Appellant Paul Forrest appeals the District Court's entry of final judgment and its denial of his motion for a new trial following a jury verdict in favor of appellee Beloit Corporation ("Beloit") in a products liability action initiated by Forrest.1 The lawsuit arises from an accident at the paper mill where Forrest was employed, in which his arm became stuck between two multi-ton rollers manufactured by Beloit, resulting in severe and permanent injuries. Forrest sued Beloit, advancing theories of negligence and strict liability under Pennsylvania law. The jury returned a special verdict in favor of Beloit, in which the jury found that Beloit's "Gloss Calender" machine was not defective, and that Beloit was not negligent in connection with the design or manufacture of the Gloss Calender. The special verdict form also addressed causality, with the jury indicating that the actions of Forrest's employer (Jefferson-Smurfit Corporation) constituted intervening forces that actively operated to cause Forrest's accident, and that these actions were so extraordinary Beloit could not reasonably have foreseen them. After the verdict, Forrest moved for a new trial. His motion was denied, and the District Court entered final judgment in favor of Beloit.

Forrest raises five issues. First, Forrest, who is African-American, presents a Batson challenge, arguing that the District Court abused its discretion in determining that the defense had proffered race-neutral reasons for striking two African-American jurors. Second, Forrest argues that counsel for Beloit engaged in "professional misconduct" in a manner that improperly influenced the jury's verdict. Third, Forrest argues that the jury's verdict was "tainted" as a result of questions and testimony relating to negligence and alleged OSHA violations purportedly committed by Jefferson-Smurfit. Fourth, Forrest argues that the District Court abused its discretion in permitting testimony concerning the alleged absence of prior accidents involving the Gloss Calender that crushed Forrest's arm. Fifth, Forrest argues that the District Court erred by permitting Beloit's expert to testify whether the presence of a guard on the Gloss Calender would have prevented Forrest's accident.

We will reverse the judgment of the District Court and remand for a new trial. While the majority of Forrest's challenges either lack merit or were not properly preserved, we believe Forrest argues correctly that the District Court abused its discretion by permitting Beloit to adduce testimony from two paper mill employees concerning the alleged absence of prior accidents involving the Gloss Calender on which Forrest was injured. The issue of the admissibility of evidence concerning the absence of prior accidents presents recurring difficulties in product liability cases, and this Court has yet to address this issue in the context of the Federal Rules of Evidence. After disposing of Forrest's other arguments, we take this opportunity to provide the district courts with guidance concerning the foundation that must be laid by a product liability defendant who seeks to introduce testimony concerning the non-occurrence of prior accidents.

Forrest's underlying lawsuit arises out of injuries he suffered on November 30, 1999, during the course of his employment at a paper mill operated by Jefferson-Smurfit. Forrest suffered his injuries while trying to clear a paper jam in an eighty-to one-hundred yard line of machines that transform wood pulp slurry into large rolls of dry paper. At the dry end of the line, the paper is run through two sets of calenders, or "dry stacks," which are large rotating rollers that feed the Gloss Calender. As the paper is propelled from the dry stacks towards the Gloss Calender, it first passes under an "air shower" and then over a lead-in roller known as a "Mount Hope roll." The air shower and Mount Hope roll were not part of the original Gloss Calender when it was designed and manufactured by Beloit in 1963. The Gloss Calender itself is an additional set of multi-ton rollers, consisting of a top roll, called the "gloss roll" or "dryer roll," and a lower roll called the "pressure roll" or "mate roll."

William Brody, Forrest's crew supervisor and a seventeen-year employee of Jefferson-Smurfit, testified that paper is generally run through the Gloss Calender regardless of whether gloss is applied, because the Gloss Calender rolls smooth the paper and support it as it moves toward the cutter at the end of the line. The opening between the Gloss Calender's two rollers is referred to as a "nip"; the size of the nip may vary depending upon whether gloss is being applied. Trial testimony indicated that Beloit's original design for the Gloss Calender called for the opening between the two rolls to be approximately eight feet, five inches off of the floor. Beloit's former chief engineer, George Wong, also testified that the Gloss Calender was originally designed to be threaded with the user standing on the floor. However, testimony from multiple Jefferson-Smurfit employees indicated that a different procedure was employed during paper breaks. In these situations, an employee would climb a set of steps located near one of the dry stacks, and would lean over the air shower and manually feed the paper through the Gloss Calender rolls to an employee waiting on the other side.

Forrest's accident occurred on November 30, 1999. It is not clear from the record whether at the time of the accident the Gloss Calender was applying gloss. There is no dispute, however, that a paper jam occurred, and that Forrest mounted the dry stack steps to feed a "tail" of paper through the Gloss Calender, in the manner described above. Forrest testified that he was working about eight to ten inches away from the nip. He testified that as he was attempting to feed the paper, his hand got pulled into the rollers, after which he had no further recollection of what occurred. Testimony from other witnesses present at the time showed that when Forrest's arm was caught between the two Gloss Calender rollers, the entire paper production line was shut down. The fire department and Forrest's co-workers eventually extricated Forrest after removing the top Gloss Calender roll. Forrest suffered severe and permanent injuries as a result of the accident.

