OPINION

N.R. SMITH, Circuit Judge:

This case requires us to once again consider the district court's admission of expert testimony at trial. We review the admission of expert testimony at trial for an abuse of discretion. Primiano v. Cook, 598 F.3d 558, 563 (9th Cir.2010). If the district court improperly admitted the expert testimony, we apply harmless error review to determine whether its decision must be reversed. United States v. Laurienti, 611 F.3d 530, 547 (9th Cir. 2010). When we find the erroneous admission of evidence actually prejudiced the defendant, such that the error was not harmless, the appropriate remedy is a new trial. See United States v. 4.85 Acres of Land, 546 F.3d 613, 620 (9th Cir.2008). Applying this well-settled precedent to the facts of this case, we vacate the judgment and remand for a new trial.1

FACTS

Henry Barabin worked at Crown-Zellerbach paper mill from 1968 until his retirement in 2001. In the mill, Crown-Zellerbach shredded logs into chips and then exposed the chips to corrosive chemicals and high pressure to create paper slurry. Paper slurry is ninety-nine percent water and one percent pulp fiber. The mill produced paper by removing water from the paper slurry. As part of that process, machines pulled the paper through dryers. Dryer felts held the paper against the dryers, so that the paper would dry properly. AstenJohnson, Inc. and Scapa Dryer Fabrics, Inc. supplied the mill with dryer felts that contained asbestos.

Barabin had a variety of jobs during the time he worked at the mill. He started as a paper sorter, working in a different building than where the dryers were located. He then moved to the technical department, where he worked as a pulp tester and a paper tester. On occasion, he worked at a test station that was about twenty feet from the dryers. After working in the technical department, he went to work on the paper machines. Part of his job was to clean the dryers. However, these jobs were not his only exposure to the dryer felts; he also took pieces of dryer felt home to use in his garden.

In 2006, Barabin was diagnosed with pleural malignant epithelial mesothelioma ("mesothelioma"). Mesothelioma is a rare cancer that affects the tissue surrounding the lungs. Alleging that this occupational exposure to asbestos from the dryer felts caused his mesothelioma, Henry Barabin and Geraldine Barabin, his wife, brought suit against AstenJohnson and Scapa.

All parties agree, and the science makes clear, that asbestos exposure from inhaling respirable fibers can cause mesothelioma. At trial, the parties argued over whether exposure to the dryer felts (provided by AstenJohnson and Scapa) substantially contributed to Barabin's mesothelioma. Of necessity, the case was to be a battle of the experts. Both parties had experts who were prepared to testify in support of their arguments.

Two of the Barabins' experts were Kenneth Cohen and Dr. James Millette. Mr. Cohen had been employed in the industrial hygiene field for several decades. He had also taught industrial toxicology courses at a university. Dr. Millette had been involved in asbestos related research since 1974. He published a number of articles dealing with asbestos, including an article dealing with asbestos fiber release from dryer felts.

Prior to trial, AstenJohnson and Scapa filed motions in limine to exclude Mr. Cohen and Dr. Millette as expert witnesses. AstenJohnson argued that Mr. Cohen was not qualified to testify as an expert and that his theory was not the product of scientific methodology. Asten-Johnson and Scapa argued that Dr. Millette's tests were unreliable, because his methodology was not generally accepted in the scientific community. They also argued that the disparity between his tests and the conditions at the mill was so great that his testimony would not help the jury. The motions also sought to exclude testimony from any expert regarding the theory that "every asbestos fiber is causative."

After receiving the motions, and without a Daubert2 hearing, the district court excluded Mr. Cohen as a witness because of his "dubious credentials and his lack of expertise with regard to dryer felts and paper mills." The district court also had concerns with Dr. Millette's testimony. Specifically, the district court was "troubled by the marked differences between the conditions of Dr. Millette's tests and the actual conditions at the [mill]." Nonetheless, the district court ruled that Dr. Millette could testify provided the jury was informed his tests were "performed under laboratory conditions which are not the same as conditions at the [mill]."

