Lawrence Sims died when the speeding car he was driving sailed off a rural road. Prior to his untimely death, he had obtained life insurance from Great American Life Insurance Company. Sims's wife made a claim on this policy, but after an investigation Great American denied the claim under its non-accidental death provision concluding that Sims committed suicide. Mrs. Sims then filed suit against Great American alleging breach of contract and bad faith for its failure to pay her husband's life insurance policy. A jury found the death was accidental and awarded Mrs. Sims $1.4 million in compensatory and punitive damages. Great American argues on appeal that the district court made numerous evidentiary and procedural errors when it precluded Great American from introducing relevant evidence pertaining to the crash and Sims's state of mind at the time of his death.

This case requires us to consider the interplay between the Erie doctrine and the Federal Rules of Evidence in diversity cases tried in federal district court. The district court refused to admit evidence that would have been inadmissible in state court, concluding that state policy trumps the Federal Rules of Evidence. Although we conclude that the Erie doctrine does not govern the admissibility of evidence under the Federal Rules, we find that the Rules still require diversity courts to analyze substantive state policy in considering the admissibility of evidence.

Exercising jurisdiction under 28 U.S.C. § 1291, we REVERSE the district court's denial of judgment as a matter of law on both the bad faith and punitive damages claims. We otherwise AFFIRM.

On September 3, 2001, Lawrence and Clara Sims attended a golf tournament at their local country club. Sims became extremely intoxicated during the course of the day. He was still quite intoxicated when the couple returned home that evening to an unkempt kitchen. Over this seemingly minor issue, Sims expressed unexpected agitation and anger in front of his wife and daughter. He stormed out of the kitchen, jumped in his car, and sped away.

Concerned for her husband's well-being, Mrs. Sims immediately called 911. She told the 911 operator that Sims "was upset and that we were afraid that he was going to get hurt or hurt himself." R. at 807. By the next morning when her husband had still failed to return home, Mrs. Sims phoned family, friends, and local hospitals to no avail. She then filed a missing persons report. In her sworn statement to the police, she described Sims's state of mind on the evening of September 3:

Drinking—but in a good mood. Got home around 9:00, got angry because the house was a mess ... Got angrier [and] angrier—slammed dining room chair to floor [and] stormed out—saying he's "out of here." Mentioning driving off a cliff. Called 911 to let you know.

R. at 85. The official missing persons report, drafted by the police and signed by Mrs. Sims without her review, editorialized on this language, noting instead that "Mr. Sims was enraged over the cleanliness of their home ... [and] may be suicidal, because when he left his residence, Mr. Sims mentioned something about driving off a cliff." R. at 84 (emphasis added).

Two days later police recovered Sims's body. Based on evidence at the accident scene, police surmised that his car had careened off the main road onto a right-of-way toward a creek. The car, traveling at an excessively high rate of speed, clipped a fence and then hit a bump that caused it to sail some 115 feet over a creek bed, before landing in a pasture. No skid marks were visible at the scene nor any other evidence suggesting that Sims swerved before clipping the fence. Although Sims habitually wore his seat belt, on this occasion his seat belt was not fastened. Post-mortem examination revealed a blood-alcohol content (BAC) of 0.19%, well over the legal limit.

Despite Mrs. Sims's sworn police statement to the contrary, she has steadfastly held in these proceedings that she intentionally lied about her husband's intent to drive off a cliff. This lie, she claims, was merely an effort to encourage the police to act quickly to locate her missing husband. Mrs. Sims adamantly asserts that she never believed her husband was truly suicidal.

B. Life Insurance Investigation and Denial of Claim

In August 2000, Sims obtained a life insurance policy from Great American in the amount of $300,000. This policy expressly excluded recovery for non-accidental death. Following Sims's death, Mrs. Sims made a claim.

As was its typical practice, Great American hired an independent claim investigator, Broyles Claim Service Agency, to review the claim. Broyles examined a number of documents as part of its investigation, including (1) the missing persons report, (2) the accident report, (3) the medical examiner's report, (4) the death certificate, and (5) Sims's medical and pharmaceutical records. Additionally, Broyles conducted telephone interviews with the medical examiner, the investigating police officer, Mrs. Sims, and Mrs. Sims's counsel.

Both the police department accident report and the county's death certificate listed suicide as the manner of death. The investigating officer and the coroner relied heavily on the missing persons report in making their cause of death determination. In preparing the accident report, for example, the investigating officer reviewed the missing persons report as well as the physical circumstances of the accident. But the officer conceded at trial that he had never examined the car to determine if the brakes failed or showed other mechanical deficiencies which could have caused a loss of control. Although the officer did not know Sims's BAC when he originally prepared the report, he was adamant that he would still have listed suicide as the cause of death.

The medical examiner also attributed the death to suicide. However, the medical examiner did not perform an autopsy to determine if Sims suffered a stroke or exhibited symptoms of some other ailment that could have caused him to lose consciousness or control of the vehicle. When asked by Broyles why he ruled the death a suicide, the medical examiner replied "that if [Broyles had seen the] missing persons report and the accident report, we would see the reason for his [conclusion]." R. at 1764.

Broyles also obtained a recorded statement from Mrs. Sims, with her counsel present, and a separate statement from counsel as he had known Sims from the country club. Unsurprisingly, both parties insisted that Sims did not commit suicide. While Mrs. Sims did not explicitly recant her sworn police statement until after filing this lawsuit, Broyles never asked why her sworn statement contradicted her statement to him. Nor did Broyles question Mrs. Sims regarding any possible motive Sims had to commit suicide. Broyles's investigation revealed that Sims's medical and pharmaceutical records showed no indication of mental depression.

Finally, Mrs. Sims claims Broyles was aware that another officer involved at the scene believed Sims's crash was accidental. Broyles denies this charge. But in any event, Broyles did not interview this officer or any other officer with the exception of the investigating officer. Broyles never interviewed Sims's daughter, who saw him the night of the incident.

Based on this investigation, Broyles recommended that Great American deny accidental life insurance coverage and pay the claim under the suicide provision. Two agents of Great American reviewed this recommendation and concurred, tendering a return of the policy premiums that had been paid to date—$2,964.67.

Following the denial of the insurance claim, Mrs. Sims filed suit in federal court, alleging both breach of contract and bad faith. The district court denied Great American's summary judgment motion on the bad faith claim and similarly denied Great American's motion for judgment as a matter of law at the close of the trial. The jury awarded Mrs. Sims $300,000 for the breach of contract claim, $600,000 for the bad faith claim, and $500,000 in punitive damages. Following the verdict, Great American renewed its motion for judgment as a matter of law or, in the alternative, a new trial, which the district court rejected. Great American's timely appeal is now before this court.

On appeal, Great American asserts that the district court committed numerous errors, each constituting grounds for reversal:

(1) the court improperly excluded relevant evidence under Oklahoma's seat belt law that would have helped support the defense theory that Sims was suicidal on the night of the crash;

(2) the court should have admitted evidence that the coroner and investigating officer believed Sims's death a suicide;

(3) the court erred in excluding evidence of Sims's father's suicide;

(4) the jury's finding of bad faith and punitive damages was not supported by sufficient evidence at trial; and

(5) the court should have permitted Great American's accident reconstruction expert to testify at trial.1

At the heart of Great American's appeal is its argument that the Federal Rules of Evidence govern questions of admissibility in federal diversity cases, irrespective of the existence of conflicting state statutes. This argument rests on the proposition that the doctrine set forth in Erie R.R. v. Tompkins, 304 U.S. 64, 58 S. Ct. 817, 82 L. Ed. 1188 (1938), is inapplicable to these Rules. Based on this premise, Great American asserts that the district court improperly applied state law, as opposed to the Federal Rules of Evidence, to exclude (1) evidence showing Sims failed to wear his seat belt on the night he died, as well as (2) reports and testimony from state officials concerning Sims's manner of death.

