Brown v. Southeastern Pennsylvania Transportation Authority

TABLE OF CONTENTS

I. INTRODUCTION 782

II. FACTS AND PROCEDURAL HISTORY 734

III. SUBJECT MATTER JURISDICTION OF THE DISTRICT COURT 737

IV. WAS THE IN LIMINE HEARING TAINTED BY UNFAIR PROCEDURES? 738

A. Inadequate Opportunity to Depose Defense Experts 738

B. Exclusion of Opinion of Experts for Whom No. 26(b)(4) Statements Were Filed 739

V. LEGAL STANDARDS FOR THE ADMISSIBILITY OF EXPERT OPINION 741

A. Rule 702 741

1. Qualifications 741

2. Reliability 742

3. Fit 742

4. The roles of judge and jury: how high a threshold for reliability? 743

5. Confusion — Rules 702 and 403 746

6. Procedural concerns regarding the Rule 403/702 balancing test 747

B. Rule 703 747

C. Standard of Review 749

D. The Significance of Pennsylvania’s Requirement of Reasonable Medical Certainty 750

VI. THE RELATIONSHIP BETWEEN PCBs AND PLAINTIFFS’ ILLNESSES 752

A. Dr. Sherman’s Qualifications 753

B. Rule 703 and Dr. Sherman’s Immunological Tests 754

C. Adequacy of Dr. Sherman’s and Dr. DiGregorio’s Use of Differential Diagnosis in Evaluating Causation of Present Illnesses 755

1. The district court’s critique 755

2. The plaintiffs’ response 757

3. Rule 702 analysis 758

a. The opinions of Dr. DiGregorio 763

b. The opinions of Dr. Sherman 764

i. Bessie Cunningham 765

ii. Amber Burrell 766

iii. Priscilla Burrell 767

iv. Monica Hilton 768

v. Matthew Cunningham 768

vi. Patricia Ingram 769

vii. John Ingram, Sr. 769

viii. John Ingram, Jr. 769

ix. George Burrell 769

x. Wallace Cummins 769

xi. April Ingram 770

xii. The Remaining Plaintiffs 770

c. Rule 403 770

d. Summary (DiGregorio and Sherman) 770

VII. EXPOSURE — THE OPINION OF IAN C.T. NISBET, Ph.D. 771

A. Recalculation of the AML Data 772

B. Recalculation of Background 773

C. Back Calculations 774

D. Remaining Issues 778

E. Conclusion 778

VIII. THE HARMFULNESS OF THE CHEMICALS 778

A. Introduction 778

B. Animal Studies 779

*732 C. Exposure to Dioxins and Furans 781

IX. THE YUSHO AND YU CHENG INCIDENTS 783

X.THE PLAINTIFFS’ PERSONAL INJURY CLAIMS — SUMMARY AND CONCLUSION 784

XI. THE MEDICAL MONITORING CLAIMS 785

A. Medical Monitoring as a Viable Claim 785

B. The Medical Monitoring Test Restated 787

C. Application of the Test 788

1. Admissibility of plaintiffs’ evidence 789

a. Dr. Sherman’s testimony 789

b. Dr. DiGregorio’s testimony 790

i. Reliability 790

ii. Failure to comply with the scheduling order 791

c. Dr. Sehecter’s testimony 1 793

2. Summary judgment on medical monitoring 793

XII. PROPERTY DAMAGE 795

XIII. CONCLUSION 798

Before BECKER, ROTH, and LEWIS, Circuit Judge.

OPINION OF THE COURT

BECKER, Circuit Judge.

I. INTRODUCTION

The plaintiffs in this toxic tort case have lived for many years in the vicinity of the Paoli Railyard, a railcar maintenance facility at which polychlorinated biphenyls (PCBs) were used in profusion for over a quarter century. They have sued to recover damages for a variety of physical ailments and for property damage against the corporations that have maintained the railyard and that sold the PCBs.