As to the "every exposure" theory, the district court found "a strong divide among both scientists and courts" on whether it is relevant in asbestos cases. However, "[i]n the interest of allowing each party to try its case to the jury," the district court allowed the testimony.

The Barabins then filed a motion to request a pretrial Daubert hearing regarding Mr. Cohen. At a pretrial conference, the district court rejected the Barabins' request for a Daubert hearing. Instead, it reversed its decision to exclude Mr. Cohen's testimony. The district court's only explanation for why it reversed its decision was that the Barabins "did a much better job" in their motion "of presenting ... the full factual basis behind Mr. Cohen testifying and his testimony in other cases."

At trial, numerous experts testified. Both Mr. Cohen and Dr. Millette testified. Dr. Brodkin, another expert, also testified for the Barabins. Part of Dr. Brodkin's testimony was about the "every exposure" theory. As each of these experts testified, AstenJohnson and Scapa objected to their testimony. The district court overruled the objections.

PROCEDURAL HISTORY

After the Barabins presented their case at trial, AstenJohnson and Scapa filed motions for judgment as a matter of law. AstenJohnson and Scapa believed they were entitled to judgment as a matter of law, because the Barabins had failed to show that their companies had manufactured the dryer felts to which Barabin had been exposed. In the alternative, they argued that the Barabins had failed to demonstrate a causal link between the dryer felts and Barabin's mesothelioma. The district court denied the motions. AstenJohnson and Scapa renewed their motions after closing arguments. The district court denied the motions again.

After deliberations, the jury found in favor of the Barabins and awarded damages totaling $10,200,000. The district court granted AstenJohnson's and Scapa's motions to vacate the judgment and scheduled a reasonableness hearing. The district court found the damages award to be reasonable, offset the judgment by a total of $836,114.61,3 and entered judgment in favor of the Barabins in the amount of $9,373,152.12.

Both Scapa and AstenJohnson then filed motions for a new trial or, in the alternative, for a remittitur. One of the grounds on which Scapa and AstenJohnson sought a new trial was the improper admission of expert testimony. The district court denied the motions in their entirety.

AstenJohnson and Scapa filed timely notices of appeal. A three judge panel consolidated the appeals. It unanimously held that the district court abused its discretion by failing to make the necessary relevancy and reliability findings under Daubert. The panel remanded for a new trial pursuant to Mukhtar v. California State University, 299 F.3d 1053 (9th Cir. 2002), amended by 319 F.3d 1073 (9th Cir.2003). The Barabins petitioned this Court to rehear the case en banc, and a majority of non-recused active judges voted to rehear the case.

STANDARDS OF REVIEW

"A district court's evidentiary rulings should not be reversed absent clear abuse of discretion and some prejudice." Grand Canyon Skywalk Dev., LLC v. `Sa' Nyu Wa Inc., 715 F.3d 1196, 1202 (9th Cir.2013) (internal quotations and citation omitted). However, we review de novo the "construction or interpretation of ... the Federal Rules of Evidence, including whether particular evidence falls within the scope of a given rule." United States v. Durham, 464 F.3d 976, 981 (9th Cir. 2006) (citations omitted). A ruling on a motion for new trial "will be overturned on appeal only for abuse of discretion." Kode v. Carlson, 596 F.3d 608, 611 (9th Cir.2010) (citation omitted).

DISCUSSION

I.

Rule 702 of the Federal Rules of Evidence governs admission of expert testimony in the federal courts:

Fed.R.Evid. 702 (2010).4 We have interpreted Rule 702 to require that "[e]xpert testimony ... be both relevant and reliable." United States v. Vallejo, 237 F.3d 1008, 1019 (9th Cir.2001). Relevancy simply requires that "[t]he evidence ... logically advance a material aspect of the party's case." Cooper v. Brown, 510 F.3d 870, 942 (9th Cir.2007).