We are persuaded that the Federal Rules of Evidence are not governed by the Erie doctrine. Having said that, concepts of federalism still guide our understanding of the interplay between the Federal Rules and state law. Although we do not ground our holding in Erie, we conclude that state substantive policy directs the admissibility of evidence under the relevancy considerations of Rule 401.

This case presents us with a classic civil procedure question—in the face of a conflicting state statute, when does a federal court sitting in diversity apply federal law? This question, of course, invokes the Supreme Court's seminal decision in Erie Railroad v. Tompkins and the landmark trilogy which followed. See Guar. Trust Co. v. York, 326 U.S. 99, 65 S. Ct. 1464, 89 L. Ed. 2079 (1945) (creating the outcome determination test); Byrd v. Blue Ridge Rural Elec. Coop., 356 U.S. 525, 78 S. Ct. 893, 2 L. Ed. 2d 953 (1958) (holding that the outcome determination test is not dispositive in the face of countervailing federal interests); Hanna v. Plumer, 380 U.S. 460, 85 S. Ct. 1136, 14 L. Ed. 2d 8 (1965) (holding Erie inapplicable to the Federal Rules of Civil Procedure). Collectively, the "broad command" flowing from these cases requires federal courts "to apply state substantive law and federal procedural law." Hanna, 380 U.S. at 465, 85 S. Ct. 1136. However, the sweep of this broad command is limited—only federal common law is governed by the Erie doctrine; congressional acts and the Federal Constitution fall outside its scope. As the Erie court famously stated:

Except in matters governed by the Federal Constitution or by acts of Congress, the law to be applied in any case is the law of the state .... There is no federal general common law. Congress has no power to declare substantive rules of common law applicable in a state whether they be local in their nature or "general," be they commercial law or a part of the law of torts.

Erie, 304 U.S. at 78, 58 S. Ct. 817 (emphasis added). This seemingly simple rule, however, has a surprisingly complex application when the federal rule at issue is one of the Federal Rules of Evidence.

Great American argues that the Federal Rules of Evidence are an "act of Congress," not federal common law, and therefore outside the purview of Erie. Accordingly, when applying these Rules, Great American continues, federal diversity courts should not look to Erie but instead should apply the test formulated in Stewart Org., Inc. v. Ricoh Corp., 487 U.S. 22, 108 S. Ct. 2239, 101 L. Ed. 2d 22 (1988), which governs the application of a federal statute to a diversity case. Following Stewart, a court asks whether the evidentiary rule is both (1) "sufficiently broad to control the issue before the Court," and (2) "a valid exercise of Congress' authority under the Constitution." Id. To be a valid exercise of congressional authority, the federal rule must be "capable of classification" as a procedural rule. Id. at 32, 108 S. Ct. 2239 (quoting Hanna, 380 U.S. at 472, 85 S. Ct. 1136). If the rule satisfies both conditions, absent other considerations, the federal evidentiary rule applies. See id. at 26, 108 S. Ct. 2239 (noting that a "district court's decision whether to apply a federal statute ... in a diversity action ... involves a considerably less intricate analysis than that which governs the `relatively unguided Erie choice'") (quoting Hanna, 380 U.S. at 471, 85 S. Ct. 1136).

A brief review of the origins of the Federal Rules of Evidence shows that Stewart provides the starting point in determining the admissibility of evidence in this case.

a. Origin of Federal Rules of Evidence

Great American's argument rises or falls on the premise that the Federal Rules of Evidence are a product of congressional action. Our review of the relevant history reveals this to be so.

Prior to the enactment of the Federal Rules of Evidence, Federal Rule of Civil Procedure 43(a) governed evidentiary matters in federal courts. 19 Charles Alan Wright, Arthur R. Miller & Edward H. Cooper, Federal Practice and Procedure § 4512 (2d ed. 1996 & Supp.2005) ("Federal Practice"). Under Rule 43, federal courts typically applied state evidentiary law to questions regarding the admissibility of evidence. Id. ("[T]he norm was to follow forum state law on evidentiary matters, not because Erie demanded it, but simply because the Civil Rule so directed."). But in 1972, the Supreme Court started drafting an independent set of rules governing the admissibility of evidence in federal courts: the Federal Rules of Evidence.

At the time the Supreme Court was considering the Rules of Evidence, Congress passed a statute requiring congressional approval before the Court could promulgate not only these Rules but also amendments that had been proposed in 1972 to the already promulgated Federal Rules of Civil and Criminal Procedure. See Act of March 30, 1973, Pub.L. No. 93-12, 87 Stat. 9. Congress's concern, in large part, was based on a desire to protect substantive state law. 19 Federal Practice § 4512. In this way, Congress became intricately involved in formulating the Federal Rules of Evidence, and the Rules themselves became an act of Congress.2 See Act of Jan. 2, 1975, Pub.L. No. 93-595, 88 Stat.1926. Remembering that Erie limits its holding to federal common law and excepts congressional actions, this history makes clear that Erie does not apply to the original Federal Rules of Evidence.

Nor can we find that Congress otherwise explicitly limited the applicability of the Federal Rules of Evidence in the face of conflicting state law. Two statutes normally limit the effect of evidentiary and procedural rules, the Rules Enabling Act3 and the Rules of Decision Act.4 But these laws do not govern the Federal Rules of Evidence as originally enacted. 19 Federal Practice § 4512; see Flaminio v. Honda Motor Co., 733 F.2d 463, 470 (7th Cir. 1984). The Rules Enabling Act proscribes the ability of the Supreme Court to alter substantive rights. Here, with the original enactment of the Federal Rules of Evidence, it was Congress, not the Supreme Court, creating the Rules. Under the Rules of Decision Act, state law provides for the rule of decision in civil cases except where Congress or the Constitution provides otherwise. Again, we are dealing with a congressional act. Put simply, there is nothing in these Acts that limits the scope of the Federal Rules of Evidence as originally enacted.5

In sum, because Erie does not apply to acts of Congress, the substance/procedure dichotomy embodied in that doctrine is inapplicable to the Federal Rules of Evidence as originally enacted. Further, because an act of Congress is subject to neither the Rules Enabling Act nor the Rules of Decision Act, these Acts do not work to limit the applicability of the unamended Federal Rules of Evidence in the face of conflicting state law.

b. Federal Court Application of Erie Doctrine to Federal Rules of Evidence

We recognize that numerous federal diversity courts have grappled with this tricky issue and reached an opposite conclusion. These courts, such as the district court below, suggest we limit application of the Federal Rules of Evidence to those matters not implicating substantive state policy. See, e.g., McInnis v. A.M.F., Inc., 765 F.2d 240, 245 (1st Cir.1985); Szantay v. Beech Aircraft Corp., 349 F.2d 60, 63 (4th Cir.1965); Potts v. Benjamin, 882 F.2d 1320, 1324 (8th Cir.1989); Moe v. Avions Marcel Dassault-Breguet Aviation, 727 F.2d 917, 930-33 (10th Cir.1984). These courts falter, though, because they ground this limitation in Erie and, as we have just discussed, Erie is inapplicable to the Federal Rules of Evidence, a product of congressional action.