This appeal is from the district court’s second grant of summary judgment for the defendants. In In re Paoli R.R. Yard PCB Litig., 916 F.2d 829 (3d Cir.1990) (“Paoli I”), cert. denied, 499 U.S. 961 , 111 S.Ct. 1584 , 113 L.Ed.2d 649 (1991), we reversed the grant of summary judgment for defendants and remanded for further proceedings. We did so because the foundation of the summary judgment — the exclusion of virtually all the plaintiffs’ expert opinion pursuant to Federal Rules of Evidence 702, 703 and 403 — was undermined by (1) the failure of the district court to permit sufficient development of the record upon which a determination to exclude expert evidence might be based, and (2) the district court’s failure to make findings, grounded on Fed.R.Evid. 702 or 703, setting forth the basis of its decision.

On remand the district court conducted five days of in limine hearings, receiving extensive evidence about the scientific reliability of plaintiffs’ expert opinions. It then filed extensive opinions (totalling 330 pages) setting forth not only findings of fact but also its reasons for again excluding the vast bulk of plaintiffs’ expert evidence. This appeal followed.

Resolution of the appeal requires that we address myriad issues of procedure, evidence, and substantive law. Primarily, however, we must consider the voluminous record concerning expert opinion and, applying Fed.R.Evid. 702 and the standards enunciated by the Supreme Court in Daubert v. Merrell Dow Pharm., Inc., — U.S. —, 113 S.Ct. 2786 , 125 L.Ed.2d 469 (1993), decide whether the district court erred in again excluding the opinions of plaintiffs’ experts in connection with its summary judgment determination. Daubert requires the district court to act as “gatekeeper” and to assure that the scientific methodology upon which the expert opinion is founded is reliable, i.e., that the expert’s conclusion is based on good grounds (the methods and principles of science). We also must deal with Fed.R.Evid. 703 because the district court’s exclusion of expert testimony rests, in part, on its conclusion that some of the hearsay evidence upon which the experts relied is not of a kind usually relied on by experts in the field. In this regard.we conclude that the precept announced in In re: Japanese Electronics Prod. Antitrust Litig., 723 F.2d 238 (3d Cir. *733 1983), rev’d on other grounds sub nom., Matsushita Electric. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574 , 106 S.Ct. 1348 , 89 L.Ed.2d 538 (1986), that the principal arbiters of the reasonableness of reliance upon inadmissible evidence are the experts and not the trial judge, does not survive Daubert .

Preliminary to an application of Rules 702-703 to the evidentiary rulings at issue, we must address the question of the proper scope of review. Although review of the district court’s fact findings that undergird its rulings on the admissibility of expert opinion is deferential, given the enormous power of the district court to foreclose submission of a party’s case to a jury on the basis of a threshold determination of nonreliability of opinion evidence, we conclude that the review requires a “hard look” to insure that the district court’s exercise of discretion was sound and that it correctly applied the several Daubert factors. Then to the extent that any of the expert opinion survives, we must determine whether the plaintiffs can avoid summary judgment on any of the claims to which the surviving opinion pertains.

For the most part we will affirm the rulings and the judgment of the district court. However, we also will reverse in part, and remand some claims for trial. In summary we rule as follows.

1. We will affirm the district court’s decision to retain jurisdiction in this case based on pendent jurisdiction stemming from the plaintiffs’ prior CERCLA claims.

2. We will affirm the district court’s discretionary ruling allowing certain of defendants’ experts to testify at the in limine hearings without being deposed; however, we point out that depositions should generally be afforded in future cases of this nature. We do not reach the propriety of its decision excluding plaintiffs’ experts for whom no Rule 26(b)(4) statements were filed because plaintiffs do not rely on these experts in opposition to summary judgment.

3. We will reverse the district court’s exclusion of the opinion of Dr. Janette D. Sherman insofar as it was grounded on the conclusion that she was an unqualified expert. However, we affirm the district court’s holding that she could not rely on immunological tests from an unaccredited laboratory. We also affirm the district court’s exclusion of the opinions of Dr. Sherman as to causation of plaintiffs’ present injuries with respect to those plaintiffs whom she did not examine and whose medical history she did not take. We believe that, in the absence of employment of the standard techniques of differential diagnosis, and in the absence of any good explanation justifying her conclusion in the light of possible alternative causes of plaintiffs’ injuries, Dr. Sherman’s methodology as to those plaintiffs was not reliable, as the district court found.