The issue here is reliability: whether an expert's testimony has "a reliable basis in the knowledge and experience of the relevant discipline." Kumho Tire Co., Ltd. v. Carmichael, 526 U.S. 137, 149, 119 S.Ct. 1167, 143 L.Ed.2d 238 (1999) (citation and alterations omitted). The "evidentiary reliability [is] based upon scientific validity." Daubert, 509 U.S. at 590 n. 9, 113 S.Ct. 2786. We are concerned "not [with] the correctness of the expert's conclusions but the soundness of his methodology." Primiano v. Cook, 598 F.3d 558, 564 (9th Cir.2010) (citations and quotations omitted). The duty falls squarely upon the district court to "act as a `gatekeeper' to exclude junk science that does not meet Federal Rule of Evidence 702's reliability standards." Ellis v. Costco Wholesale Corp., 657 F.3d 970, 982 (9th Cir.2011).

The reliability inquiry is "a flexible one." Kumho Tire, 526 U.S. at 150, 119 S.Ct. 1167. The Supreme Court has suggested several factors that can be used to determine the reliability of expert testimony: "1) whether a theory or technique can be tested; 2) whether it has been subjected to peer review and publication; 3) the known or potential error rate of the theory or technique; and 4) whether the theory or technique enjoys general acceptance within the relevant scientific community." United States v. Hankey, 203 F.3d 1160, 1167 (9th Cir.2000) (citing Daubert, 509 U.S. at 592-94, 113 S.Ct. 2786). However, whether these specific factors are "reasonable measures of reliability in a particular case is a matter that the law grants the trial judge broad latitude to determine." Kumho Tire, 526 U.S. at 153, 119 S.Ct. 1167.

The trial judge also has broad latitude in determining the appropriate form of the inquiry. See United States v. Alatorre, 222 F.3d 1098, 1102 (9th Cir. 2000) ("Nowhere ... does the Supreme Court mandate the form that the inquiry into relevance and reliability must take."). While pretrial "Daubert hearings" are commonly used, see, e.g., United States v. Lukashov, 694 F.3d 1107, 1112 (9th Cir. 2012), they are certainly not required, United States v. Jawara, 474 F.3d 565, 582 (9th Cir.2007).

Nevertheless, Rule 702 "clearly contemplates some degree of regulation of the subjects and theories about which an expert may testify." Daubert, 509 U.S. at 589, 113 S.Ct. 2786 (emphasis added). Applying these principles to the facts before us, we find the district court abused its discretion by failing to make appropriate determinations under Daubert and Federal Rule of Evidence 702.

The district court first excluded Mr. Cohen's testimony based on his "dubious credentials and lack of expertise." The district court's only explanation for reversing its decision, without a Daubert hearing or findings, was, "I think the plaintiffs did a much better job of presenting to me the full factual basis behind Mr. Cohen testifying and his testimony in other cases." Absent from the explanation is any indication that the district court assessed, or made findings regarding, the scientific validity or methodology of Mr. Cohen's proposed testimony. Therefore, the district court failed to assume its role as gatekeeper with respect to Mr. Cohen's testimony.

The district court also failed to act as gatekeeper for Dr. Millette's testimony. After acknowledging various arguments as to whether the testimony was admissible, the district court concluded that it could be admitted, so long as the jury was informed of the "marked differences" between conditions of the tests and the actual conditions of the mill. Rather than making findings of relevancy and reliability, the district court passed its greatest concern about Dr. Millette's testimony to the jury to determine.

The district court took the same approach with respect to expert testimony regarding the "every exposure" theory:

(emphasis added). Just as the district court cannot abdicate its role as gatekeeper, so too must it avoid delegating that role to the jury.

Here, the district court delegated that role by giving each side leeway to present its expert testimony to the jury. Before allowing the jury to hear the expert testimony, the district court should have first determined that the "expert meets the threshold established by Rule 702," Primiano, 598 F.3d at 564-65; that is the entire purpose of Daubert. The district court abused its discretion by admitting the expert testimony without first finding it to be relevant and reliable under Daubert.

II.