Under their approach, federal courts apply the Stewart analysis only if the conflicting state evidentiary rule is procedural. If, however, the state evidentiary rule is substantive, these courts have held that Erie controls and state law prevails over conflicting federal evidentiary rules. State substantive rules, then, are governed by an Erie analysis. Examples of substantive state evidentiary rules where courts have applied Erie include such diverse topics as (1) the parol evidence rule; (2) evidence concerning the amount of a plaintiff's settlement with other defendants; (3) findings of medical malpractice screening panels; (4) the objective evidence requirement; (5) burden of proof rules; and (6) as we have here, the failure to use seat belts. 17A James Wm. Moore et al., Moore's Federal Practice ¶ 124.09 (3d ed.2006).

Yet, as previously explained, this Erie approach turns Stewart on its head. Stewart requires the court to look first at the federal evidentiary rule. If that rule is arguably procedural, it governs even when a state substantive law is on point. Simply put, because Erie is inapplicable, we must first look to the Federal Rules of Evidence.

c. Tenth Circuit Approach

Our precedent suggests that where a conflict exists between the Federal Rules of Evidence and state law, we apply the Federal Rules unless the state law reflects substantive concerns or policies.

The admissibility of evidence in diversity cases in federal court is generally governed by federal law. Nevertheless, it is well recognized that Congress did not intend the procedural rules to preempt the so-called substantive state rules of evidence, such as the parol evidence rule, the collateral source rule, or the statute of frauds; although the application of these rules will affect the admissibility of some evidence, they in reality serve substantive state policies regulating private transactions.

Macsenti v. Becker, 237 F.3d 1223, 1241 (10th Cir.2001) (quotations omitted); see Romine v. Parman, 831 F.2d 944, 944-45 (10th Cir.1987) (noting that while "admissibility of evidence in diversity cases in federal court is generally governed by federal law," state law must be applied when "an evidentiary question is so dependent on a state substantive policy").

In our most noteworthy and extensive discussion of this issue, Moe v. Avions Marcel Dassault-Breguet Aviation, 727 F.2d 917, 930-33 (10th Cir.1984), we addressed the applicability of the Federal Rule of Evidence governing subsequent remedial measures. We held that state law rather than Rule 407 prevails, thus rejecting the notion that the "admissibility of evidence in diversity actions is governed exclusively by federal law—that is, the Federal Rules of Evidence." Id. at 931. We employed instead an Erie analysis and categorized the state law as one of substance, not procedure. In doing so, we stated such distinction was necessary to effect the twin goals of Erie—(1) uniformity, and (2) prevention of forum-shopping. Id. at 932. We concluded,

If the law of the state supplies the rule of decision, there is no justification for reliance on Rule 407. We recognize that, by its terms, Rule 407, when read in conjunction with Rules 401 and 402, does appear to apply in these cases. However, such a result is an unwarranted incursion into the Erie doctrine.

Id.

We are careful to point out, however, that the court's discussion regarding the application of the Federal Rules of Evidence was dicta. As the court observed: "Notwithstanding our view that the trial court erred in ruling that ... Rule 407 applies in diversity actions without regard to state law, we hold that no harm resulted therefrom because the trial court's actions were proper and correct on other grounds." Id. at 933. Indeed, one judge in a concurring opinion refused to join the court's Erie discussion, stating,

I concur in [the] result and in the opinion of the court except to the extent that it purports to resolve the difficult question of possible conflict between state and federal rules of evidence. As the court's opinion makes clear, that discussion is not necessary to our decision which properly rests on the independent ground of balancing mandated by Rule 403 and parallel state authority.

Id. at 936 (McKay, J., concurring).

d. Effect of Rule 401

Although the Moe court's reasoning is tempting, we are persuaded that Erie is inapplicable to the Federal Rules of Evidence. We recognize, of course, that a federal court's apparently unchecked ability to apply the Federal Rules of Evidence without any regard for state substantive policy creates considerable tension with notions of federalism. But this tension is overstated. Congress did not give federal courts unbridled discretion to preempt state substantive law on all arguably procedural matters. Instead, the Rules of Evidence expressly provide for the application of state law in numerous circumstances. See Fed.R.Evid. 302 (presumptions); id. 501 (privileges); id. 601 (competency of witnesses).

Most pertinent to this case, Rule 401 protects against incursion into state substantive policy as part of its admissibility assessment.6 Rule 401 defines relevant evidence as "evidence having any tendency to make the existence of any fact that is of consequence to the determination of the action more probable or less probable than it would be without the evidence." (emphasis added).

This definition requires a dual inquiry: (1) whether the evidence is probative or factually relevant to the proposition asserted (i.e., whether the evidence tends to make the existence of that fact more or less probable), and (2) whether the proposition for which the evidence is offered is properly provable in the case (i.e., the fact is material—of consequence—to the question of state law).7 19 Federal Practice § 4512; see Fed.R.Evid. 401 advisory committee's notes ("Relevancy is not an inherent characteristic of any item of evidence but exists only as a relation between an item of evidence and a matter properly provable in the case."); Olin Guy Wellborn III, The Federal Rules of Evidence and the Application of State Law in the Federal Courts, 55 Tex. L.Rev. 371, 388-89 (1977). The first inquiry is a procedural question of evidence law and the second is a substantive question regarding the materiality of the evidence. Wellborn, supra, at 389; see Telum, Inc. v. E.F. Hutton Credit Corp., 859 F.2d 835, 838 (10th Cir. 1988) ("Substantive law ... determines which facts are of consequence in a given action.").

Under the second inquiry, a fact may be "of consequence" to the determination only if it is "the kind of fact to which proof may properly be directed." Fed.R.Evid. 401 advisory committee's notes; Wellborn, supra, at 406.

Where a state law excludes certain evidence in order to effect substantive policy considerations, Rule 401 acts to exclude the evidence since the proposition for which the evidence is submitted is not properly provable and, therefore, irrelevant to the claim.8 See Huff v. Shumate, 360 F. Supp. 2d 1197 (D.Wyo.2004) (holding that although the Federal Rules of Evidence apply, the Rules nonetheless bow to state substantive policy based on the relevancy requirement of Rules 401 and 402); see also Jack B. Weinstein & Margaret A. Berger, Weinstein's Federal Evidence, § 401.04[3][b] (Joseph M. McLaughlin ed.2006). For example, state law defines the elements and defenses of a cause of action in a diversity case. If, in such a case, a defendant proffers evidence supporting a defense that is no longer permitted by state statute, that proffer is of no consequence to the action and therefore not properly provable.

In sum, although we find that the Federal Rules of Evidence are not subject to Erie's substance/procedure distinction, this distinction still has meaning in making evidentiary determinations in federal diversity cases.

e. Substance versus Procedure

The distinction between state substantive and procedural law has proven elusive, largely because many state rules generally reflect a mix of substantive and procedural concerns. See D. Michael Risinger, "Substance" and "Procedure" Revisited with Some Afterthoughts on the Constitutional Problems of "Irrebuttable Presumptions," 30 UCLA L.Rev. 189, 190 (1982) (noting that "the closest thing there is to a developed school of thought concerning the meaning of the procedure-substance dichotomy is really an abdication of analysis, wilfully embraced."). Nonetheless, some fundamental elements consistently define the substance-procedure dichotomy.