4. We will affirm the district court’s exclusion of the opinion of Dr. G. John DiGre-gorio with respect to causation of the injuries of those plaintiffs for whom he testified. Because Dr. DiGregorio neither examined these plaintiffs nor took their medical histories, he did not have a reliable basis from which to conclude they were ill. Moreover, as with Dr. Sherman, in the absence of employment of standard techniques of differential diagnosis, and failure to provide any explanation as to why he concluded that PCBs rather than alternative possibilities had caused plaintiffs’ illnesses, we conclude that Dr. DiGregorio’s assessment of causation was unreliable.

5. We will reverse the district court’s exclusion from evidence of the opinions of Dr. Sherman with respect to most of the injuries of Bessie Cunningham and Amber Burrell, whom she actually examined and whose medical histories she took. In light of: (1) Dr. Sherman’s employment of the standard techniques of differential diagnosis with respect to those plaintiffs, and (2) defendants’ failure to point to possible alternative causes of their injuries which Dr. Sherman failed to consider, we are satisfied that, with respect to them, Dr. Sherman’s scientific methodology was reliable and that the district court’s exclusion of her opinion represented an impermissible jury-like determination that it preferred the defendants’ experts’ opinion to that of Dr. Sherman.

6. We will affirm the district court’s ex-elusion from evidence of the opinion of Dr. Ian Nisbet regarding plaintiffs’ exposure to PCBs: (1) insofar as that opinion is based on *734 a recalculation of American Medical Laboratory test (“AML”) results to account for the degree to which the laboratory ostensibly underestimated its results; and (2) insofar as it is grounded on “back calculations” of plaintiffs’ PCB blood levels in 1986 based upon the results of studies made by the Eco Logic laboratory in 1992. We conclude that Dr. Nisbet’s methodology in recalculating the AML tests was unreliable, and that the Eco Logic data from which he began his back calculations was not a type reasonably relied on by experts in the field due to the quality control problems in Eco Logic’s methodology-

7. We will reverse the district court’s exclusion of the remainder of Dr. Nisbet’s opinion, including his conclusion that plaintiffs had PCBs in their blood at levels above the background PCB levels in the general population. We are satisfied that Dr. Nisbet’s calculation of background blood levels from National Human Adipose Tissue Survey (“NHATS”) fat data was based on a reliable methodology, and that the NHATS data was a type of data reasonably relied on by experts.

8. We will reverse the district court’s exclusion of animal studies. Where, as here, these studies were supported by some epidemiological data and had been used by EPA to conclude that PCBs were a probable human carcinogen, plaintiffs’ experts had good grounds for determining that they “fit” a conclusion regarding human causation.

9. We will affirm the district court’s exclusion of data on chemicals other than PCBs and its exclusion of testimony related to the incidents at Yusho (in Japan) and Yu Cheng (in Taiwan). The district court’s conclusion under Fed.R.Evid. 403 that the probative value of this evidence would have been substantially outweighed by unfair prejudice (and would have resulted in significant waste of trial time for resolution of collateral issues) was based on a more than adequate record (one justifying the rare making of the pretrial ruling based on Rule 403), and on findings of fact which, under the applicable standard of review we cannot disturb.

10. We will reverse the grant of summary judgment with respect to those personal injury plaintiffs for whom Dr. Sherman’s testimony is admissible, concluding that they have demonstrated the existence of a genuine issue of material fact as to whether defendants exposed them to PCBs which caused their present injuries.

11. We will reverse the grant of summary judgment with respect to the medical monitoring claims. Although recent changes in Pennsylvania law have interred plaintiffs’ increased risk claims, we still believe that the Pennsylvania Supreme Court would recognize the plaintiffs’ medical monitoring claims, and are satisfied that Dr. Sherman’s testimony as to the plaintiffs’ future risk is sufficient to get them over the summary judgment threshold.