Forrest sued Beloit, advancing theories of strict liability and negligence under Pennsylvania law. Two of Forrest's pretrial motions in limine relate to issues presented in this appeal. Forrest's first motion in limine sought to exclude references at trial to (1) alleged negligence on the part of Jefferson-Smurfit; (2) Jefferson-Smurfit's alleged violations of or non-compliance with OSHA standards and regulations; and (3) any OSHA investigations, proceedings, findings, reports or adjudications. Forrest's second motion in limine sought to exclude all references at trial to the alleged absence of prior accidents involving Beloit's Gloss Calender machines, including the Gloss Calender on which Forrest suffered his injuries. Forrest argued that Beloit had failed to establish an adequate foundation for the admissibility of such evidence, given that Beloit's witnesses admitted during deposition testimony that they were unaware of any databases or incident logs used by Beloit to track whether users of the Gloss Calender or other similar Beloit machines suffered injuries in circumstances similar to those surrounding Forrest's accident. Beloit responded by arguing that evidence reflecting the absence of prior accidents involving the Gloss Calender was admissible on the contested issue of causation. Beloit also asserted that it would first lay an adequate foundation as required under the Federal Rules of Evidence before introducing such testimony. The District Court denied Forrest's motions in limine, while preserving Forrest's right to raise his evidentiary objections in context at trial.2

Jury selection commenced on January 14, 2004, and trial ended on February 9, 2004, when the jury returned a defense verdict. The special verdict form reflects four specific findings. The jury found: (1) that the Gloss Calender machine was not defectively designed in 1963; (2) that Beloit was not negligent in its design, manufacture, or sale of the Gloss Calender machine in 1963; (3) that Forrest's employer, Jefferson-Smurfit, had taken intervening actions that actively operated to cause Forrest's accident; and (4) that these actions were so extraordinary they could not reasonably have been foreseen by Beloit. Following the jury's verdict, Forrest moved for a new trial pursuant to Rule 59(a) of the Federal Rules of Civil Procedure. Forrest also filed a supplemental memorandum regarding his Batson challenge. On April 15, 2004, the District Court denied Forrest's motion for a new trial and again rejected his Batson challenge. This appeal followed.

The District Court had subject matter jurisdiction pursuant to 28 U.S.C. § 1332(a) and 28 U.S.C. § 1441(a). We have jurisdiction under 28 U.S.C. § 1291.

With respect to Forrest's Batson challenge, the District Court's finding concerning the absence of intentional discrimination is reviewed for clear error. See United States v. Casper, 956 F.2d 416, 419 (3d Cir.1992) (citing Batson v. Kentucky, 476 U.S. 79, 98 n. 21, 106 S. Ct. 1712, 90 L. Ed. 2d 69 (1986)). We review the District's Court's allegedly inadequate response to supposed attorney misconduct for an abuse of discretion. See Wagner v. Fair Acres Geriatric Center, 49 F.3d 1002, 1017 (3d Cir.1995). The District Court's determinations concerning the admissibility of evidence are reviewed for an abuse of discretion as well. See In re Merritt Logan, Inc. v. Fleming Companies, 901 F.2d 349, 359 (3d Cir.1990). An abuse of discretion arises where the District Court's decision "rests upon a clearly erroneous finding of fact, errant conclusion of law or an improper application of law to fact." Oddi v. Ford Motor Co., 234 F.3d 136, 146 (3d Cir.2000).

To the extent an evidentiary issue turns on the interpretation of a Federal Rule of Evidence, rather than the mere application of the rule, our review is plenary. See In re Paoli R.R. Yard PCB Litigation, 35 F.3d 717, 749 (3d Cir.1994). Likewise, the propriety of the District Court's interpretations of substantive state law are subject to plenary review. See Waldorf v. Shuta, 896 F.2d 723, 728 (3d Cir.1990). Where an appellant's arguments for a new trial implicate questions of fact, we view "all the evidence and inferences reasonably drawn therefrom in the light most favorable to the party with the verdict." See Marino v. Ballestas, 749 F.2d 162, 167 (3d Cir.1984).

Even if Forrest establishes an error by the District Court, Forrest must also show that the error was prejudicial. See 28 U.S.C. § 2111; McQueeney v. Wilmington Trust Co., 779 F.2d 916, 924 (3d Cir.1985). An error will be deemed harmless only if it is "highly probable" that the error did not affect the outcome of the case. See McQueeney, 779 F.2d at 924. Harmless error analysis, however, does not apply to Forrest's Batson challenge. Ramseur v. Beyer, 983 F.2d 1215, 1225 n. 6 (3d Cir.1992) (en banc).

In Batson, the Supreme Court held that the 14th Amendment's equal protection clause barred the use of peremptory challenges to exclude prospective jurors on the basis of race. The Supreme Court extended Batson's rule to civil cases in Edmonson v. Leesville Concrete Co., Inc., 500 U.S. 614, 631, 111 S. Ct. 2077, 114 L. Ed. 2d 660 (1991). Forrest challenges Beloit's use of its peremptory challenges here, arguing that Beloit improperly used two of its challenges to exclude African-American jurors on the basis of their race. The District Court ruled that Forrest had failed to satisfy the third prong of the Batson test, which requires that the District Court conduct an independent assessment concerning whether the striking party has advanced a non-pretextual, race-neutral reason for the challenge. See Hernandez v. New York, 500 U.S. 352, 359-60, 111 S. Ct. 1859, 114 L. Ed. 2d 395 (1991).

Notably, the reason advanced by the striking party in support of the peremptory challenge need not be especially persuasive from a tactical standpoint. See Purkett v. Elem, 514 U.S. 765, 767-68, 115 S. Ct. 1769, 131 L. Ed. 2d 834 (1995). Instead, a race-neutral explanation is simply one that is based on "something other than the race of the juror" and is free of discriminatory animus. See Hernandez, 500 U.S. at 360, 111 S. Ct. 1859. The trial judge must then evaluate whether the reason proffered by the striking party is indeed race-neutral, and also whether it is non-pretextual, in the sense that it is not being used merely to cover the striking party's discrimination. See United States v. Casper, 956 F.2d 416, 419 (3d Cir.1992). Because the trial court's evaluation turns in large part upon the credibility and demeanor of the attorney exercising the challenge, see id. at 419, the trial judge's determination is afforded considerable deference, and "will not be reversed unless it is completely devoid of minimum evidentiary support displaying some hue of credibility, . . . or bears no rational relationship to the supportive evidence." United States v. Milan, 304 F.3d 273, 281 (3d Cir.2002).