"A pure rule of evidence ... is concerned solely with accuracy and economy in litigation ...." Barron v. Ford Motor Co., 965 F.2d 195, 199 (7th Cir.1992). Such rules "are addressed to lawyers and judges in their professional roles and govern the means by which disputes regarding the content or application of substantive rules should be resolved. The purpose of these rules is to achieve accuracy, efficiency, and fair play in litigation, without regard to the substantive interests of the parties." Michael Lewis Wells, The Impact of Substantive Interests on the Law of Federal Courts, 30 Wm. & Mary L.Rev. 499, 504 (1989).

As John Hart Ely observed,

[A] procedural rule is ... one designed to make the process of litigation a fair and efficient mechanism for the resolution of disputes. Thus, one way of doing things may be chosen over another because it is thought to be more likely to get at the truth, or better calculated to give the parties a fair opportunity to present their sides of the story, or because, and this may point quite the other way, it is a means of promoting the efficiency of the process.

John Hart Ely, The Irrepressible Myth of Erie, 87 Harv. L.Rev. 693, 724-25 (1974).

In contrast, "a substantive rule is concerned with the channeling of behavior outside the courtroom." Barron, 965 F.2d at 199. Substantive rules "are directed at individuals and governments and tell them to do or abstain from certain conduct on pain of some sanction. Substantive rules are based on legislative and judicial assessments of the society's wants and needs, and they help to shape the world of primary activity outside the courtroom." Wells, supra, at 504. Ely elaborates upon this idea:

The most helpful way ... of defining a substantive rule ... is as a right granted for one or more nonprocedural reasons, for some purpose or purposes not having to do with the fairness or efficiency of the litigation process. Thus, in attempting to give content to the notion of substance, the literature has focused on those rules of law which characteristically and reasonably affect people's conduct at the stage of primary private activity.

Ely, supra, at 725 (internal citations and quotations omitted).9

In short, although the distinction between substance and procedure is not always clear, we can distinguish a substantive rule from a procedural rule by examining the language and the policy of the rule in question. If these inquiries point to achieving fair, accurate, and efficient resolutions of disputes, the rule is procedural. If, however, the primary objective is directed to influencing conduct through legal incentives, the rule is substantive.

f. Summary

The Federal Rules of Evidence are an act of Congress and, thus, subject neither to the dictates of the Erie doctrine nor to the Rules Enabling Act or Rules of Decision Act. Yet, we are mindful that Erie "recognized that the scheme of our Constitution envisions an allocation of law-making functions between state and federal legislative processes which is undercut if the federal judiciary can make substantive law affecting state affairs beyond the bounds of congressional legislative powers." Hanna, 380 U.S. at 474-75, 85 S. Ct. 1136 (Harlan, J. concurring).

But we also recognize that Congress may "prescribe housekeeping rules for federal courts even though some of those rules will inevitably differ from comparable state rules." Id. at 473, 85 S. Ct. 1136. This is not to say, however, that Congress has done so in a manner that disrespects "one of the modern cornerstones of our federalism." Id. at 474, 85 S. Ct. 1136 (Harlan, J. concurring). Although Erie does not apply, its reasoning remains instructive. In formulating Rule 401, Congress maintained the delicate balance between the state and federal systems. We keep with that goal: a fact may be "of consequence" in a diversity action only if substantive state policy so allows.

With this analysis behind us, we now address the district court's exclusion of evidence regarding Sims's failure to use his seat belt and testimony from state officials concerning Sims's manner of death. In doing so, the first question we resolve is whether, under Rule 401, the evidence offered is "of consequence" to the action. This inquiry necessarily requires us to consider if the state law excluding this evidence reflects substantive policy considerations.

Great American's principal evidentiary argument on appeal is that it should have been permitted to offer evidence showing Sims did not fasten his seat belt on the night he died.

The district court relied on Oklahoma's Mandatory Seat Belt Act to exclude this evidence. See Okla. Stat. tit. 47, § 12-420 (barring evidence of the use or non-use of a seatbelt in any civil suit). The court reasoned that in a diversity action, although the admissibility of evidence is generally governed by the Federal Rules of Evidence, when the "evidentiary question is so dependent on state substantive policy, state law must be applied." R. at 467 (quoting Romine v. Parman, 831 F.2d 944 (10th Cir.1987)). Great American contends this logic is flawed for two independent reasons: (1) because the Federal Rules of Evidence govern, this evidence is admissible, and (2) even if the Rules are not applicable, the statute does not apply where a defendant is not attempting to insinuate fault but, as here, is only attempting to prove a driver's state of mind.

Although we hold that the Rules exclusively govern in federal diversity cases, we still must determine, in accordance with our Rule 401 relevancy assessment, whether § 12-420 is a procedural rule or a substantive one. Because this determination necessarily involves interpretation of Oklahoma law, we apply the most recent pronouncement of the Oklahoma Supreme Court. If, however, the Oklahoma Supreme Court has not yet spoken on the issue, we predict how that court would address the issue. Blackhawk-Central City Sanitation Dist. v. Am. Guar. & Liab. Ins. Co., 214 F.3d 1183, 1188 (10th Cir.2000). We review de novo the district court's interpretation of Oklahoma law. Burton v. R.J. Reynolds Tobacco Co., 397 F.3d 906, 910 (10th Cir.2005).

a. Mandatory Seat Belt Act

Oklahoma enacted its Mandatory Seat Belt Act to encourage seat belt use on the roads of Oklahoma. See Bishop v. Takata Corp., 12 P.3d 459, 464 (Okla.2000) (noting that "the obvious purpose[] of the Act [is] to codify public policy of encouraging seat belt use"); see generally §§ 12-416 to 420. To this effect, the Act requires drivers and front-seat passengers to fasten their seat belts. However, the Act is careful to limit the civil court ramifications for failure to wear a seat belt, directing that evidence of the use or non-use of seat belts is inadmissible "in any civil suit in Oklahoma." Okla. Stat. tit. 47, § 12-420.

Even though the Oklahoma Supreme Court has not squarely addressed whether § 12-420 is a substantive rule, a careful review of decisions from that court persuades us that it is.10 In Bishop v. Takata Corp., the court described the purpose and scope of the Act in holding it inapplicable to products liability cases:

Considering the Act as a whole, and the context from which it was enacted, the obvious purposes of the Act are to codify public policy of encouraging seat belt use and to make seat belt use mandatory for drivers and front-seat passengers by providing a penalty for nonuse. Section 12-420 expressly clarifies that the sole legal sanction for the failure to wear a seat belt is the fine imposed by the Act and that a person will not be penalized in a civil proceeding, by connotations of fault, for choosing to refrain from wearing a seat belt. To read the statute any more broadly would defeat the legislative intent. The entire Act is confined to the regulation of the conduct of the driver or the passenger of a passenger vehicle as seat belt users or nonusers.

12 P.3d at 464 (emphasis added). A few years later, the court in Clark v. Mazda Motor Corp., 68 P.3d 207, 208 (Okla.2003), further clarified the scope of the Act, stating that the public policy encouraging seat belt use "protects persons in civil proceedings from connotations of fault."