12. We will reverse the grant of summary judgment on plaintiffs’ claim with respect to diminution in property value. Based on an extrapolation from existing Pennsylvania case law, we predict that where (1) the defendants have caused some physical harm to plaintiffs’ property; (2) repair of this damage will not restore the value of the property to its prior level; and (3) there is at least some ongoing risk to the plaintiffs’ land, Pennsylvania would allow plaintiffs to recover damages for the diminution of their property value. On the present record, there remains a genuine issue of material fact on the permanency of damage issue.

13. With respect to all other claims, which represent the bulk of plaintiffs’ ease, the exclusion of the expert opinion leaves the plaintiffs with inadequate evidence to survive summary judgment. We will therefore affirm the grant of summary judgment on these remaining claims.

II. FACTS AND PROCEDURAL HISTORY

Beginning in the 1950s, if not before, PCBs were used at the Paoli Railroad Yard (the ‘Yard”) as a fire-resistant insulating fluid in railroad car transformers. PCBs gradually accumulated in the Yard and leaked off the Yard into groundwater and ultimately into the soil of nearby residences. (A1122-1123). Several studies from 1979-1986 documented high levels of PCBs in the Yard and in *735 nearby water and land. The National Institute for Occupational Safety and Health (NIOSH) identifies the site as containing the worst PCB contamination it has ever encountered during a health inspection.

In 1986, the United States sued SEPTA, Conrail and Amtrak under the Comprehensive Environmental Response Compensation and Liability Act (CERCLA), 42 U.S.C. § 9601 , to compel the cleanup of the Yard. After the entry of five consent decrees, which primarily required the control of leakage from the Yard, EPA adopted a final plan, the Record of Decision, in July of 1992. This plan requires extensive excavation and treatment of soils both at the Yard and on nearby, residential property and streams, and erosion controls at the Yard.

The plaintiffs in this action consist of individuals who lived near the Yard in areas identified by EPA and by the railroad defendants’ contractor, Groundwater Technologies, Inc., as having experienced the most severe run-off from the Yard. The plaintiffs have adduced evidence of significant levels of PCBs in the soil at their homes. Many of the plaintiffs played in the soil at their homes while growing up, gardened in it, and ate vegetables grown from it. Many also regularly traversed the Yard on foot as a' short cut to their destinations.

In 1986, thirty eight plaintiffs brought suit in the District Court for the Eastern District of Pennsylvania. Some sought recovery for present injuries, (allegedly caused by exposure to PCBs and other assorted chemicals from the Yard, including polychlorinated dibenzo furans (“furans”) and polychlorinated dibenzo-p-dioxins (“dioxins”)). Some plaintiffs, many of whom have not yet suffered any physical injury, brought claims for emotional distress caused by fear of future injury, and for medical monitoring to decrease the likelihood of the future development of serious diseases. Finally, some of the plaintiffs have brought claims for the decrease in value to their property caused by the presence of PCBs on the land.

The defendants (in some or all of the cases) are Monsanto Corporation, the nation’s leading manufacturer of PCBs; General Electric Company, manufacturer of some of the railroad car transformers in which the PCBs were used; Amtrak, owner of the Yard since 1976; Conrail, which operated the facility between 1976 and 1983; the Southeastern Pennsylvania Transit Authority (SEPTA), which has operated the facility since 1983; and the City of Philadelphia, which is linked with SEPTA. SEPTA filed third-party complaints against Westinghouse Electric Corporation, manufacturer of some of the transformers used at the Yard, and the Budd Company, manufacturer of some of the railroad ears. Conrail filed a third part complaint against the Penn Central Corporation.

In November of 1988 the district court granted defendants’ motion for summary judgment against all plaintiffs on the personal injury claims because it held that all of plaintiffs’ evidence pertaining to significant exposure and causation, necessary elements of their claims, was inadmissible. The court excluded the evidence under Federal Rule of Evidence 702, which requires expert witnesses to be qualified and to use reliable scientific techniques, Rule 703 which requires experts to rely on data of a type reasonably relied upon by experts in the field, and Rule 403 which requires the probative value of evidence not to be substantially outweighed by unfair prejudice or undue delay. See In re Paoli R.R. Yard PCB Litig., 706 F.Supp. 358 (E.D.Pa.1988).