When we conclude evidence has been improperly admitted, "we consider whether the error was harmless." United States v. Bailey, 696 F.3d 794, 802-03 (9th Cir. 2012). We treat the erroneous admission of expert testimony the same as all other evidentiary errors, by subjecting it to harmless error review. See United States v. Rahm, 993 F.2d 1405, 1415 (9th Cir. 1993). We reverse "only if the error affect[ed] a substantial right of the party." Fed.R.Evid. 103(a). "In other words, we require a finding of prejudice." Obrey v. Johnson, 400 F.3d 691, 699 (9th Cir.2005).

"[T]he burden [is] on the beneficiary of the error either to prove that there was no injury or to suffer a reversal of his erroneously obtained judgment." Id. at 700 (citation omitted). Thus, "we begin with a presumption of prejudice. That presumption can be rebutted by a showing that it is more probable than not that the jury would have reached the same verdict even if the evidence had not been admitted." Jules Jordan Video, Inc. v. 144942 Canada Inc., 617 F.3d 1146, 1159 (9th Cir.2010) (citation and internal quotations omitted).

As the beneficiaries of the erroneously admitted evidence, the Barabins fail to rebut the presumption of prejudice. Indeed, they admit they cannot win without this expert testimony.5 Prejudice is at its apex when the district court erroneously admits evidence that is critical to the proponent's case. The improper admission of the expert testimony severely prejudiced AstenJohnson and Scapa because the Barabins' claim depended wholly upon the erroneously admitted evidence. Given these circumstances, there is no doubt the error was not harmless.

The dissent contends that we must decide whether the evidence would be admissible before engaging in harmless error review. Dissent at 469. However, the dissent is reading a non-existent step into our evidentiary-error case law.

The dissent cites only two cases addressing the use of harmless error review in these circumstances.6 They both support our decision. See 4.85 Acres of Land, 546 F.3d at 620; Simpson v. Thomas, 528 F.3d 685, 691 (9th Cir.2008). In 4.85 Acres of Land, we found that the district court abused its discretion by excluding all post-taking sales from consideration without first making any findings regarding the comparability of the excluded sales to the condemned property. 546 F.3d at 620. Despite being asked to do so, we refused to address on appeal whether "some of the post-taking comparable sales would have been admissible." Id. Instead, we engaged in harmless error review, found that the error was not harmless, and remanded for a new trial. Id.

In Simpson, we found that the district court abused its discretion when it admitted three prior convictions that were more than ten years old without engaging in proper balancing under rule 609(b) of the Federal Rules of Evidence. 528 F.3d at 690-91. The district court identified the correct rule, but it abused its discretion when it inverted the requirement of the rule, failed to offer specific facts to support its conclusion, and did not find that the probative value substantially outweighed the prejudice. Id. at 690. We did not engage in 609(b) balancing on appeal to determine whether the prior convictions would have been admissible. Instead, we went straight to harmless error review, found the evidence to be prejudicial, and remanded for a new trial. Id. at 690-91.

As both 4.85 Acres of Land and Simpson illustrate, when the district court abdicates its responsibility to answer a threshold question of admissibility, we need not determine whether the evidence would have been admissible before we determine the district court abused its discretion and proceed to harmless error review. In both cases we engaged in harmless error review, as we always do, by asking if erroneously admitting or excluding the evidence affected the outcome of the trial.7 See 4.85 Acres of Land, 546 F.3d at 620; Simpson, 528 F.3d at 691. We reject the dissent's attempt to insert a new step into our review of evidentiary errors.

III.

When the district court has erroneously admitted or excluded prejudicial evidence, we remand for a new trial. See, e.g., B.K.B. v. Maui Police Dep't, 276 F.3d 1091, 1109 (9th Cir.2002). We do so even if the district court errs by failing to answer a threshold question of admissibility. See, e.g., 4.85 Acres of Land, 546 F.3d at 620 (excluding other sales without determining comparability); Simpson, 528 F.3d at 691 (admitting convictions without Federal Rule of Evidence 609(b) balancing). We have no precedent for treating the erroneous admission of expert testimony any differently.