While not completely clear, the language in Bishop and Clark suggests that § 12-420 reflects substantive state policy to the extent that drivers and passengers, while required by law to wear their seat belts, should not be penalized, through connotations of fault, beyond a small statutory fine.11 The rule is not concerned with how seat-belt evidence might affect the accurate resolution of litigated disputes. Nor is it concerned with the efficient resolution of disputes. To the contrary, the rule was designed to protect drivers from being blamed in a court of law for failure to wear a seat belt. Cf. Barron, 965 F.2d at 199 (A state law that bars admission of seat belt evidence "is a rule of evidence if it is motivated by concern that jurors attach too much weight to a plaintiff's failure to wear his seatbelt [but] is a substantive rule if it is designed not to penalize persons who fail to fasten their seatbelts.").

A negligence action arising from a car accident illustrates how § 12-420 acts to avoid penalizing litigants with connotations of fault. In this scenario, a driver not wearing his seat belt is injured in a car accident caused by another vehicle. The injured driver sues. At trial, the defendant wants to argue that had the plaintiff been wearing a seat belt, plaintiff's injuries would have been minimized and, accordingly, plaintiff's recovery should be reduced. Section 12-420 clearly precludes this argument.

We are not presented with such a case here. Instead, Great American, the defendant, wishes to present evidence of non-use of a seat belt as proof of Sims's, the driver's, suicidal intent. Section 12-420 was not created to apply to such facts. The Oklahoma Supreme Court currently limits the application of § 12-420 to negligence cases. See Bishop, 12 P.3d at 466 (noting that the Oklahoma Supreme Court held that the Act clearly "precludes the introduction of evidence of the use or nonuse of seat belts to support a claim of negligence in a wrongful death action ....") (citing Comer v. Preferred Risk Mut. Ins. Co., 991 P.2d 1006, 1008 (Okla. 1999)); Clark, 68 P.3d at 208 (Okla.2003) ("Section 12-420 ... prohibits introduction of [evidence of the use or non-use of seat belts] to impute negligence or fault to a person who elects not to wear a seat belt.").

That is not to say that the court might not later expand its application, but the court has indicated a preference to limit the Act to disputes where the conduct of the driver is at issue. See Bishop, 12 P.3d at 464 (noting that the Act "is confined to the regulation of the conduct of the driver."). For that reason, a number of Oklahoma decisions find the Act inapplicable in cases involving products liability. Clark, 68 P.3d at 210 (Opala, J. concurring); Bishop, 12 P.3d at 466.

Great American does not argue that Sims was negligent for failing to wear his seat belt or otherwise insinuate fault for this omission. It argues he was suicidal. Whether a mandatory seat belt law exists in Oklahoma is irrelevant to this argument. In a negligence case, the existence of a seat belt law could be critical to a damages assessment because it imparts a duty on the injured party to wear a seat belt. See, e.g., Gardner ex rel. Gardner v. Chrysler Corp., 89 F.3d 729, 733 (10th Cir.1996) (noting that the Kansas legislature modified earlier law in order to clarify the extent of the duty owed under a mandatory seat belt law). But here duty is not an element of the claim. Great American frankly is unconcerned whether Sims could have avoided his injuries had he been wearing his seat belt. Great American is only concerned with how his failure to wear a seat belt reflects on his mental state. In this sense, the evidence does not speak to the conduct of the driver but instead to the driver's state of mind.

Because Great American introduced this evidence for purposes of showing Sims's state of mind, not to insinuate fault, we hold that Oklahoma's Mandatory Seat Belt Act is inapplicable to the present case, and the district court should have admitted this evidence at trial. Cf. Barron, 965 F.2d at 198-200 (holding that North Carolina's seat belt exclusionary rule, while a substantive law, is inapplicable to the present case because the use of evidence "was not to show that [Plaintiff] failed to mitigate the consequences of the accident but to show that [Defendant] had been reasonable").

Having concluded the seat belt evidence was admissible, we turn to what effect the court's preclusion of this evidence had at trial.

b. Harmless Error

Although we find error here, reversal is not appropriate unless the error prejudicially affected a substantial right of a party or if we reasonably conclude that the exclusion of this evidence led the jury to reach a contrary result. Praseuth v. Rubbermaid, Inc., 406 F.3d 1245, 1253 (10th Cir.2005); see Fed. R.Civ.P. 61. We look to "the particular jury in the particular circumstances of the trial." United States v. Hedgepeth, 418 F.3d 411, 421 (4th Cir.2005) (quotation omitted); see also United States v. Crawford, 130 F.3d 1321, 1324 (8th Cir.1997). Put another way, after reviewing the entire record, we must reverse unless we find that this jury's verdict more probably than not was unaffected by the error. Obrey v. Johnson, 400 F.3d 691, 699-701 (9th Cir.2005); see Mehojah v. Drummond, 56 F.3d 1213, 1215 (10th Cir.1995) (requiring review of entire record). We note that an "important factor in determining whether an error was harmless is the strength of the case in support of the verdict.... The risk is greater that a particular error tipped the scales in a close case than in one in which the evidence was extremely one-sided."12 11 Charles Alan Wright, Arthur R. Miller & Edward H. Cooper, Federal Practice and Procedure § 2883 (2d ed. 1995 & Supp.2005).

Here, Great American planned to offer testimony that Sims habitually wore his seat belt but failed to do so on the night he died.13 We agree this evidence is probative of Sims's state of mind. While the jury surely could have found Sims failed to buckle up, not because he was suicidal but because he was intoxicated, the very nature of habit evidence is that it is done reflexively. Habit "is the person's regular practice of meeting a particular kind of situation with a specific type of conduct." Fed.R.Evid. 406 advisory committee's notes (quotation omitted); accord United States v. Serrata, 425 F.3d 886, 906 (10th Cir.2005). Accordingly, the jury might have weighed this habit evidence along with the lack of skid marks at the scene, the car's high rate of speed, and Sims's alleged mention of driving off a cliff, and found Sims intended to commit suicide.

We nonetheless find the exclusion harmless. At trial, Great American had the burden of proving Sims committed suicide and it, along with Mrs. Sims, presented extensive evidence on this issue. Indeed, this issue was the crux of the case. Great American offered testimony that Sims was traveling at an excessive rate of speed, testimony that he did not attempt to avoid the accident, testimony from Mrs. Sims's daughter that Sims left the house crying, and Mrs. Sims's sworn statement to police that Sims intended to drive off a cliff. Yet, the jury unequivocally rejected Great American's characterization of the evidence when it not only found for Mrs. Sims on the breach of contract claim but also on the bad faith and punitive damages claims. It is fair to say that if the question had been close, the jury would not have awarded both bad faith and punitive damages.

Therefore, we cannot reasonably conclude that the seat belt evidence, by itself, would have led the jury to find in favor of Great American on the breach of contract claim.14

3. Admissibility of Evidence from State Officials

Great American next asserts the district court improperly limited the jury's consideration of evidence offered to refute the breach of contract claim: (1) the medical examiner's report, (2) the death certificate, (3) the accident report, and (4) statements made by the investigating officer concerning Sims's manner of death.15 The court erred, Great American argues, in applying Oklahoma law rather than the Federal Rules of Evidence. As with the seat belt evidence, this claim requires us to determine whether the evidence is of consequence under Rule 401, or stated differently, whether the Oklahoma law at issue reflects substantive state policy.

a. Expert Reports

We begin with an examination of the two expert reports: the medical examiner's report and the death certificate. Pursuant to Oklahoma law, such reports generally are excluded either as hearsay or improper opinion testimony. See Travelers Protective Ass'n v. Mansell, 540 P.2d 1178, 1180 (Okla.1975) (excluding death certificate as hearsay); Fed. Life Ins. Co. v. Maples, 204 Okla. 195, 228 P.2d 363 (1951) (excluding proof of death as hearsay and conclusion); N.Y. Life Ins. Co. v. Gibbs, 176 Okla. 535, 56 P.2d 1179, 1180 (1936) (excluding verdict of coroner's jury as hearsay); Metro. Life Ins. Co. v. Plunkett, 129 Okla. 292, 264 P. 827, 828 (1928) (excluding statement by attending physician in proof of death and death certificate as hearsay); Okla. Aid Ass'n v. Thomas, 125 Okla. 190, 256 P. 719, 721-22 (1927) (excluding death certificate as hearsay).