In Paoli I, 916 F.2d at 854 , this Court reversed the Rule 703 holdings because the district court had not created a sufficiently detailed factual record on which to base them and had not given “the plaintiffs sufficient opportunity to explore the issues upon which they were ultimately denied relief.” We reversed the district court’s Rule 702 holdings because it had construed the qualification requirement of Rule 702 too strictly, see id. at 855-56 , and because it had not created a detailed enough factual record to justify exclusion of the experts’ opinion under the reliability prong of Rule 702. See id. at 858-59 . Finally, we reversed the district court’s Rule 403 determinations, holding that Rule 403 exclusions should not be granted pretrial absent a record which is “a virtual surrogate for a trial record.” Id. at 859-60 . We strongly suggested that the district court *736 should have held an in limine hearing before excluding plaintiffs’ expert opinion. Id. at 859 .

Following remand, the district court issued a cáse management order in June 1991, giving plaintiffs until March 2, 1992 to complete their discovery, and until March 17, 1992 to designate their trial experts and provide defendants with their experts’ reports. The court required the defendants to do likewise one month after plaintiffs.

The list of expert witnesses designated by the plaintiffs in March of 1992 was considerably different from the list at the time of the 1988 summary judgment. The twenty-three plaintiffs represented by the Kohn, Nast, & Graf, and Klehr, Harrison, Harvey, Branz-burg, Ellers, & Weir law firms (the “Kohn/Klehr” plaintiffs) named three experts supporting their personal injury claims: Mel-vyn Kopstein, Ph.D., to testify about the Paoli residents’ opportunities for exposure to PCBs from the Yard; Ian C.T. Nisbet, Ph. D., to testify about the plaintiffs’ exposure to PCBs; and Janette Sherman, M.D., to testify that PCBs had caused plaintiffs’ injuries and that plaintiffs require medical monitoring to detect and treat future PCB-related illnesses.

Defendants moved in limine to exclude the opinions of each of these experts under Fed.R.Evid. 702, 703 and 403. On the same grounds, defendants filed motions in limine to preclude the plaintiffs’ experts from testifying about animal studies purporting to show the harmful effects of PCBs, evidence concerning the harm of dioxins and furans (chemicals that are sometimes present in transformer fluids), and evidence concerning the ‘Yusho” incident in Japan and the ‘Yu Cheng” incident in Taiwan in which many individuals suffered adverse effects after consuming rice oil contaminated with PCBs and furans.

Bruce Hanes, who represented the nine remaining plaintiffs (the “Hanes plaintiffs”), designated Dr. G. John DiGregorio, M.D. to testify that PCBs had caused their injuries, and Arnold Schecter, M.D., M.P.H., to testify that their fear of future illness was reasonable. In a joint pretrial memorandum filed on July 17, 1992, the Kohn/Klehr and Hanes plaintiffs designated three additional trial experts — Deborah Barsotti, Ph.D., Benjamin Calesnick, M.D., and Arthur Zahalsky, Ph. D. — who had offered opinions for the plaintiffs in 1988. They also listed William J. Nicholson, Ph.D. and Cate Jenkins, Ph.D. as fact witnesses. Defendants moved to exclude these witnesses because plaintiffs had failed to designate them as experts in March, as the district court’s pre-trial order had required.

The district court held five days of in limine hearings between July 28 and August 7,1992. Drs. Kopstein, Nisbet, and Sherman testified for the plaintiffs. Ten physicians and scientists testified for the defense. On October 21, 1992, the, court entered orders excluding all but one of plaintiffs’ experts (Dr. Kopstein). The court also excluded evidence concerning animal studies, dioxins and furans, and the incidents at Yusho and Yu Cheng. The court then granted summary judgment against the plaintiffs on their personal injury claims.

The court explained that under our holding in Paoli I, the plaintiffs had to establish the following elements to make out a prima facie case: (1) that the defendants had released PCBs into the environment; (2) that the plaintiffs had somehow ingested those PCBs into their bodies; (3) that the plaintiffs have an injury; and (4) that PCBs are the cause of their .injury. See Paoli I, 916 F.2d at 860 . Not surprisingly, after excluding plaintiffs’ experts, the court found that plaintiffs had no evidence on either exposure or causation and thus could not prove elements 2, 3, and 4. For similar reasons, the court granted summary judgment against the plaintiffs on their medical monitoring claims.