We find that the District Court did not abuse its discretion in determining that Beloit's attorney advanced non-pretextual, race-neutral reasons in support of Beloit's challenge of two African-American jurors. Forrest seeks to vindicate his Batson challenge by focusing upon a comparison of characteristics possessed by the two stricken African-American jurors that are purportedly identical to characteristics possessed by white jurors who were not stricken. This approach is consistent with the mode of analysis we have embraced in evaluating prior Batson challenges. See, e.g., Holloway v. Horn, 355 F.3d 707, 724 (3d Cir.2004); Riley v. Taylor, 277 F.3d 261, 282 (3d Cir.2001) (en banc). However, Forrest's challenge fails on the merits, because the District Court reasonably determined, based on the record before it, that the reasons cited by Beloit in support of its challenges to the stricken African-American jurors were not reflected in equal measure in various white jurors who were not challenged.

The first African-American juror against whom Beloit allegedly exercised an improper challenge was Juror No. 38. Beloit's counsel indicated on the record that Juror 38 was struck because she was a nurse, and Beloit anticipated putting on testimony that would be critical of the wound care received by Forrest following his accident. Forrest complains that Juror 38 was never questioned concerning whether her occupation as a nurse would affect her ability to serve as a fair and impartial juror. However, Forrest cites no authority for the proposition that Beloit was required to make such an inquiry prior to exercising its peremptory challenge. Forrest also argues that Beloit's reliance on Juror 38's occupation as a nurse was pretextual, noting that a white female juror also employed as a nurse was not stricken. Forrest acknowledges, however, that the white juror was so far down the list that she was not seated on the jury in any event, and Beloit observes that this juror was "far enough down the list that it was unnecessary for the defense to use one of its peremptory strikes."3 On this record, we believe it is clear that the District Court did not abuse its discretion in concluding that Beloit offered an acceptable, race-neutral, non-pretextual reason for striking Juror 38.

The second African-American juror against whom Beloit allegedly exercised an improper challenge was Juror No. 29. Beloit's counsel indicated that Juror 29 was struck for a combination of two reasons: (a) she appeared inattentive and did not participate during voir dire other than to indicate that she preferred not to sit on the jury; and (b) she was from Philadelphia, and regardless of race Beloit was concerned that jurors from Philadelphia were more likely to award large verdicts than jurors from other parts of the Eastern District. Forrest argues that Beloit's cited reasons were pretextual, because it purportedly failed to strike a number of similarly situated white jurors. However, most of the jurors cited by Forrest shared neither of the two characteristics that Beloit cited as together prompting it to strike Juror 29. Five of the six jurors cited by Forrest were not Philadelphia residents, and contrary to Forrest's characterization, a number of these jurors participated actively in the voir dire.

Forrest notes that one white juror who was a Philadelphia resident indicated that he preferred not to sit on the jury. However, this juror participated actively in the voir dire, his statement concerning the preventability of workplace accidents provided a reasonable basis for Beloit to distinguish between this juror and Juror 29, notwithstanding that both were residents of Philadelphia. On this record, we cannot say that the District Court abused its discretion by accepting as non-pretextual Beloit's explanation concerning its basis for striking Juror 29.

D. Forrest's Allegations Of "Attorney Misconduct"

Forrest argues for a new trial on the basis of alleged attorney misconduct by counsel for Beloit.4 The arguments grouped by Forrest under the heading of attorney misconduct cover a wide range of issues, including concerns regarding evidentiary rulings, defense counsel's facial expressions, an in-court demonstration performed by defense counsel, and the tenor and content of defense counsel's questions to witnesses and statements in closing arguments. We review the District Court's decisions concerning alleged attorney misconduct under an abuse of discretion standard. Due to his superior vantage point, the trial judge is entrusted with wide discretion in matters relating to the conduct of counsel during trial. See Greenleaf v. Garlock, Inc., 174 F.3d 352, 363 (3d Cir.1999) (citing Fineman v. Armstrong World Indus., Inc., 980 F.2d 171, 207 (3d Cir.1992)). Accordingly, under our deferential review, we will grant a new trial only where the allegedly improper statements or conduct make it "reasonably probable" that the verdict was influenced by the resulting prejudice. See Greenleaf, 174 F.3d at 363-64; Waldorf, 142 F.3d at 627-28; Greate Bay Hotel & Casino v. Tose, 34 F.3d 1227, 1236 (3d Cir.1994).

Our consideration of the record, coupled with our deferential standard of review, compels the conclusion that the District Court did not err in refusing to grant Forrest a new trial on the basis of the alleged attorney misconduct by counsel for Beloit. Forrest's scattered assertions concerning alleged "editorializing" by counsel for Beloit for the most part lack record support. In the one instance where Forrest requested a curative instruction, the District Court reasonably noted that the jury had heard the Court's admonitions with respect to editorializing, and that an additional instruction would serve only to highlight the questioned statements for the jury.5

We need not address Forrest's remaining allegations of attorney misconduct in detail, as we are remanding for a new trial. We note, however, that Forrest has identified certain actions of Beloit's trial counsel which may reasonably be questioned. In particular, we believe counsel for Beloit should not have invoked in his closing argument a prior courtroom demonstration in which he attempted to simulate Forrest's efforts to clear the paper jam on the night Forrest was injured. This attempted demonstration, involving a variety of poles and a ladder, was subject to repeated sustained objections, and the District Court eventually ordered Beloit's trial counsel to discontinue the demonstration.