In holding these types of records inadmissible for the purpose of showing suicide, the Oklahoma Supreme Court has stated,

It is our opinion that the Legislature provided for the keeping of vital statistics ... for the purpose of keeping an accurate record of births and deaths and of the diseases causing death, and so that the health authorities may be better enabled to combat diseases. The attending physician or coroner might be able to state the cause of death, just as was stated here, gunshot wound. But to go further and state by whom inflicted would change all the rules of evidence in cases in which this certificate could be admitted.

Thomas, 256 P. at 721-22 (emphasis added). The court was clearly troubled that a jury in such circumstances could be influenced by hearsay testimony. See id. at 721 (noting that admitting such testimony would "permit hearsay evidence"). Indeed, the court later reiterated: "To know ... how the deceased received the injury that caused his death would require the testimony of an eyewitness, because the injury was of the type which could have been incurred either accidentally or intentionally." Maples, 228 P.2d at 367. Therefore, evidence as to cause of death is "inadmissible for the reason that they [are] based on hearsay and conclusion." Id.

Based on this precedent, we are persuaded that Oklahoma law barring such testimony reflects a procedural rule. The court's objective in precluding such evidence is to promote accuracy in its courtrooms. See Thomas, 256 P. at 721 (noting that the case illustrates "the danger of permitting [a] death certificate to be introduced to make out a prima facie case of suicide."). Therefore, because the state rules are procedural under Federal Rule of Evidence 401, the district court should not have relied on state law in conducting its admissibility determination.

Nevertheless, we find no error in the exclusion of these reports. Under Rule 702, reports from experts, such as a medical examiner, are admissible only if necessary to aid in the interpretation of scientific, technical, or other specialized facts. Although any witness may offer an opinion as to an ultimate issue to be decided by a jury, this opinion should not unduly invade the province of the jury when the assistance of the witness is unnecessary. See United States v. Dazey, 403 F.3d 1147, 1172 (10th Cir.2005); see Fed.R.Evid. 704 advisory committee's notes ("Under Rules 701 and 702, opinions must be helpful to the trier of fact, and Rule 403 provides for exclusion of evidence which wastes time. These provisions afford ample assurances against the admission of opinions which would merely tell the jury what result to reach, somewhat in the manner of the oath-helpers of an earlier day.").

Here, the jury was fully capable of assessing the facts to determine causation. The experts did not offer any opinion that was based on scientific or technical facts outside the jurors' common knowledge or experience. To the contrary, this is the very type of fact determination a jury is equipped to make. Accordingly, although the district court incorrectly applied state law, the court's analysis nonetheless conforms to the appropriate analysis under Federal Rule of Evidence 702, which would have limited the admissibility of the medical examiner's report and the death certificate.

b. Evidence from Investigating Officer

The district court also limited the investigating officer's accident report as well as his testimony relating to Sims's cause of death.16 The court relied on Gabus v. Harvey, 678 P.2d 253, 257 (Okla.1984), to exclude this evidence. In that case, the court emphasized the capability of the jury in assessing certain facts:

[The evidence] concerned facts that could be readily appreciated by any person who drives an automobile or crosses streets. No special skill or knowledge was needed to understand these facts and draw a conclusion from them. In such a case as this, where the normal experiences and qualifications of laymen jurors permit them to draw proper conclusions from the facts and circumstances, expert conclusions or opinions are inadmissible. The expert conclusion here was not helpful and should not have been admitted.... It was not useful since the jury was just as capable of drawing a proper conclusion from those facts as was the officer.

Id. at 256-57; see Maben v. Lee, 260 P.2d 1064 (Okla.1953) (holding that an investigating officer without personal knowledge cannot offer an opinion as to causation at trial when the jury is as capable of doing so as is the officer); see also Diaz v. State, 728 P.2d 503, 514 (Okla.Crim.App.1986) ("It is only where the fact of death and its cause is beyond the understanding of the average layman that expert testimony may be necessary.") (quotation omitted).

Based on this language, we believe that the Oklahoma rule precluding a police officer's opinion about causation is a procedural rule concerned solely with accuracy and economy in the courtroom. Therefore, the district court should not have grounded its conclusion in Oklahoma procedural law, but instead should have relied exclusively on the appropriate Federal Rule of Evidence to guide its admissibility determination.

Nonetheless, we find no error in the exclusion of this evidence. The holding from the Gabus court relied extensively on Oklahoma Rules of Evidence § 2403 and § 2702. Gabus, 678 P.2d at 254-55; see Okla. Stat. tit. 12, § 2403 ("Although relevant, evidence may be excluded if its probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, misleading the jury, undue delay, needless presentation of cumulative evidence, or unfair and harmful surprise."); id. § 2702 ("If scientific, technical or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue, a witness qualified as an expert by knowledge, skill, experience, training or education may testify in the form of an opinion or otherwise."). These rules are identical to Federal Rules of Evidence 403 and 702. Under the Federal Rules, statements by investigating officers as to accident causation are generally not admitted unless (1) perceived by the witness and helpful to a determination of a fact in issue, see Fed.R.Evid. 701, or (2) in the case of an expert, necessary to aid the jury in the interpretation of scientific, technical, or other specialized facts, see id. 702. Neither of these conditions are met here. As we stated earlier, the jury was fully capable of concluding whether or not Sims committed suicide. See Dazey, 403 F.3d at 1172. The investigating officer's opinion on this matter was no more than speculation based on the same facts that the jury had before it. We therefore find that had the district court properly applied the Federal Rules, it would have excluded the officer's testimony concerning causation. In short, there was no error in the court's exclusion of the testimony under state law.

Great American's last evidentiary argument is that the district court improperly excluded evidence showing that Sims's father committed suicide. The court concluded that the danger of unfair prejudice substantially outweighed the probative value of the evidence under Federal Rule of Evidence 403. Because this evidence is both irrelevant17 and highly prejudicial, the district court did not abuse its discretion. See Christiansen v. City of Tulsa, 332 F.3d 1270, 1283 (10th Cir.2003).

In addition to its evidentiary claims, Great American asserts that the evidence was insufficient to support the bad faith and punitive damages awards. We review de novo a district court's denial of a motion for judgment as a matter of law, viewing the evidence in a light most favorable to the opposing party and drawing all reasonable inferences therefrom. Hardeman v. City of Albuquerque, 377 F.3d 1106, 1112 (10th Cir.2004). We do "not weigh the evidence, pass on the credibility of the witnesses, or substitute [our] conclusions for that of the jury." Id. at 1116 (internal quotation marks omitted). Only if "there is no legally sufficient evidentiary basis for a reasonable jury to find for the issue against that party" do we reverse a jury's verdict. Fed.R.Civ.P. 50(a)(1).