Nine plaintiffs — Mabel Brown, K. Louise Jones, James Lament, Christopher Brown, Margherita Barbetta, Mary Retta Johnson, John Ingram, William Butler, arid Matthew Cunningham — also brought property damage claims against the defendants. The district court granted summary judgment in favor of the defendants on these claims as well. The court held that Pennsylvania law only allows recovery for diminution of a property’s market value if there has been permanent damage to the property; and the court found *737 that plaintiffs could not prove such permanent damage in light of EPA’s proposed cleanup plan.

The plaintiffs challenge the district court’s subject matter jurisdiction, all of its admissibility decisions, its legal conclusions regarding diminution of a property’s market value, and the court’s grant of summary judgment. We will affirm most of the district court’s decisions. We will reverse only its grant of summary judgment on a number of the present injury claims of Bessie Cunningham and Amber Burrell, and its grant of summary judgment on the plaintiffs’ claims for medical monitoring and diminution in property value.

III. SUBJECT MATTER JURISDICTION OF THE DISTRICT COURT 1

Plaintiffs’ original claims in 1986 included a CERCLA claim, 42 U.S.C. § 9601 et seq., which conferred federal question jurisdiction, 28 U.S.C. § 1331 , on the district court. The court possessed pendent jurisdic-. tion over plaintiffs’ other claims. However, plaintiffs voluntarily dismissed the CERCLA claim after our ruling in Paoli I. Plaintiffs contend that, after this dismissal, the district court no longer had subject matter jurisdiction over the case because, absent the CERCLA claim, there was no federal question jurisdiction; nor was there complete diversity. 2

In United Mine Workers v. Gibbs, 383 U.S. 715 , 86 S.Ct. 1130 , 16 L.Ed.2d 218 (1966), the Supreme Court held that in evaluating whether to exercise pendent jurisdiction, a district court must weigh “considerations of judicial economy, convenience and fairness to [the] litigants.” 383 U.S. at 726 , 86 S.Ct. at U39. 3 In holding that these factors favored retention of jurisdiction here, the district court acted within the ambit of its discretion.

Considerations of judicial economy clearly weighed in favor of the district court retaining jurisdiction. Although considerations of judicial economy alone are generally insufficient to justify a district court’s decision to retain jurisdiction, see Shaffer v. Board of Sch. Directors , 730 F.2d 910 , 912 (3d Cir.1984); Lovell Mfg. v. Export-Import Bank of U.S., 843 F.2d 725 , 734-35 (3d Cir.1988), we have held such concerns sufficient when they are especially strong. In Lentino v. Fringe Employee Plans, Inc., 611 F.2d 474, 480 (3d Cir.1979), we upheld a district court’s decision to retain jurisdiction over a state law claim when plaintiffs dropped their federal claims on the morning of trial after there had already been a year of pretrial proceedings.

The concerns of judicial economy here are at least as strong as they were in Lentino . Although the plaintiffs dropped their CERC-LA claim much earlier than the morning before trial, far more energy had been invested in the proceedings here than in Lenti-no. Six years of federal proceedings had occurred before the plaintiffs dropped their CÉRCLA claim, including extensive discovery. Moreover, the concern over judicial economy was combined here with two other *738 factors strongly militating in favor of retention of jurisdiction.

First, the district court would have continued to have jurisdiction over plaintiffs’ claims against Amtrak, see 28 U.S.C. § 1349 (providing district court jurisdiction over all corporations incorporated by an act of Congress in which the United States owns more than half the stock) even if it had dismissed most of plaintiffs’ claims for lack of jurisdiction. Amtrak doubtless would then have impleaded the other defendants, all of whom had deep pockets, thus changing the alignment of the parties but not the nature of the litigation. In such circumstances, a district court acts within its discretion in retaining jurisdiction. See, e.g., New Jersey Dept. of Env. Prot. v. Gloucester Env. Mgmt. Services, Inc., 719 F.Supp. 325, 337 (D.N.J.1989) (retaining jurisdiction in order to prevent “a litigation merry-go-round”).