Notwithstanding the District Court's ruling, counsel for Beloit invoked the demonstration repeatedly in his closing, arguing that it validated the defense's theory of causation. Two of the references to this demonstration occurred after the District Court had informed Forrest's trial counsel that it would not permit further objections during closing arguments. We recognize both the breadth of the District Court's discretion with respect to trial proceedings and the desirability of permitting each party to present its closing statement free from undue interruption. However, this freedom is not a license to flout a district court's earlier rulings restricting reliance on misleading courtroom demonstrations or inadmissible evidence. Although Beloit's counsel "crossed the line," we do not consider his conduct so severe as to warrant a new trial.

Forrest also seeks a new trial on the basis of the alleged improper introduction of evidence concerning OSHA standards by Beloit during the course of the trial. A product manufacturer in Pennsylvania has a non-delegable duty to provide a safe product. See Walton v. Avco Corp., 530 Pa. 568, 610 A.2d 454, 458 (1992). Thus, a manufacturer in a products liability action may not invoke industry or OSHA standards to argue that the plaintiff's employer, rather than the manufacturer, had the responsibility to provide the equipment or instructions necessary to make a product safe for its intended use. See Sheehan v. Cincinnati Shaper Co., 382 Pa.Super. 579, 555 A.2d 1352, 1355 (1989); Majdic v. Cincinnati Machine Co., 370 Pa.Super. 611, 537 A.2d 334, 336-38 (1988). However, this rule is of limited applicability here, because the District Court did not admit OSHA and industry standard evidence for such purposes. Indeed, the first OSHA reference cited by Forrest occurred in a question directed to Forrest's expert, Widas, during cross-examination. Forrest objected, a lengthy sidebar ensued, and the District Court directed Beloit's counsel to proceed without referencing Jefferson-Smurfit's citation for an OSHA violation in connection with Forrest's accident. The District Court also agreed to strike from the record the OSHA references that had occurred thus far. Notably, Forrest did not move for a mistrial. We see no error in the District Court's actions, and thus Forrest's appeal with respect to this issue lacks merit.6

Forrest argues that the District Court improperly permitted testimony from Beloit's biomechanical engineering expert, Kelly Kennett. Forrest first argues that Kennett was improperly permitted to testify concerning the ultimate issue in the case. Second, Forrest maintains that Beloit failed to establish an adequate foundation for Kennett's testimony concerning whether the presence of a particular type of guard would have prevented Forrest's accident. Both of Forrest's objections lack merit.

As Beloit correctly points out, Kennett was admitted to testify as an expert witness, and under FRE 704 an expert witness may offer testimony concerning the ultimate issue in the case. See Salas v. Wang, 846 F.2d 897, 905 (3d Cir.1988). Moreover, Beloit is also correct that the question of whether the presence of a guard would have prevented Forrest's injury is a question of fact, and is distinct from the ultimate issue of whether the Gloss Calender was defectively or negligently designed and manufactured. The admissibility of expert opinion testimony with respect to such issues is well established. See Wilburn v. Maritrans GP, Inc., 139 F.3d 350, 356 (3d Cir.1998).

Kennett's testimony focused upon the physical posture that Forrest allegedly must have been in to access the Gloss Calender nip, and also whether the presence of a particular guard would have prevented Forrest's injury. Forrest argues that Kennett failed to lay an adequate foundation for this testimony. However, Kennett's testimony set forth his methodology and described at some length the various measurements relevant to his calculations. This foundation adequately supported Kennett's expert testimony, and thus the District Court did not abuse its discretion in permitting Kennett to testify.

G. Evidence Concerning The Absence Of Prior Gloss Calender Accidents

Forrest argues that the District Court erred by permitting Beloit to introduce evidence concerning the alleged absence of prior accidents involving the Gloss Calender at the Jefferson-Smurfit mill. Beloit introduced this evidence through testimony extracted on cross-examination from former Jefferson-Smurfit employees William Brody and Edward Marshall, who had been employed at Jefferson-Smurfit (and its corporate predecessor CCA) for seventeen years and thirty-five years, respectively. They testified that the way Forrest attempted to thread the Gloss Calender on the night of the accident was the same as that used for years by other employees. Both Brody and Marshall also indicated that they were unaware of any prior similar accidents involving the Gloss Calender during their years at Jefferson-Smurfit. Beloit invoked this testimony in its closing, arguing that "as far as the evidence is concerned, the only accident we know of, in thirty-six years, on the Gloss Calender was Mr. Forrest's."

The foregoing testimony came in over Forrest's repeated objections, including a pretrial motion in limine. Forrest's objections centered on Beloit's alleged failure to establish an adequate foundation for introducing this testimony concerning the alleged absence of prior accidents involving the Gloss Calender at the Jefferson-Smurfit mill. Forrest noted that George Wong, Beloit's former chief engineer, had admitted in his deposition that Beloit kept no records relating to either safety complaints by Beloit customers or past accidents involving Beloit's Gloss Calender machines. Invoking Federal Rules of Evidence 402 and 403, Forrest argued that the lack of records precluded Beloit from satisfying the foundation-laying requirement traditionally imposed on a product liability defendant seeking to introduce testimony concerning the alleged absence of prior accidents involving its products. Forrest's motion in limine argued that "[b]ecause Beloit cannot establish a foundation for the admissibility of evidence concerning an absence of prior substantially similar accidents, any reference to such alleged evidence. . . would be unfairly prejudicial to Forrest[.]"

