Sanchez v. People: Colorado Upholds COLJI-Crim. E:03 (2022) “Firmly Convinced/Real Possibility” Reasonable-Doubt Instruction
1. Introduction
In Sanchez v. People, 2026 CO 60, the Colorado Supreme Court addressed a constitutional challenge to the
revised model reasonable-doubt instruction, COLJI-Crim. E:03 (2022) (the “2023 Instruction” as described by the
court). The petitioner, Alexis Teran Sanchez, had been convicted in county court of offenses arising from an
alcohol-impaired driving incident (including leaving the scene and failing to report an accident, plus the lesser
included offense of driving while ability impaired). On appeal, he argued that the revised language—requiring jurors
to be “firmly convinced” of guilt and to acquit if they believe there is a “real possibility” of non-guilt—(1)
lowered the prosecution’s burden of proof, (2) shifted the burden to the defense, and (3) violated due process and
fair-trial rights. He also argued that certain comments by the trial court during voir dire and by the prosecutor
during closing argument further diluted the burden of proof.
The case thus presented a high-stakes question of statewide importance: whether Colorado’s updated model reasonable-doubt
instruction is constitutionally sound and, if so, how reviewing courts should evaluate surrounding trial statements
alleged to erode the beyond-a-reasonable-doubt standard.
2. Summary of the Opinion
The Colorado Supreme Court affirmed the district court’s judgment and held:
-
The 2023 Instruction (COLJI-Crim. E:03 (2022)) is constitutional; it neither lowers the prosecution’s burden of proof
nor shifts that burden to the defendant.
-
Considering the instructions and record as a whole, there was no “reasonable likelihood” the jury understood the
instruction—or the trial court’s and prosecutor’s additional statements—as allowing conviction on a standard lower
than beyond a reasonable doubt.
-
Jury studies offered for the first time on appeal did not warrant altering the constitutional analysis; they were not
proper subjects for judicial notice and, in any event, would require trial-level evidentiary development.
3. Analysis
A. Precedents Cited
The court’s reasoning is built around a cluster of U.S. Supreme Court, Colorado, and federal circuit authorities
governing (i) the minimum constitutional content of reasonable-doubt instructions, (ii) how to review allegedly
burden-lowering statements, and (iii) the role of model instructions.
1) Constitutional baseline for reasonable doubt
-
Victor v. Nebraska, 511 U.S. 1 (1994): The court relied on Victor for the proposition that
“the Constitution does not require that any particular form of words be used” so long as jurors are instructed that
guilt must be proved beyond a reasonable doubt. The opinion also drew on Justice Ginsburg’s concurrence endorsing the
Federal Judicial Center’s “firmly convinced” and “real possibility” formulation as “succinct and comprehensible.”
-
Boyde v. California, 494 U.S. 370 (1990): Used to emphasize that jurors apply “commonsense
understanding” rather than parsing instructions like lawyers for “subtle shades of meaning.”
-
Cage v. Louisiana, 498 U.S. 39 (1990), and Estelle v. McGuire, 502 U.S. 62 (1991):
Cited as a contrast—Cage condemned “grave uncertainty”/“moral certainty” phrasing that risked lowering the
standard, underscoring that not all attempts to define reasonable doubt are safe.
2) Colorado’s burden-of-proof jurisprudence and review framework
-
Pettigrew v. People, 2022 CO 2: Reaffirmed that due process requires proof beyond a reasonable doubt
and that a trial court must “correctly define” reasonable doubt while retaining “reasonable discretion” in phrasing.
-
Garcia v. People, 2023 CO 30: Supplied the de novo standard for reviewing constitutionality and the
correctness of jury instructions.
-
Tibbels v. People, 2022 CO 1: Central to the opinion. Tibbels held that burden-lowering
instructions are structural error and adopted the “reasonable likelihood” test for whether jurors understood trial
court statements, in context, to permit conviction on less than beyond a reasonable doubt.