Second, plaintiffs were the ones who initially filed their claims in federal court and obtained a 1986 ruling that the CERCLA claims created federal jurisdiction. Plaintiffs dismissed their CERCLA claims and argued against federal jurisdiction only after the district court had first granted summary judgment against them. Thus, plaintiffs’ dismissal of their CERCLA claims may well constitute an attempt to forum shop by divesting the district court of jurisdiction. In Carnegie-Mellon Univ. v. Cohill, 484 U.S. 343 , 108 S.Ct. 614 , 98 L.Ed.2d 720 (1988), the Supreme Court held that the possibility of forum shopping is one factor a district court should take into account in exercising its discretion with respect to pendent jurisdiction. See id. at 351, 356 , 357 & n. 12, 108 S.Ct. at 619 , 622-23 & n. 12.

Cohill , to be sure, was not about whether a district court should retain jurisdiction, but rather about what a district court should do after deciding not to retain jurisdiction, i.e. whether the court should dismiss the case, which might end it due to statute of limitations problems in state courts, or whether the court should remand the case to the state court from whence it had been removed. But Cohill did hold that one factor the district court should consider in deciding whether to remand the case to state courts is whether plaintiffs are attempting to forum shop. We hold that the same principle applies when the district court decides whether to retain jurisdiction in the first place.

In sum, fairness to defendants along with judicial economy and convenience justify the district court’s decision to retain jurisdiction here.

IV. WAS THE IN LIMINE HEARING TAINTED BY UNFAIR PROCEDURES?

Before addressing the central issues in the case (regarding the admissibility of the opinions of the plaintiffs’ experts under Rules 702, 703 and 403), we must dispose of some preliminary questions. The plaintiffs contend that the district court should not have relied on the testimony of the defendants’ experts in excluding the testimony of the plaintiffs’ experts, because the defendants failed to designate these experts in accord with the time schedule announced by the district court. We review for abuse of discretion, and find none with respect to this ruling.

Plaintiffs also submit that the district court should not have relied on its scheduling order to exclude the testimony of some of plaintiffs’ proposed experts. We do not have to reach the merits of this contention, because plaintiffs did not point to the testimony of these experts in opposition to summary judgment.

A. Inadequate Opportunity to Depose Defense Experts

On June 4, 1991, the district court issued an order requiring defendants to “designate all trial expert witnesses” by April 17, 1992 and giving the parties “until June 17, 1992 to depose expert witnesses.” (emphasis added). On June 25, 1992, well after the deadline the district court had set for identification of experts, defendants submitted affidavits from a new set of experts. Defendants intended to have these experts testify at the August in limine hearing but not at trial. Because discovery was closed, plaintiffs were unable to depose them.

*739 Plaintiffs moved to prevent these experts from testifying. The district court never ruled on this motion, allowed the experts to testify at the in limine hearing, and relied on their testimony in excluding much of plaintiffs’ evidence. Plaintiffs contend that this was error. On a purely formal level, we disagree for, as defendants correctly assert, the district court’s June 1991 order setting deadlines for the identification of experts was only directed at trial witnesses and said nothing about witnesses who were to testify only at the in limine hearing. But that conclusion does not settle the question. By coincidence, plaintiffs’ more significant claim of error is encapsulated in a recent law review article by Professor Berger, who argues that “[e]ourts should not permit the defendant to obtain a hearing on a motion in limine by relying on affidavits from experts unless their identity and reports have been supplied to the plaintiff in the course of discovery and the plaintiff had an opportunity to depose [them].” Margaret A. Berger, Procedural Paradigms for Applying the Daubert Test, 78 Minn.L.Rev. 1345, 1372 (1994).

