the evidence or deter- mining the existence of probable cause are more “legal” than the approximate-balance inquiry. Ante, at 384–385. The majority further emphasizes that evaluating whether the ev- idence is approximately balanced for § 5107(b) purposes is a “fact intensive” inquiry that is better left to the “tribunal below” than the appeals court. Ante, at 382–383. It also unfurls a surplusage argument of its own: If de novo review 3 In criminal cases, too, whether evidence is legally suffcient to support a conviction is reviewed de novo, considering the evidence in the light most favorable to the prosecution. Jackson v. Virginia, 443 U. S. 307, 318–319 (1979). Cite as: 604 U. S. 369 (2025) 403

applies to the VA's approximate-balance determination, the majority says, then subsection (b)(1) is superfuous under petitioners' interpretation because that determination “would have been subject to de novo review even before the enactment of § 7261(b)(1).” Ante, at 387. None of this reasoning is persuasive. To start, the major- ity's efforts to distinguish suffciency and probable-cause re- view from the approximate-balance inquiry fall fat. For ex- ample, according to the majority, the probable-cause inquiry is distinguishable because reviewing courts assess probable cause from the standpoint of an objectively reasonable off- cer, and courts are required to consider and “refne over time” what a “hypothetical person” would deem reasonable as a matter of law. Ante, at 384–385. The majority says that suffciency challenges likewise require courts to use a “legal lens, applying a hypothetical, objective standard”— i. e., what a reasonable jury could fnd. Ante, at 386. But the approximate-balance inquiry is not meaningfully different. Before the Veterans Court can determine whether the VA erred in deciding that the positive and nega- tive evidence is not in “approximate balance,” it must frst have an understanding of what “balanced” evidence looks like in this context. Identifying “balance” might involve a hypothetical and objective examination of the evidence—just like a review of the suffciency of the evidence or probable cause. Imagine two doctors testify in favor of a claimant's diagnosis and two doctors testify against it. Does the mere fact that the same number of witnesses are presented on both sides mean that the evidence is “balanced,” as a legal matter? The Veterans Court would need to develop a legal rule to decide. It would also need to decide how the quality of the evidence factors in: What if one of the doctors is the Nobel laureate who discovered the condition at issue while others are recent medical-school graduates? Does the testi- mony of a doctor who has examined the claimant balance evenly against one who has not? 404 BUFKIN v. COLLINS

It is thus quite likely that legal standards will need to be developed to govern the assessment of “approximate balance”—which is why the majority is wrong to contend that resolving approximate-balance challenges will rarely produce generally applicable precedents. Ante, at 385–386. Regardless, the mere fact that the assignment here involves a fact-intensive and “multi-faceted” decision—such that “ `one determination will seldom be a useful “precedent” for another' ”—should not impede our recognition of the signif- cant questions of law that are embedded in the application of the approximate-balance standard. Ornelas, 517 U. S., at 698; see also Dupree v. Younger, 598 U. S. 729, 734 (2023) (observing that suffciency challenges “depend on . . . the facts”). The majority's attempt to distinguish probable cause from the approximate-balance standard because the latter is “a creature of statute,” ante, at 385, is similarly off base. To be sure, application of the beneft-of-the-doubt rule is not constitutional in nature. But this principle is not a mere statutory creation, either—it predates § 5107(b) and in fact “dates back to the post-Civil War era.” Ante, at 373 (citing 50 Fed. Reg. 34454 (1985)); see also, e. g., H. R. Rep. No. 387, 46th Cong., 3d Sess., 132–133 (1881) (testimony of a Bureau of Pensions physician explaining that when the evi- dence left “room for doubt” regarding the extent of a veter- an's disability, the practice was to “give him the beneft of the doubt” if “he appear[ed] like an honest man”). In any event, the majority does not explain why the statutory- versus-constitutional distinction has any rational bearing on the amount of deference the Veterans Court owes to the VA's beneft-of-the-doubt determinations. Speaking of statutes, it is also signifcant that, under the judicial review scheme Congress has crafted for veterans' benefts claims, the Veterans Court is no ordinary appellate tribunal. The thrust of the majority's reasoning seems to be that, as an “appellate court,” the Veterans Court should Cite as: 604 U. S. 369 (2025) 405