For seventy years prior to Daubert, the dominant standard for determining admissibility of novel scientific evidence was the "general acceptance" test. Daubert, 509 U.S. at 585-86, 113 S.Ct. 2786 (citing Frye v. United States, 293 F. 1013 (D.C.Cir. 1923)). Under Frye, we required the proponent of novel scientific evidence to "lay a proper foundation" by demonstrating "general acceptance" of the evidence "in the particular field in which it belong[ed]." United States v. Boise, 916 F.2d 497, 503 (9th Cir.1990).

In Daubert, the Supreme Court set "the standard for admitting expert scientific testimony in a federal trial" by holding that the Federal Rules of Evidence superseded the Frye test. Daubert, 509 U.S. at 582, 586-87, 113 S.Ct. 2786. Daubert continues to require that the proponent of expert testimony lay a proper foundation, but now laying a proper foundation means establishing relevancy and reliability rather than mere general acceptance. Id. at 597, 113 S.Ct. 2786.

Initially, in grappling with the effects of Daubert, we remanded two cases to district courts to conduct post-hoc Daubert hearings. See United States v. Cordoba, 104 F.3d 225, 229 (9th Cir.1997); United States v. Amador-Galvan, 9 F.3d 1414, 1418 (9th Cir.1993).8 After the dust of Daubert had settled, we held that an erroneous admission of expert testimony, absent a showing the error was harmless, requires a new trial. See Mukhtar, 299 F.3d at 1066-67. To the extent Mukhtar requires anything more, it is overruled.

AstenJohnson and Scapa contend that a reviewing court should have the authority to make Daubert findings based on the record established by the district court. We agree and overrule Mukhtar to the extent that it required that Daubert findings always be made by the district court. See Mukhtar, 299 F.3d at 1066 n. 12. If the reviewing court decides the record is sufficient to determine whether expert testimony is relevant and reliable, it may make such findings. If it "determines that evidence [would be inadmissible] at trial and that the remaining, properly admitted evidence is insufficient to constitute a submissible case[,]" the reviewing court may direct entry of judgment as a matter of law. Weisgram v. Marley Co., 528 U.S. 440, 446-47, 120 S.Ct. 1011, 145 L.Ed.2d 958 (2000).

Citing Weisgram, AstenJohnson and Scapa argue we should enter judgment in this case. We decline their invitation. In Weisgram, the Eighth Circuit found, based on a fully developed record, that the expert testimony was not reliable. Weisgram v. Marley Co., 169 F.3d 514, 517-18 (8th Cir.1999). We cannot speak to the admissibility of the expert testimony at issue here because the record before us is too sparse to determine whether the expert testimony is relevant and reliable. We can only say with certainty that the district court erred by failing to make that determination.

The Barabins and the dissent argue that we should remand for a post-hoc Daubert hearing. Even assuming that a limited remand is available post-Mukhtar, see 319 F.3d at 1074, it would not be appropriate under the circumstances here, where the district court abused its discretion by erroneously admitting expert testimony, and the evidence was prejudicial. We therefore remand for a new trial.

CONCLUSION

The district court failed to make findings of relevancy and reliability before admitting into evidence the expert testimony of Mr. Cohen and Dr. Millette and expert testimony regarding the theory that "every asbestos fiber is causative." The district court's failure to make these gateway determinations was an abuse of discretion. The error was prejudicial because the erroneously admitted evidence was essential to the Barabins' case. Due to the district court's abdication of its role as gatekeeper and the severe prejudice that resulted from the error, the appropriate remedy is a new trial. We vacate the judgment and remand for a new trial.

VACATED and REMANDED.

The parties shall bear their own costs on appeal.

Circuit Judge NGUYEN, with whom Judges McKEOWN, W. FLETCHER, BYBEE, and WATFORD join, concurring in part and dissenting in part:

I concur in Part I of the majority's opinion, which concludes that the district court failed to fulfill its gatekeeping function with regard to the expert testimony at issue. I also agree with the majority that we are unable to determine based on the record before us whether the expert testimony is admissible. See Maj. Op. at 467 ("We cannot speak to the admissibility of the expert testimony at issue here because the record before us is too sparse to determine whether the expert testimony is relevant and reliable."). Further, to the extent the majority overrules Mukhtar v. California State University, 299 F.3d 1053 (9th Cir.2002), amended by 319 F.3d 1073 (9th Cir.2003), I am in accord. There is no reason to require a new trial whenever a district court fails to conduct a Daubert analysis, regardless of whether on remand the district court would determine that disputed expert testimony is relevant and reliable.