Under Oklahoma law, "an insurer has an implied duty to deal fairly and act in good faith with its insured." Christian v. Am. Home Assurance Co., 577 P.2d 899, 904 (Okla.1977). If an insurer fails to fulfill this duty, a bad faith action in tort can arise. Id. This is not to say, however, that an insurer may never dispute coverage; to the contrary, an insurer needs only a legitimate basis for doing so. Id. at 905; Timberlake Constr. Co. v. U.S. Fid. & Guar. Co., 71 F.3d 335, 343 (10th Cir.1995). The Oklahoma Supreme Court cautions:

We recognize that there can be disagreements between insurer and insured on a variety of matters such as insurable interest, extent of coverage, cause of loss, amount of loss, or breach of policy conditions. Resort to a judicial forum is not per se bad faith or unfair dealing on the part of the insurer regardless of the outcome of the suit.

Christian, 577 P.2d at 905.

The simple presence of a legitimate dispute does not necessarily end the inquiry. Instead, it shifts the burden to the insured to present additional evidence of bad faith. Most commonly, the insured asserts an insurer's failure to "conduct an investigation reasonably appropriate under the circumstances." Buzzard v. Farmers Ins. Co., 824 P.2d 1105, 1109 (Okla.1991). An investigation does not meet this standard if (1) the manner of investigation hints at a sham defense or otherwise suggests that material facts were overlooked, or (2) the insurer intentionally disregarded undisputed facts supporting the insured's claim. Oulds v. Principal Mut. Life Ins. Co., 6 F.3d 1431, 1442 (10th Cir.1993).

The "essence" of such a claim, though, is always the "insurer's unreasonable, bad-faith conduct." McCorkle v. Great Atl. Ins. Co., 637 P.2d 583, 587 (Okla.1981); accord Willis v. Midland Risk Ins. Co., 42 F.3d 607, 612 (10th Cir. 1994). "The decisive question is whether the insurer had a good faith belief, at the time its performance was requested, that it had justifiable reason for withholding payment under the policy." Buzzard, 824 P.2d at 1109 (quotation omitted). Put simply, "tort liability may be imposed only where there is a clear showing that the insurer unreasonably, and in bad faith, withholds payment of the claim of its insured." Christian, 577 P.2d at 905.

"When there is conflicting evidence from which different inferences may be drawn regarding the reasonableness of the insurer's conduct, then what is reasonable is always a question to be determined by the trier of fact ...." Peters v. Am. Income Life Ins., 77 P.3d 1090, 1097 (2002).

With these principles in mind, we now turn to the facts of this case.

a. Legitimate Dispute

Our first inquiry is whether Great American had a legitimate basis to deny coverage. The facts here allow two plausible interpretations of the events that transpired: (1) Sims perished in an unintended and tragic accident, as the jury found, or (2) Sims committed suicide. Great American presented ample evidence to suggest the latter. Foremost, the jury reviewed Mrs. Sims's sworn statement to the police, where she stated that her husband "mention[ed] driving off a cliff" and had become "angrier and angrier." The police officer who took this statement interpreted it to imply that Sims was not only enraged but suicidal. Indeed, Sims's apparent conduct lends credence to this interpretation. No skid marks or other evidence at the scene indicated any intent to avoid the accident. Nor did Sims take his routine precaution to avoid injury—he failed to fasten his seat belt. Finally, although he did not drive off a cliff per se, within moments of leaving his home, Sims did launch his vehicle off an incline that propelled him some 115 feet across a river bed.

Aside from this evidence, every official report listed suicide as the cause of death: the accident report, the medical examiner's report, and the death certificate. Even the official missing persons report indicated that Sims might be suicidal. While it may be true that these reports were primarily based on Mrs. Sims's sworn statement to the police, a statement she later recanted, this fact does not negate the conclusion reached in these reports for the purpose of determining whether a legitimate dispute existed. Notably, the conclusions in these reports were not made at the behest of Great American—the medical examiner and investigating officer had no connection with Great American. An insurer can reasonably rely on such evidence when making its preliminary decision to dispute coverage.

Given the evidence before Great American at the time a decision on payment was required, the evidence clearly demonstrates a legitimate dispute concerning coverage. Therefore, Mrs. Sims had to present additional evidence of bad faith to survive a motion for judgment as a matter of law.

b. Adequate Investigation

Mrs. Sims argues that Great American did not perform an investigation reasonably appropriate under the circumstances. Great American counters that a more thorough investigation would not have produced additional relevant information that would nullify the undisputed evidence of suicide.

Great American cites to our decision in Timberlake Construction Co. v. United States Fidelity & Guaranty Co. for this proposition:

[W]hen a bad faith claim is premised on inadequate investigation, the insured must make a showing that material facts were overlooked or that a more thorough investigation would have produced relevant information. [The insured] made no such showing. For example, while [the insured's] expert, testified that [the insurer] breached the duty of good faith by not questioning [the insured's] representatives [concerning the dispositive issue], [the representatives'] views on this subject would not have changed the underlying facts already known to [the insurer], facts from which [the insurer] was entitled to form a reasonable belief [about the dispositive issue].

71 F.3d at 345 (emphasis added).

Extending this logic here, Great American asserts that it conducted an investigation reasonably appropriate under the circumstances and that no amount of additional evidence would have changed the underlying facts. It overlooked no material facts; it did not intentionally disregard undisputed facts in favor of the insured; and any additional facts could not have changed the conclusions drawn in the various state officials' reports. We agree.

The key failure, according to Mrs. Sims, is the lack of any investigation into motive. In particular, she argues, Great American should have explored this issue with her and her daughter.18 To be sure, Great American was aware the Sims family adamantly rejected any assertion that Sims committed suicide. Great American also knew that Sims's medical history did not reveal any evidence of depression and that he was drunk when he left in his car that evening. And it is hardly necessary to concede that committing suicide over a messy kitchen is improbable.

Yet, the operative inquiry is whether additional questioning of Mrs. Sims regarding a motive and questioning her daughter for the first time would have led to new or significant information. We think not. In the first place, Mrs. Sims and her daughter had every reason to contest a finding of suicide: a $300,000 life insurance benefit hung in the balance. Moreover, Mrs. Sims told police her husband mentioned "driving off a cliff." It is true she told Broyles that Sims did not commit suicide, and Broyles did not ask for further clarification. But even if it had, only a few additional questions could have been asked, and these would have revealed nothing new.

Great American could have further investigated by asking Mrs. Sims about the obvious inconsistency between her sworn police statement and her statements to Broyles. Yet this inquiry would not have changed the underlying facts upon which Great American was entitled to rely: the missing persons report, the death certificate, the medical examiner's report, and the accident report. See Timberlake, 71 F.3d at 345.

Mrs. Sims also contends that the fact an officer added language to a police form she signed in blank should have changed the underlying facts already known to Great American. But the officer testified that he interpreted her body language and her statement to suggest Sims possessed a suicidal state of mind. Just as importantly, this fact does not present anything new, as Mrs. Sims repeatedly told Broyles she did not think her husband was suicidal. We cannot draw from this alleged failure any suggestion that Broyles or Great American conducted a superficial investigation.

In sum, we conclude that the district court improperly submitted the issue of bad faith to the jury and reverse the jury verdict awarding damages on this claim.

Great American argues that the evidence was also insufficient to support a punitive damages verdict. Because a finding of punitive damages necessarily entails a finding of bad faith, Willis, 42 F.3d at 614-15, and because we find the evidence insufficient to support bad faith, we reverse the punitive damages verdict.