do what appeals courts typically do: defer to the lower tribu- nal's assessment of these “ `case-specifc factual issues.' ” Ante, at 383. But the “contrast between ordinary civil liti- gation” and the statutory requirements for adjudicating veterans' benefts claims “could hardly be more dramatic.” Henderson, 562 U. S., at 440. To begin, the institutional advantages that generally war- rant deference to trial courts on factual issues do not exist in this context, because the fnal decisionmaker for the VA— the Board of Veterans' Appeals—is itself an appellate body. See id., at 431 (citing §§ 7101, 7104(a)). What is more, the Veterans Court is “an Article I tribunal” that Congress placed within “a unique administrative scheme.” Id., at 437–438. Such tribunals often develop “special `expertise' ” that guides them “ `in making complex determinations in a specialized area of the law.' ” Sanders, 556 U. S., at 412. And that is particularly true of the Veterans Court, which regularly reviews “case-specifc raw material” in veterans' cases. Ibid. This feature distinguishes the Veterans Court from Article III appellate courts because it gives that court the “experience” and “expertise,” Anderson, 470 U. S., at 574, that “enable it to make empirically based” judgments on fact-specifc issues, Sanders, 556 U. S., at 412—such as whether the beneft-of-the-doubt rule applies. Considering these unique features of the Veterans Court and the legal landscape in which it operates, there is good reason to believe that the Veterans Court is actually “better suited to resolve” the approximate-balance inquiry than the VA itself. U. S. Bank, 583 U. S., at 395. This is especially so given the need for fairness and consistency in the applica- tion of the “approximate balance” standard. Thus, even though review of the VA's approximate-balance determination requires applying a legal standard to a set of facts, the Veterans Court is well equipped to do so, and de novo review of the VA's determination best promotes the “sound administration of justice.” Miller, 474 U. S., at 114. 406 BUFKIN v. COLLINS

In concluding otherwise, the majority minimizes the “singu- lar characteristics” of the review scheme at issue, Hender- son, 562 U. S., at 440, and disregards Congress's choice to task the Veterans Court with ensuring that the VA complies with the beneft-of-the-doubt rule. Finally, because § 7261(a) unquestionably authorizes de novo review of questions of law, the majority contends that applying de novo review to the VA's approximate-balance determinations would render subsection (b)(1) superfuous. Ante, at 386. But that reasoning assumes that it was always clear that an approximate-balance challenge raises a legal question (an assumption belied by the very holding that the majority announces today). Quite to the contrary, before Congress enacted subsection (b)(1) to require that the Veter- ans Court “take due account” of the VA's beneft-of-the-doubt determinations, that court—erroneously, in my view—re- viewed beneft-of-the-doubt challenges with deference. See supra, at 390–391, 396–397. Properly understood, subsection (b)(1) accomplishes the clear and important mission of abro- gating the Federal Circuit and Veterans Court cases requir- ing such deference. The majority simply ignores this point. See ante, at 387. * * * In response to complaints that the Veterans Court was not doing enough to ensure that the VA was, in fact, giving the beneft of the doubt to veterans, Congress inserted into a demonstrably pro-claimant administrative scheme the requirement that the Veterans Court “take due account” of the VA's application of the beneft-of-the-doubt rule. § 7261(b)(1). The reading that the majority adopts today re- duces that provision to a rhetorical fourish and all but en- sures that the Veterans Court will continue rubberstamping the VA's application of the beneft-of-the-doubt rule. If the majority's holding is correct, then “Congress went through an awful lot to achieve relatively little.” Wilkinson v. Garland, 601 U. S. 209, 227 (2024) (Jackson, J., concurring Cite as: 604 U. S. 369 (2025) 407

in judgment). The far better reading of this statute—one that fully comports with the text, structure, context, history, and purpose of the provision Congress wrote—is that the Veterans Court must review without deference the VA's approximate-balance determination, and thereby fully assess that agency's compliance with Congress's commands.

Reporter’s Note

The attached opinion has been revised to refect the usual publication and citation style of the United States Reports. The revised pagination makes available the offcial United States Reports citation in advance of publication. The syllabus has been prepared by the Reporter of Decisions for the convenience of the reader and constitutes no part of the opinion of the Court. A list of counsel who argued or fled briefs in this case, and who were members of the bar of this Court at the time this case was argued, has been inserted following the syllabus. Other revisions may include adjustments to formatting, captions, citation form, and any errant punctuation. The following additional edits were made:

p. 369: “Together with Thornton v. Collins, Secretary of Veterans Affairs (see this Court's Rule 12.4), also on certiorari to the same court” is inserted as a footnote to “Decided March 5, 2025” p. 378, line 10 from bottom: “Bufkin v. McDonough” is inserted before 75 F. 4th 1368” p. 378, line 9 from bottom: “Thornton v. McDonough” is inserted before 2023 WL 5091653” p. 395, line 7 from bottom: “interpretation” is changed to “construction”