-
Johnson v. People, 2019 CO 17: Used as a comparative example—confusing or offhand embellishments do
not necessarily amount to reversible error where correct instructions are repeatedly and properly given.
-
Domingo-Gomez v. People, 125 P.3d 1043 (Colo. 2005): Provided the plain-error framework for reviewing
unobjected-to prosecutorial remarks: reversal only if fundamental fairness is undermined and the verdict reliability
seriously doubted.
3) Model instructions as guideposts, not constitutional guarantees
-
People v. Flockhart, 2013 CO 42: Pattern jury instructions are “nonbinding” but “time-trusted
guideposts.”
-
People v. Hoskin, 2016 CO 63: Pattern instructions can still be unconstitutional if they misstate or
lower the standard of proof.
-
People v. Schlehuber, 2025 COA 50, and People v. Melara, 2025 COA 48: Recent Colorado
Court of Appeals decisions upholding the same “firmly convinced”/“real possibility” formulation. The Supreme Court
expressly aligned with their reasoning that the two phrases operate together and are not tautological.
4) Federal circuit “consensus” on “firmly convinced”/“real possibility”
A major move in Sanchez is its synthesis of a broad body of federal circuit authority treating the twin
formulation as acceptable:
-
United States v. Hunt, 794 F.2d 1095 (5th Cir. 1986), and United States v. Alonzo,
681 F.2d 997 (5th Cir. 1982): Approved “firmly convinced” as substantively similar to other acceptable formulations
and warned against “abstract semantics.”
-
United States v. Velasquez, 980 F.2d 1275 (9th Cir. 1992): Treated “firmly convinced” as essentially
equivalent in function to “hesitate to act.”
-
United States v. Barrera-Gonzales, 952 F.2d 1269 (10th Cir. 1992): Suggested “firmly convinced”
conveys strong persuasion, arguably stronger than some alternative phrases.
-
United States v. Porter, 821 F.2d 968 (4th Cir. 1987): The key “skeptical” case. It criticized
potential “confusion” from “real possibility” where the court failed to explain it and did not expressly clarify the
defendant had no burden to show a “real possibility” of innocence—yet it still found no burden shift and held that
other instructions cured the problem.
-
United States v. Taylor, 997 F.2d 1551 (D.C. Cir. 1993): Found no structural error where presumption
of innocence and the prosecution’s burden were properly explained, dispelling any implication the defendant must
prove “real possibility.”
-
Cases approving the formulation post-Victor: United States v. Conway, 73 F.3d 975 (10th Cir.
1995); United States v. Williams, 20 F.3d 125 (5th Cir. 1994); United States v. Brand,
80 F.3d 560 (1st Cir. 1996); United States v. Petty, 856 F.3d 1306 (10th Cir. 2017);
United States v. Watkins, 66 F.4th 1179 (8th Cir. 2023) (quoting Harris v. Bowersox,
184 F.3d 744 (8th Cir. 1999)).
5) Out-of-state critiques and partial endorsements
-
State v. Perez, 976 P.2d 427 (Haw. Ct. App. 1998): Cited by Sanchez to argue “real possibility” and
“if you think” language invites abandonment of the presumption of innocence and that “firmly convinced” is compatible
with lingering reasonable doubt. The Colorado Supreme Court acknowledged the critique but declined to adopt it.
-
State v. Jackson, 925 A.2d 1060 (Conn. 2007), and State v. Putz, 662 N.W.2d 606
(Neb. 2003): Cited as state courts approving similar language when viewed in the context of the charge as a whole.
-
United States v. Artero, 121 F.3d 1256 (9th Cir. 1997): Approved the twin formulation relying on
Justice Ginsburg’s approval in Victor.
6) Record development, judicial notice, and expert-evidence “gatekeeping”
-
Mun. Subdistrict, N. Colo. Water Conservancy Dist. v. OXY USA, Inc., 990 P.2d 701 (Colo. 1999):
Cited for the limits of judicial notice under CRE 201(b).
-
Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579 (1993): Used to explain that trial courts, not
appellate courts, are the proper “gatekeeper” for expert evidence like jury studies.