To assess Forrest's challenge to the disputed evidence, we must first determine the applicable law. The parties and the District Court focused primarily on Pennsylvania law concerning this issue, and in particular, on the decision of the Pennsylvania Supreme Court in Spino v. John S. Tilley Ladder Co., 548 Pa. 286, 696 A.2d 1169 (1997). While the well-reasoned decision in Spino provides useful guidance, the question presented is governed by federal rather than state law. The admissibility of the evidence ultimately turns on a balancing of its probative value versus its prejudicial effect, and we have held that in a federal court the Federal Rules of Evidence govern procedural issues of this nature. See, e.g., Diehl v. Blaw-Knox, 360 F.3d 426, 431 n. 3 (3d Cir.2004) (stating in product liability diversity action governed by Pennsylvania law that "assessment of the dangers of unfair prejudice and confusion of the issues are procedural matters that govern in a federal court notwithstanding a state policy to the contrary") (emphasis added); Kelly v. Crown Equipment Co., 970 F.2d 1273, 1277-78 (3d Cir.1992) (noting that relevancy provision in Federal Rules of Evidence is "arguably procedural" and therefore governs in diversity action notwithstanding contrary Pennsylvania law); Espeaignnette v. Gene Tierney Co., 43 F.3d 1, 9 (1st Cir.1994) (federal rather than state law governs admissibility of "no prior accident" evidence in a diversity action).7

Under the Federal Rules of Evidence, subject to certain limitations, all evidence is admissible if it is relevant, i.e., if it tends to make the existence or nonexistence of a disputed material fact more probable than it would be without that evidence. See Fed.R.Evid. 401, 402. Pursuant to Rule 403 of the Federal Rules of Evidence, a district court may nonetheless exclude relevant evidence if the probative value of the evidence is "substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury, or by considerations of undue delay, waste of time, or needless presentation of cumulative evidence." Fed.R.Evid. 403. Rule 403 is an "`umbrella rule' spanning the whole of the Federal Rules of Evidence," and as such trial judges must apply Rule 403 "in tandem with other Federal Rules under which evidence would be admissible." See Coleman v. Home Depot, Inc., 306 F.3d 1333, 1343 (3d Cir.2002). Where, as here, a district court fails explicitly to articulate the Rule 403 balancing, "we either decide the trial court implicitly performed the required balance; or, if we decide the trial court did not, we undertake to perform the balance ourselves." Ansell v. Green Acres Contracting Co., 347 F.3d 515, 525 (3d Cir.2003) (quoting Glass v. Philadelphia Electric Co., 34 F.3d 188, 192 (3d Cir.1994)). In sum, "Rule 403 recognizes that a cost/benefit analysis must be employed to determine whether or not to admit evidence; relevance alone does not ensure its admissibility." Coleman, 306 F.3d at 1343. However, "there is a strong presumption that relevant evidence should be admitted, and thus for exclusion under Rule 403 to be justified, the probative value of evidence must be `substantially outweighed' by the problems in admitting it." Id. at 1343-44.

Federal and state courts addressing the admissibility of evidence concerning the absence of prior accidents have recognized that the probative value of such evidence is determined in large measure by the foundation laid by the offering party. In Espeaignnette, the First Circuit observed that as a general rule, "evidence of the absence of prior accidents may not be admitted unless the offering party first establishes that the `lack of accidents was in regard to products that are substantially identical to the one at issue and used in settings and circumstances sufficiently similar to those surrounding the machine at the time of the accident.'" 43 F.3d at 10 (quoting Klonowski v. International Armament Corp., 17 F.3d 992, 996 (7th Cir.1994)). Accordingly, most courts admitting evidence of the absence of prior accidents in product liability cases have done so only where the testifying witness, usually an employee of the product manufacturer, has testified that (a) a significant number of substantially identical products have been used in similar circumstances over a period of time; (b) the witness would likely be aware of prior accidents involving these products; and (c) to the witness's knowledge, no such prior accidents have occurred. See, e.g., Pandit v. American Honda Motor Co., 82 F.3d 376, 380-81 (10th Cir.1996) (permitting testimony where allegedly defective feature had been included in nearly 1.9 million automobiles over eight-year period, and these other automobiles had been used in substantially similar settings and circumstances); Espeaignnette, 43 F.3d at 10 (permitting testimony where manufacturer's president testified that eighty-seven similar products had been sold in the previous fifteen years, and that as president of company any prior accidents or claims involving these products would have come to his attention); Bilski v. Scientific Atlanta, 964 F.2d 697, 700 (7th Cir.1992) (permitting testimony where defendant's expert testified that 4,000 identical satellite dishes had been sold and were used under circumstances similar to those at plaintiff's place of employment); Hines v. Joy Mfg. Co., 850 F.2d 1146, 1154 (6th Cir.1988) (permitting testimony where defendant's expert testified that original design of product dated back to the 1950s and that 200 substantially identical units had been sold by defendant); Spino, 696 A.2d at 1174 (permitting testimony where defendant's president indicated that over 100,000 identical ladders had been sold, and that company claims log did not reveal the existence of any prior accidents involving the allegedly defective ladder).8

With respect to the conceptual underpinnings of this foundation requirement, Espeaignnette stated that it was unclear "[w]hether such preliminary requirements are aimed at preventing the admission of irrelevant evidence under Rule 402, excluding relevant evidence that is unfairly prejudicial and confusing under Rule 403, or both. . . ." Id. We think the foundation requirement discussed in these cases is best described as a tool meant to aid in the balancing inquiry under Rule 403 or its state analog. There is little doubt that as a general matter evidence concerning the absence of prior accidents can satisfy the relevance threshold established by Rule 402. Courts have indicated that such evidence may be relevant to show (1) the absence of the alleged defect; (2) the lack of a causal relationship between the injury and the defect or condition charged; and (3) the nonexistence of an unduly dangerous situation. See, e.g., Pandit, 82 F.3d at 380; Espeaignnette, 43 F.3d at 9-10; Hines, 850 F.2d at 1152. Notwithstanding the potential relevance of such evidence under Rule 402, its probative value must be carefully balanced, pursuant to Rule 403, against its possible prejudicial effect. Testimony concerning an alleged absence of prior accidents, if offered without a proper foundation, can create risks of unfair prejudice that may substantially outweigh whatever probative value the evidence otherwise has. Thus, courts assessing the admissibility of such evidence emphasize the contextual nature of the inquiry, which turns upon the facts and circumstances of each particular case. See Espeaignnette, 43 F.3d at 10; Walker, 487 F.2d at 599; Spino, 696 A.2d at 1173-74; Jones v. Pak-Mor Mfg. Co., 145 Ariz. 121, 700 P.2d 819, 824-25 (1985).