I part ways with the majority, however, in its application of harmless error review. The majority's analysis is seriously flawed because it conflates a district court's gatekeeping error with a district court's erroneous determination of admissibility. Here, assuming inadmissibility — a question we cannot answer at this juncture — the majority applies harmless error review and concludes that a new trial is needed because the "improper admission of the expert testimony severely prejudiced [defendants]." Maj. Op. at 465 (emphasis added). The majority thus unnecessarily burdens both the parties and the judicial system by ordering a new trial without having a sufficient basis to determine whether the disputed expert testimony was admissible. Further, the majority's approach undercuts its effort to open the door to a limited remand occasioned by overruling Mukhtar. Because I would conditionally vacate the judgment and remand with instructions to the district court to conduct a Daubert determination in the first instance, I respectfully dissent from Parts II and III of the majority opinion.

I.

A district court must "ensure the reliability and relevancy of expert testimony" and "make certain that an expert, whether basing testimony upon professional studies or personal experience, employs in the courtroom the same level of intellectual rigor that characterizes the practice of an expert in the relevant field." Kumho Tire Co., Ltd. v. Carmichael, 526 U.S. 137, 152, 119 S.Ct. 1167, 143 L.Ed.2d 238 (1999); see also Fed.R.Evid. 702. In short, a district court serves an essential "gatekeeping" function in evaluating proffered expert testimony. Kumho Tire, 526 U.S. at 141, 119 S.Ct. 1167.

In this case, the district court abdicated its gatekeeping role by failing to evaluate the relevancy and reliability of the expert testimony at issue. Given this oversight, we must determine whether the district court's misstep resulted in the admission of evidence that should have been excluded. In the past, when the record before us was sufficient to make this determination, we have proceeded to evaluate whether the erroneous admission or exclusion of evidence was harmless. See, e.g., United States v. Morales, 720 F.3d 1194, 1197 (9th Cir.2013) (district court's error "was harmless because the erroneously admitted hearsay did not materially affect the verdict"); United States v. 4.85 Acres of Land, 546 F.3d 613, 620 (9th Cir.2008) (district court's error was not harmless because improper exclusion of evidence was prejudicial). This approach makes perfect sense — once we determine that inadmissible evidence was presented, or that admissible evidence was excluded, we can then analyze whether the error materially affected the verdict. See United States v. Bailey, 696 F.3d 794, 803 (9th Cir.2012).

Here, however, we face a markedly different scenario. As the majority correctly concludes, "the admissibility of the expert testimony at issue" cannot be determined "because the record before us is too sparse." Maj. Op. at 467. Given this conclusion, harmless error review is simply not possible at the current juncture. Indeed, we cannot even say whether there was an "error" to "materially affect the verdict." If the disputed expert testimony was admissible pursuant to Rule 702 and Daubert, despite the district court's failure to fulfill its gatekeeping function, then no harm, no foul. On the other hand, if the testimony was inadmissible, then a harmless error analysis would be appropriate. Thus, in light of the outstanding question regarding the admissibility of the expert testimony at issue, a remand to the district court for a Daubert analysis is the proper course.

II.

The majority goes awry in adopting an approach that ignores this antecedent question of admissibility. In considering the district court's gatekeeping failure, the majority asserts that "[w]hen we conclude evidence has been improperly admitted, `we consider whether the error was harmless.'" Maj. Op. at 464 (quoting Bailey, 696 F.3d at 802-03). Though innocuous at first glance, this remark harbors a grave oversight: it equates an incorrect determination of admissibility with a failure to conduct a Daubert analysis.