-
Hines v. Oliver, 291 P.2d 693 (Colo. 1955) (quoting Davis v. Pursel, 134 P. 107
(Colo. 1913)): Invoked for the principle that trial courts weigh evidence and find facts, including interpreting
statistical studies.
-
Petition of Edilson, 637 P.2d 362 (Colo. 1981): Reinforced that evidence not presented below will not
be considered on review.
B. Legal Reasoning
1) The core holding: the 2023 Instruction does not lower or shift the burden
The court framed Sanchez’s challenge as attacking two phrases: (i) “firmly convinced” allegedly dilutes the required
certainty; and (ii) “real possibility” allegedly shifts the burden by asking jurors to assess whether non-guilt is
“substantially” likely.
The court rejected both arguments through a combined doctrinal and linguistic approach:
-
Doctrinal validation through precedent: The court emphasized the “overwhelming weight” of federal
authority approving the twin formulation and noted that no federal circuit has found it to be reversible error. It
also relied on Colorado appellate approval in People v. Schlehuber and People v. Melara.
-
Textual “whole instruction” analysis: The court repeatedly returned to the opening sentences of the
2023 Instruction: “The burden of proof in this case is upon the prosecution,” and the prosecution must prove “each
and every element necessary.” That allocation, in the court’s view, defeats the “burden shift” narrative that
concerned courts in cases like United States v. Porter.
-
Complementarity rather than tautology: The court endorsed the Court of Appeals’ view that “firmly
convinced” sets a high threshold for conviction while “real possibility” prevents acquittal based on fanciful,
imaginary, or merely possible doubts.
2) Commonsense meaning: “firmly convinced” and “real possibility” are ordinary and stringent
Sanchez attempted to reframe the case as an empirical/modern-language dispute, offering pop-culture examples and jury
studies. The court answered by anchoring meaning in everyday dictionary definitions:
-
“Firmly” = “strongly” or “steadfastly.”
-
“Real” = “genuine” or having “objective independent existence.”
Applying Boyde v. California, the court reasoned that jurors can apply commonsense meanings without the
instruction insinuating a lesser standard. It also pointed out that Sanchez’s own examples could be rewritten with
“strongly” and “genuine” without changing the core meaning—undercutting the claim that modern usage necessarily implies
a lower proof threshold.
3) The jury studies: why the Supreme Court refused to rely on them
The opinion is notable for how it cordons off social science evidence offered for the first time on appeal. The court
gave multiple, mutually reinforcing reasons:
-
Not part of the trial record: Under Petition of Edilson, evidence not presented below is not
considered on review.
-
Not appropriate for judicial notice: Under CRE 201(b) and Mun. Subdistrict, N. Colo. Water Conservancy Dist. v. OXY USA, Inc.,
the studies were “subject to reasonable dispute” and not the kind of indisputable fact suitable for notice.
-
Requires trial-court gatekeeping and adversarial testing: Citing Daubert v. Merrell Dow Pharms., Inc.,
the court stressed that evaluating methodology, reliability, and relevance is a trial-court task; the People had no
opportunity to confront the studies with competing experts.
-
Limited ecological validity: Particularly for the defense-funded online study, the court emphasized
that survey conditions do not replicate live instruction by a judge and collective deliberation dynamics.
Importantly, the court did not say that empirical research can never matter. Rather, it held that on this record and
posture, it would not treat appellate-filed studies as dispositive of constitutional meaning.
4) Additional statements: applying the Tibbels “reasonable likelihood” test and plain-error review
Sanchez also argued that comments during voir dire (“figure out your best guess”) and a closing remark that the jury
“need[ed] to find him guilty” diluted the standard.
-
Trial court’s voir dire exchange: The Supreme Court treated the “best guess” remark as a stray phrase
immediately followed by emphatic, repeated correct statements that the prosecution bears the burden, the burden never
shifts, and the defense need not present evidence. Unlike Tibbels v. People, the trial court did not
undermine the model instruction or replace it with a concrete but misleading analogy (like the “crack in the
foundation” example). The record instead resembled Johnson v. People, where correct instructions and
repetition neutralized any confusing aside.