The importance of the foundation requirement is underscored by the potential for unfair prejudice that may result from such evidence. The Arizona Supreme Court's thorough opinion in Jones summarized the concerns at issue. First, the mere fact that a witness does not know of any prior accidents does not prove that no such accidents occurred. See Jones, 700 P.2d at 824. Second, generalized assertions concerning an alleged absence of accidents over an extended period of time can be directly rebutted only with specific evidence of prior occurrences, but such evidence may be difficult or impossible for a plaintiff to obtain in cases where the defendant has not kept records concerning the safety history of its products. See id. at 824-26. Third, the absence of prior accidents may simply mean that the plaintiff was the first to be injured; there is always a first victim. See id. at 825; Spino, 696 A.2d at 1173. Fourth, testimony concerning the absence of prior accidents "does not tell us how many near-accidents, nor how many fortuitous escapes from injury, may have occurred[.]" See Jones, 700 P.2d at 826.

This fourth concern is especially salient in product liability cases arising under Pennsylvania law, which deems a product defective if it "left the supplier's control lacking any element necessary to make it safe for its intended use." Lewis v. Coffing Hoist Div., Duff-Norton Co., 515 Pa. 334, 528 A.2d 590, 593 (1987) (quoting Azzarello v. Black Bros. Co., 480 Pa. 547, 391 A.2d 1020, 1027 (1978)). The Pennsylvania Supreme Court has stated that "products are to be evaluated at the time of distribution when examining a claim of product defect." Duchess v. Langston Corp., 564 Pa. 529, 769 A.2d 1131, 1142 (2001). Pennsylvania's approach is reflected in the fact that risk-utility analysis concerning whether a product is unreasonably dangerous (which is required under § 402A of the Restatement (Second) of Torts) is performed by the trial judge rather than the jury. See Azzarello, 391 A.2d at 1026. "In answering this question a court is essentially making a social policy determination and acting as both a social philosopher and a risk-utility economic analyst." Riley v. Warren Mfg. Inc., 455 Pa.Super. 384, 688 A.2d 221, 224 (1997) (citing Fitzpatrick v. Madonna, 424 Pa.Super. 473, 623 A.2d 322, 324 (1993)). Thus, where the plaintiff has surmounted this initial hurdle and the case has reached the jury, the jury's focus is on the product in se, and specifically on whether the product as designed presents a potential danger to the intended user.

The nature of this inquiry is such that evidence of near-misses or fortuitous escapes would be highly probative of the existence of a danger, and thus of the existence of a defect. Such evidence, however, is by definition extremely difficult to obtain, if for no other reason than that a user who has fortuitously escaped injury may not even recognize that he was exposed to danger in the first place. Permitting a product liability defendant to introduce testimony concerning an alleged absence of prior accidents may thus create a misleading impression as to whether a defect exists, due to the potential inaccessibility of contrary probative evidence that would cast doubt upon the product's safety. It may also divert the jury's focus onto a balancing of the product's proven costs vis-a-vis its proven benefits, notwithstanding that this issue will already have been resolved in the plaintiff's favor by the trial judge's earlier risk-utility analysis.9

To summarize the applicable analytical framework, in federal court the admissibility of evidence concerning an absence of prior accidents is governed by federal law. The admissibility of such evidence turns on the facts and circumstances of each case. Testimony concerning an alleged absence of prior accidents will usually satisfy the relevance threshold established by Rule 402. Such testimony, however, by its very nature, raises significant concerns regarding unfair prejudice to the plaintiff, and these concerns are heightened in product liability cases arising under Pennsylvania law. District courts are required under Rule 403 to balance the probative value of such evidence against its likely prejudicial effect, but the evidence may not be excluded unless the unfair prejudice created by admitting the evidence would substantially outweigh its probative value. In an effort to ascertain probative value and minimize undue prejudice, other courts considering such evidence have consistently insisted that the offering party lay a proper foundation. In most cases the required foundation has involved three elements: (a) similarity — the defendant must show that the proffered testimony relates to substantially identical products used in similar circumstances; (b) breadth — the defendant must provide the court with information concerning the number of prior units sold and the extent of prior use; and (c) awareness — the defendant must show that it would likely have known of prior accidents had they occurred.

The facts and circumstances surrounding the disputed testimony at issue in this case present an uncommon scenario. Prior cases have usually involved a product liability defendant's attempt to introduce evidence concerning the absence of prior accidents through the testimony of its own witness, typically a corporate officer or an expert. Here, in contrast, Beloit sought to introduce safety history evidence by extracting testimony during the cross-examination of two witnesses who were long-time employees of the Jefferson-Smurfit paper mill. Beloit also restricted its questions to the safety history of the specific Gloss Calender that was installed at the Jefferson-Smurfit mill. This narrower focus was understandable, because Wong, Beloit's corporate designee, admitted in his deposition that he knew of no records or databases relating to either safety complaints by Beloit's customers or past accidents involving Beloit's Gloss Calender machines. Thus, any attempt by Beloit to introduce through its own witness a broad claim with respect to the safety history of Beloit's Gloss Calender machines would likely have been foreclosed by the witness's inability to show that he or she would have known of prior accidents had they occurred.