The distinction between the two is crucial. With the former, we know whether a party was wrongfully permitted or denied the opportunity to present certain evidence, and we can determine whether that error was prejudicial. With the latter, we cannot gauge prejudice unless we are able to determine what the jury would have been permitted to hear had the district court properly discharged its gatekeeping duties.1

By skipping over the question of admissibility and heading straight for prejudice, the majority's analysis results in two key missteps. First, the majority dubs the Barabins "the beneficiaries of ... erroneously admitted evidence." Id. at 465. But, as the majority acknowledges, we have no idea whether the expert testimony at issue was in fact "erroneously admitted." On a proper Daubert analysis — a task we decline to engage in on appeal — the testimony might indeed have been admissible. In this circumstance, the Barabins would merely be the beneficiaries of evidence they were entitled to present in the first place.

Second, in bypassing admissibility, the majority engages in a perplexing prejudice analysis that emphasizes the fact that the disputed expert testimony was "critical to the proponent's case."2 Id. Indeed, based on its conclusion that "the Barabins' claim depended wholly upon the erroneously admitted evidence," the majority finds "no doubt the error was not harmless." Id. Not so. Even if the Barabins' claim depended on the expert testimony at issue, we have no idea whether the testimony was "erroneously admitted," let alone whether any error "materially affect[ed] the verdict." Morales, 720 F.3d at 1197. If the expert testimony was admissible, then the jury simply reached a verdict based on evidence it was properly permitted to consider, despite the district court's error.

The flaw in the majority's logic is high-lighted by the fact that not a single case it cites supports the type of harmless error analysis it applies. Rather, in each case, we engaged in harmless error review only after we determined that evidence had been improperly deemed admissible or inadmissible.3 Indeed, this is true of Simpson v. Thomas and United States v. 4.85 Acres of Land — two cases specifically discussed by the majority in support of its decision. Contrary to the majority's assertion, neither case involved merely a situation where a "district court abdicate[d] its responsibility to answer a threshold question of admissibility." Maj. Op. at 466. Rather, in both cases, we found actual error by the district court in deeming evidence admissible or inadmissible before proceeding to harmless error review. See Simpson, 528 F.3d at 689 ("The district court erred in admitting the evidence of Simpson's three prior felony convictions."); 4.85 Acres of Land, 546 F.3d at 620 ("[T]he [district] court simply excluded all post-taking sales based on `the erroneous premise... that evidence of subsequent sales is never proper for consideration in arriving at fair market value.'" (quoting United States v. 1,129.75 Acres of Land, 473 F.2d 996, 999 (8th Cir.1973)) (final alteration in original)).

In fact, our case law suggests a notably different harmless error inquiry when a district court fails to fulfill its gatekeeping function. For example, in Hangarter v. Provident Life & Accident Insurance Co., 373 F.3d 998 (9th Cir.2004), the district court incorrectly concluded that Daubert was inapplicable to the non-scientific testimony of an independent consultant. Id. at 1015-16, 1018. We found the error harmless because "the [district] court's probing of [the consultant's] knowledge and experience was sufficient to satisfy its gatekeeping role under Daubert." Id. at 1018. In short, the district court's failure to analyze the expert testimony pursuant to Daubert was harmless because the analysis it nonetheless conducted satisfied Daubert and the testimony was thus correctly admitted. This was a proper application of harmless error review — where the error identified on appeal pertains to the gatekeeping function, the reviewing court should consider whether the gatekeeping error was harmless.4

In contrast, the majority here finds a gatekeeping error, but embarks on a prejudice inquiry that focuses on how crucial the disputed expert testimony was to the prevailing party's success. In doing so, the majority effectively treats the testimony as inadmissible, even as it professes to reserve judgment on the question. The majority cannot have it both ways.

III.

I would conditionally vacate the judgment and remand to the district court with instructions to determine whether the disputed expert testimony was admissible pursuant to the requirements of Rule 702 and Daubert. If the testimony is determined to be admissible, the district court may reinstate the verdict. If, however, the testimony is inadmissible, the district court should ascertain whether the wrongful admission of that expert testimony prejudiced the defendants and, if so, order a new trial. In the former case, the system will not be unreasonably burdened with a retrial. In either case, the parties retain their right to appeal. This solution makes practical and legal sense.