-
Prosecutor’s closing remark: Because there was no contemporaneous objection, the court reviewed only
for plain error under Domingo-Gomez v. People. It found no plain error given the remark’s brevity and the
surrounding correct instructions and statements—including the prosecutor’s own reminder that jurors needed to be
“firmly convinced.”
The unifying theme is contextualism: the court refused to isolate single phrases and instead asked whether the trial
record made it reasonably likely the jury convicted on less than beyond a reasonable doubt.
C. Impact
1) Statewide validation of Colorado’s revised model instruction
Sanchez effectively settles—at the Colorado Supreme Court level—the constitutionality of the 2023 Instruction’s
signature language. Trial courts can give COLJI-Crim. E:03 (2022) “verbatim” with strong confidence that it will
withstand due process challenge, so long as the instructions and court statements clearly allocate the burden to the
prosecution and maintain the beyond-a-reasonable-doubt requirement.
2) A practical ceiling on “semantic” attacks
By emphasizing ordinary dictionary definitions, juror commonsense (per Boyde), and a large body of approving
authority, the court signaled skepticism toward challenges that depend primarily on abstract semantics divorced from
the instruction’s full context.
3) How future litigants must build records for empirical challenges
The opinion also draws a clear procedural roadmap: if a defendant wishes to use jury comprehension studies to attack an
instruction, the effort likely must begin in the trial court, with expert disclosures, admissibility litigation, and
an evidentiary record that permits adversarial testing and findings. Appellate attachment of studies—particularly
defense-funded ones—will not substitute for trial-level Daubert gatekeeping.
4) Reinforcing contextual review under Tibbels
Sanchez strengthens Tibbels as Colorado’s governing framework for evaluating whether a trial court’s
comments, in context, created a reasonable likelihood of a lowered standard. But it simultaneously narrows
Tibbels factually: improper, vivid “real life” analogies that displace the instruction remain dangerous, while
fleeting remarks followed by repeated correct burden statements are unlikely to warrant reversal.
4. Complex Concepts Simplified
-
“Beyond a reasonable doubt”: The highest burden of proof in law. It does not require absolute
certainty, but it does require a very high level of confidence before convicting.
-
“Burden shifting”: An unconstitutional move where jurors are led to believe the defendant must prove
innocence (or prove some level of doubt) rather than the prosecution proving guilt.
-
Structural error: A fundamental constitutional error affecting the framework of the trial. If a
jury instruction actually lowers the burden of proof, reversal is automatic (as referenced via Tibbels v. People).
-
“Reasonable likelihood” test: A contextual inquiry asking whether it is reasonably likely jurors took
the court’s statements and instructions to allow conviction on less than the proper standard.
-
Plain error: A demanding appellate standard when no timely objection was made at trial; reversal
occurs only if the error undermined fundamental fairness and casts serious doubt on the verdict’s reliability (per
Domingo-Gomez v. People).
-
Judicial notice (CRE 201(b)): A mechanism allowing courts to accept indisputable facts without proof.
Disputed, method-dependent social science findings typically do not qualify.
-
Daubert gatekeeping: The trial judge’s role in ensuring expert testimony rests on reliable
methods and is relevant—one reason appellate courts resist deciding empirical disputes in the first instance.
5. Conclusion
Sanchez v. People establishes a clear Colorado rule: the “firmly convinced” and “real possibility” language in
COLJI-Crim. E:03 (2022) is constitutional and does not, by itself, lower or shift the burden of proof. The decision
also demonstrates how Colorado will evaluate alleged burden dilution from surrounding trial statements: courts will
apply Tibbels’ contextual “reasonable likelihood” inquiry to judicial comments and will apply
Domingo-Gomez plain-error review to unobjected-to prosecutorial remarks. Finally, the opinion signals that
empirical challenges to jury comprehension must be built in the trial court record, not introduced for the first time
on appeal.