The question now before us is whether Beloit, by focusing solely upon the single Gloss Calender at the Jefferson-Smurfit mill, so diluted the probative value of the testimony in question as to render it inadmissible in light of the potential for unfair prejudice that inheres in all such testimony. We answer this question in the affirmative, and hold that the testimony should have been excluded pursuant to Rule 403.

We reach this conclusion for several reasons. Our primary concern is that notwithstanding the disputed testimony, we have no idea whether there were prior accidents involving Beloit's allegedly defective Gloss Calenders. The record is clear that Beloit designed and sold its Gloss Calenders to many customers over a period of several decades. Wong, who at one time personally led Beloit's Gloss Calender design group, testified that to his knowledge Beloit kept no records concerning whether injuries or accidents involving these Gloss Calenders might have occurred during the decades prior to Forrest's accident. The combination of (a) the existence of multiple other Beloit Gloss Calenders of similar or identical design; (b) the likely use of these Gloss Calenders in similar circumstances over a period of several decades; and (c) the absence of any evidence concerning the safety history of these other Gloss Calenders, leaves us with no reliable way to determine the probative value of what is essentially anecdotal testimony from two former Jefferson-Smurfit employees concerning a single Gloss Calender installed at a single mill. Thus, we can do little more than engage in rank speculation concerning the "probative value" side of the Rule 403 balancing equation.

The same uncertainty that hampers our ability to ascertain the probative value of the disputed testimony also undermines Forrest's ability to respond. Forrest could of course speculate that other accidents might have occurred on one or more of the Beloit Gloss Calenders used at other mills over the past forty years. Such speculation, however, is unlikely to have anywhere near the same effect on the jury when compared to the concrete testimony from two witnesses concerning the specific Gloss Calender involved in Forrest's accident.

The asymmetry in the persuasive force of the cross-examination testimony extracted by Beloit and the speculative nature of Forrest's potential response highlights two ways in which Forrest was unfairly prejudiced. First, Forrest's inability to address the issue in a more concrete fashion is traceable in large measure to Beloit's failure to maintain records concerning the safety history of its own products. Second, the advantage Beloit gains over Forrest in this situation is not primarily the result of the natural probative force of the disputed testimony; indeed, the disputed testimony leaves us no way of knowing whether the absence of prior accidents involving the Jefferson-Smurfit Gloss Calender was an aberration, as opposed to a typical example of industry experience with substantially identical Beloit Gloss Calenders. This problem is basically a variation of a general concern applicable to all similar evidence from which a jury is asked to draw a negative inference: Witnesses testify from limited knowledge, and the fact that a particular witness is unaware of prior accidents does not mean such accidents have not occurred. We believe that given these considerations, the potential harm Forrest suffered as a result of Beloit's reliance on the disputed testimony constitutes the sort of unfair prejudice that Rule 403 is meant to combat. See Coleman, 306 F.3d at 1343, n. 6.10

The disputed testimony at issue is also troubling in light of Rule 403's reference to "confusion of the issues" and "misleading the jury." Isolated testimony concerning the alleged safety history of the Gloss Calender on which Forrest was injured tends naturally to focus the jury's attention upon that specific Gloss Calender. This focus may lead the jury to generalize from the limited experience surrounding one Gloss Calender to a broader conclusion concerning the overall safety of Beloit's Gloss Calender design. Pennsylvania law, however, focuses on the design of the product in the abstract, rather than the safety history of a particular unit. See Duchess, 769 A.2d at 1142. Thus, to the extent an inference concerning the safety of a product's design can be drawn from a product's safety history, the reliability of such an inference is determined in large measure by the scope of the available safety history information. Here, of course, the information relied upon by Beloit does not cover all of Beloit's prior Gloss Calenders, or even a majority of them. Thus, to the extent this evidence could lead the jury to an inference concerning the overall safety of Beloit's Gloss Calender design, we cannot discount the possibility that the inference would be based on either false assumptions, unsupported speculation, or both.

All of the foregoing concerns with respect to possible unfair prejudice and jury confusion are in addition to the generally applicable concerns discussed earlier. Of particular significance is that the evidence concerning the absence of prior accidents does not account for "near accidents" and "fortuitous escapes." See Jones, 700 P.2d at 826. The risk of jury confusion and unfair prejudice arising as a result of this issue is especially acute under Pennsylvania law, where the jury's defect determination turns not upon a risk-utility analysis, but instead upon whether the product as designed lacks a necessary safety feature. See Lewis, 528 A.2d at 593.

In a risk-utility analysis, avoidance of accidents through extra care by product users, and post-purchase employer precautions such as additional safety training for workers, may affect the analysis of whether the product as designed was "unreasonably dangerous." See Surace v. Caterpillar, Inc., 111 F.3d 1039, 1046 (3d Cir.1997) (discussing risk-utility analysis under Pennsylvania law, and citing factors including "the user's ability to avoid danger by the exercise of care in the use of the product" and "the user's anticipated awareness of the dangers inherent in the product and their avoidability, because of . . . the existence of suitable warnings or instruction") (quoting Dambacher v. Mallis, 336 Pa.Super. 22, 485 A.2d 408, 423 n. 4 (1984)). These same factors, combined with good fortune, may also contribute to a lack of prior accidents involving the allegedly defective product. Consideration of such factors, however, is not within the province of the jury in Pennsylvania; instead, the jury is to focus on the design of the product as it existed when it left the supplier's control.11 See Duchess, 769 A.2d at 1142; Lewis, 528 A.2d at 590; Azzarello, 391 A.2d at 1027.

These characteristics of Pennsylvania law create a heightened risk that testimony concerning the alleged absence of prior accidents may confuse or mislead the jury. Under Pennsylvania law, the lack of a necessary safety feature when the product leaves the factory determines whether the product is defective. In arguing to the jury in a case governed by Pennsylvania law, a product liability defendant may not invoke an alleged absence of sufficiently severe or frequent injuries in support of an assertion that a product's social utility outweighs its otherwise defective design. There is a danger, however, that testimony concerning the alleged absence of prior accidents may tend to lead the jury towards forbidden inferences of this sort.