We note that even were we to have left the bad faith verdict intact, punitive damages still would have been inappropriate. Punitive damages are awarded "only when the evidence plainly shows oppression, fraud, malice, or gross negligence." McLaughlin v. Nat'l Ben. Life Ins. Co., 772 P.2d 383, 386 (Okla.1988) (quotation omitted). Stated differently, an insurer must "recklessly disregard" or "intentionally and with malice breach[] its duty to deal fairly and act in good faith with its insured." Okla. Stat. tit. 23, § 9.1(B), (D) (2005); see Badillo v. Mid Century Ins. Co., 121 P.3d 1080, 1106 (Okla.2005).

The evidence in this case plainly does not support this conclusion. In addition to the evidence supporting its claim denial, Great American did not engage in conduct that has supported punitive damages in other cases. It did not falsify any documents. See, e.g., Timmons v. Royal Globe Ins. Co., 653 P.2d 907, 917 (Okla. 1982). Nor did it offer to settle the claim if Mrs. Sims fired her counsel. See, e.g., id. Great American did not simply deny coverage based on skeletal information. It hired an experienced, independent claim investigation agency to review the evidence and interview witnesses. The investigator reviewed Sims's medical and pharmaceutical history, interviewed numerous individuals with knowledge of the accident, and obtained relevant documentation from government officials. See, e.g., Buzzard, 824 P.2d at 1115 ("Farmers was notified of the claim the day after the accident but failed to conduct any investigation.") (emphasis added).

Great American remained steadfast in its reasoning for denying the claim. It never argued that the denial was appropriate for any reason other than the alleged suicide. See, e.g., id. ("At trial, [the insurer] raised the issue of [the insured's] negligence, asserting that denial of the claim was based on [this negligence]. However, nowhere in [the insurer's] file was comparative negligence mentioned as a reason for denial."). Finally, Great American did not attempt to conceal the identity of its investigator or obtain confidential information without Mrs. Sims's permission. See, e.g., Timmons, 653 P.2d at 917. In short, because we find that the evidence does not support a finding of bad faith, we also cannot affirm a finding of punitive damages. Even were we to have affirmed the jury's finding on bad faith, Great American's conduct does not demonstrate the elements of fraud or evil intent to which punitive damages are directed. We therefore find no legally sufficient evidentiary basis for a reasonable jury to award punitive damages.

Great American finally contends that the district court improperly excluded its accident reconstruction expert as being untimely disclosed. Federal Rule of Civil Procedure 26(a)(2)(C) requires that

disclosures shall be made at the times and in the sequence directed by the court. In the absence of other directions from the court or stipulation by the parties, the disclosures shall be made at least 90 days before the trial date or the date the case is to be ready for trial ....

(emphasis added). Here, Great American disclosed its expert more than 90 days before the trial actually began. Based on this fact, Great American urges that its disclosure was timely because the trial date is the actual, as opposed to the scheduled, date of the trial.

We find Great American's disclosure untimely in either case. Rule 26 requires disclosure 90 days prior to the trial date unless otherwise directed by the court. Because the date of disclosure is initially determined per court order, the district court's scheduling orders are dispositive. Here, the court initially scheduled witness lists to be disclosed by March 7, 2003. The court amended this order twice. The final scheduling order set the date for expert disclosures for June 6, 2003. Great American filed a final witness list on that date without including the accident reconstruction expert. Despite ample opportunity to request an amendment to the scheduling order,19 Great American never did so, instead opting to file an Amended Witness List on November 12, 2003, five months past the original deadline.

Because Great American did not comply with the court's scheduling order, the district court properly excluded Great American's accident reconstruction expert as untimely disclosed.

Accordingly, we REVERSE the district court's denial of judgment as a matter of law on both the bad faith and punitive damages claims. We otherwise AFFIRM.

HARTZ, Circuit Judge concurring/dissenting.

I join Judge Tymkovich's impressive opinion except in two respects. First, I would not recognize any restrictions on the validity of Fed.R.Evid. 401 in diversity cases. Second, I would not hold that it was harmless error to exclude the seatbelt evidence.

It seems to me that my view of the validity of Rule 401 follows inexorably from the logic of the opinion itself. The opinion convincingly establishes that Rule 401 is not subject to the Rules of Decision Act or the Rules Enabling Act. And Stewart Organization, Inc. v. Ricoh Corp., 487 U.S. 22, 108 S. Ct. 2239, 101 L. Ed. 2d 22 (1988), imposes no limits on Rule 401 because it is "capable of classification as a procedural rule," id. at 32, 108 S. Ct. 2239. Indeed, it is a quintessential procedural rule.

This is not to say that Rule 401 can be applied in diversity cases without reference to state substantive law. Far from it. State law governs the plaintiff's cause of action and the defendant's defenses. The trial court must look to that law when deciding whether evidence meets Rule 401's definition of relevant evidence: "evidence having any tendency to make the existence of any fact that is of consequence to the determination of the action more probable or less probable than it would be without the evidence." If state law renders evidence of no consequence to resolution of the claim, then the evidence is irrelevant under Rule 401.

For example, if Oklahoma law says that failure to wear a seatbelt is not negligence for the purpose of determining a plaintiff's comparative fault in a tort case, then evidence of such failure is not relevant under Rule 401 when offered to prove such comparative fault. Similarly, if, under a state parol-evidence rule, the contract between two parties does not incorporate oral agreements that contradict the terms of the written contract, then evidence of the oral agreement is not relevant under Rule 401. Thus, in a sense, application of Rule 401 in a diversity case requires examining whether a state rule of evidence is one of procedure or of substance. If the state rule is one of substance because it determines that a fact (such as seatbelt use) is or is not, in the words of Rule 401, "of consequence to the determination of the action," then the state rule governs the application of Rule 401. If, however, the state rule is "substantive" only in some other sense, it does not affect application of Rule 401.

The approach that I prefer leads to the same conclusions reached by the majority opinion regarding the admissibility of evidence in this case. Perhaps application of the majority's formulation will never differ from what I conceive to be the proper rule. Future cases will tell. I do, however, think that a virtue of my approach (besides being, in my view, correct) is that it makes the trial judge's task more straightforward. Just apply Rule 401 on its own terms, and the ruling will be correct.

Turning now to harmless error, the majority and I agree that the district court erred in excluding the evidence that Mr. Sims had a habit of fastening his seatbelt but had not done so when he drove to his death. In what I view as a close case, this significant evidence could have changed the result. I would therefore reverse the breach-of-contract judgment and remand for a new trial.

My chief disagreement with the majority opinion on this point is not, however, its conclusion. Reasonable people can differ on whether the failure to admit the seatbelt evidence was harmless error. My particular concern is one aspect of the majority's analysis-the reference to the jury's verdicts on bad faith and punitive damages. The majority reasons that this jury must not have found the suicide issue to be a close call because it proceeded to rule against the insurer on bad faith and punitive damages. Because, as the majority opinion notes, the general rule is that one looks at the specific jury in assessing whether error is harmless, the majority's conclusion appears to follow. But I am not persuaded. To be sure, the jury did find bad faith and award punitive damages. On appeal, however, we set aside both findings for lack of evidence. In other words, the jury's fact finding was unreasonable. It seems peculiar to base our harmless-error ruling on a jury's unreasonable, irrational finding on another issue. In this circumstance, I think the proper approach is not to look at this particular jury (or at least this jury's findings on bad faith and punitive damages) in assessing whether error was harmless but to consider what a hypothetical rational jury would do.