Moreover, as discussed above, the close relationship under Pennsylvania law between the existence of danger and the existence of a defect demonstrates that to the extent the absence of prior accidents is probative, the presence of prior near-accidents or fortuitous escapes is equally probative. Such evidence, however, is inherently difficult to obtain, and thus plaintiffs such as Forrest may be left at an unfair disadvantage, in that safety history testimony proffered by defendants such as Beloit may appear to be more probative than it actually is, but its shortcomings will not be fully exposed before the jury.

A number of the concerns set forth above are aspects of prejudice that arise from the unique facts of this case. Others, however, are more or less equally applicable to all evidence concerning an alleged absence of prior accidents. We note this to emphasize that the mere existence of potential unfair prejudice to the plaintiff does not in and of itself justify the exclusion of safety history evidence offered by a product liability defendant. Nonetheless, that such concerns are always present will invariably implicate the strength of the showing a defendant must make in terms of establishing probative value under Rule 403. We believe this is why other federal courts that have admitted such evidence have done so in the context of a foundational showing that incorporates the elements of similarity, breadth, and awareness.

The "breadth" aspect of the foundation need not always incorporate knowledge of the safety history of every unit of a particular product; there may be gaps in even the most thorough record-keeping system. Minor gaps can legitimately be said to go to the weight of the evidence, rather than its admissibility. In the present case, however, we are not dealing with disputed testimony predicated upon a solid foundation containing isolated gaps; we are dealing instead with a complete absence of records that Beloit has attempted to remedy using a small fragment of anecdotal testimony.

We also note that our reference to the breadth of a proponent's evidentiary foundation should not be taken as automatically barring evidence concerning an alleged absence of prior accidents in cases involving a one-of-a-kind product, or cases where only a small number of substantially identical units have been sold. In cases involving a unique, one-of-a-kind product, there is less concern that testimony regarding the safety history of the single unit will present a distorted picture of the overall risks associated with the product's design. In cases involving unique products or products with a small number of identical units, concerns regarding the plaintiff's access to contrary probative evidence are lessened as well, because the universe in which plaintiffs can seek such evidence is smaller and more manageable.12

Other concerns endemic to safety history evidence may still exist, but we do not purport today to create a categorical rule for all circumstances, instead leaving these issues in the first instance to the sound discretion of district judges. We are confident, however, that where Beloit has manufactured and sold multiple Gloss Calenders over a span of several decades, and where Beloit admits that it has compiled no information concerning the safety history of these Gloss Calenders, it was prejudicial error for the District Court to permit Beloit to extract anecdotal testimony concerning a single Gloss Calender, and then to use that testimony to argue in closing that "as far as the evidence is concerned, the only accident we know of in 36 years on the Gloss Calender was Mr. Forrest's."

The judgment of the District Court will be reversed and the case remanded to the District Court for a new trial consistent with this opinion.

ALITO, Circuit Judge, concurring in the judgment.

I agree that evidence of the absence of prior accidents involving the gloss calendar at the plant in question should not have been admitted and that the plaintiff is therefore entitled to a new trial. If we were not constrained by prior circuit law, I would join the majority in holding that the admission of such evidence should be analyzed under the Federal Rules of Evidence, but our court's decision in Greiner v. Volkswagenwerk Aktiengeselleschaft, 540 F.2d 85 (3d Cir.1976), points toward the application of state law. In addition, except for one passing reference in his District Court papers, the plaintiff's arguments at the trial level and on appeal focused exclusively on Pennsylvania law and in particular on the Pennsylvania Supreme Court's decision in Spino v. John S. Tilley Ladder Co., 548 Pa. 286, 696 A.2d 1169 (1997), and therefore I am doubtful that an argument based on Rule 403 of the Federal Rules of Evidence is properly before us. However, because I see no conflict between state and federal law on the point in question, I would follow the path we took in Schulz v. Celotex Corp., 942 F.2d 204, 207 (3d Cir.1991), and I would refrain from deciding which law applies.

Under either federal or state law, Beloit did not establish an adequate foundation. Beloit did not introduce its own safety records. Nor did it introduce evidence regarding the safety history of a large number of similar machines. Instead, in cross-examining two long-time employees, Beloit elicited testimony about the absence of prior accidents involving the machine in question. One of the employees, Edward Marshall, testified as follows:

Q.... You've never had an accident on the gloss calendar, correct?

A. I never had an accident, no.

Q. Okay. And you're not aware of anyone other then (sic) Mr. Forrest that's ever had an accident where his hand went through the gloss calendar correct?....

A. No.... I'm only aware of Paul, that's all.

A-709. This was the entirety of Marshall's testimony on this point. Thus, he was not asked and did not state whether he would have been aware of any accidents that occurred when he was not present. Moreover, he stated only that he had not been personally involved in any accidents on the gloss calendar and that he was not aware of any accidents in which a worker's hand "went through the gloss calendar." A-709. He was not asked about and did not state whether he knew of other types of accidents that might be relevant. In my view, Marshall's brief testimony did not provide the foundation required by either Spino or the Federal Rules of Evidence.

The other employee, William Brody, provided a bit more information than Marshall, but his testimony was far from ideal. Specifically, Brody's brief testimony on this point focused solely on injury-producing accidents and not on any other accidents that might have been relevant. See A-669. In any event, even if Brody's testimony provided a barely adequate foundation, the admission of Marshall's testimony was erroneous and could not be regarded as harmless. The foundation requirement in cases like this is important, and this requirement cannot be met in the casual way that Beloit attempted here. For these reasons, I agree that the judgment of the District Court must be reversed and the case must be remanded for a new trial.