The district court certainly had the power to provide that the experts who were to testify at the in limine hearing be subject to discovery. See Berger, supra, at 1372 n. 141. (“A court undoubtedly has power pursuant to Federal Rule of Civil Procedure 16 to require parties to provide each other with expert reports and an opportunity for the deposition of experts retained in connection with in limine motions.”). And we generally agree with Professor Berger that because under Daubert a judge at an in limine hearing must make findings of fact on the reliability of complicated scientific methodologies and this fact-finding can decide the case, it is important that each side have an opportunity to depose the other side’s experts in order to develop strong critiques and defenses of their expert’s methodologies. See Paoli I, 916 F.2d at 854 (“[T]he detailed factual record requirement, firmly entrenched in our jurisprudence, requires adequate process at the evidentiary stage, particularly when a summary judgment may flow from it.” (internal citations omitted)). Given the “liberal thrust” of the federal rules, see Daubert, — U.S. at -, 113 S.Ct. at 2794 (internal quotation omitted), it is particularly important that the side trying to defend the admissibility of evidence be given an adequate chance to do so. Moreover, fairness suggests that each side should have an equal opportunity to depose the other side’s experts. 4

In this ease, however, where we had yet to articulate our specific concern with depositions prior to the ruling of the district court, we conclude that the district court did not abuse its discretion in allowing experts, who had provided affidavits, to testify at the in limine hearing without being deposed. 5 Moreover, the defendants submitted the affidavits of their experts four weeks prior to the hearing, so that the plaintiffs had a significant opportunity to prepare for the testimony of defendants’ experts. They also had extensive leeway at the in limine hearing, where the conservation of the time of the jury was not implicated, to consult with their experts regarding any new arguments presented by defendants’ experts. Under these circumstances, whatever unfairness to the plaintiffs that resulted from their inability to depose defendants’ experts is not sufficient to warrant a reversal for depositions and a new in limine hearing.

B. Exclusion of Opinion of Experts for Whom No. 26(b) (U) Statements Were Filed

In the June 4, 1991, scheduling order setting deadlines for designation of defense ex *740 perts discussed above, the district court required plaintiffs to designate on or before March 17, 1992, “all trial expert witnesses and provide defendants with the substance of their trial testimony and the basis of their opinions at the time of designation in accordance with Rule 26(b)(4).” On July 17, 1992, four months after the deadline for designating experts and two weeks before the scheduled trial date, the plaintiffs filed a joint pretrial memorandum in which they named three expert witnesses, Drs. Calesnick, Bar-sotti, and Zahalsky, who had not been identified in plaintiffs’ March 1992 Rule 26(b)(4) statements. Plaintiffs also named Drs. Nicholson and Jenkins as fact witnesses even though defendants argue that they have no connection with this case and are really expert witnesses. The district court excluded all of these witnesses on the basis that plaintiffs had not named them by March 17,1992.

Plaintiffs assert that because Drs. Cales-niek, Barsotti, and Zahalsky provided opinions in 1988 before the first summary judgment decision, the plaintiffs had designated them as experts before March 17, 1992, as required by the district court’s scheduling order. Moreover, plaintiffs argue, even if they violated the pre-trial order, the district court abused its discretion in sanctioning them by excluding the witnesses under the factors we set forth in Meyers v. Pennypack Woods Home Ownership Ass’n, 559 F.2d 894 (3d Cir.1977). We do not reach the merits of the district court’s ruling.

Plaintiffs did not rely on these witnesses to oppose summary judgment. In fact, the plaintiffs did not even list these witnesses as trial witnesses until July 17, nine days after they had filed their opposition to summary judgment. As a result, our consideration of whether the district court was correct to grant summary judgment does not depend on whether the district court properly excluded these witnesses. “Where a party opposing a motion for summary judgment has the burden of persuasion, and the moving party has identified sufficient facts of record to demonstrate that no genuine issue of material fact remains, the nonmoving party is obliged to identify those facts of record which would contradict the facts identified by the movant.” Childers v. Joseph, 842 F.2d 689 (3d Cir.1988). See also Celotex Corp. v. Catrett, 477 U.S. 317, 322 , 106 S.Ct. 2548, 2552-53 , 91 L.Ed.2d 265 (1986). If the plaintiffs wished to rely on the opinions of Drs. Cales-nick Barsotti, Zahalsky, Nicholson and Jenkins to oppose summary judgment, they were required to identify the relevant facets of this testimony in their opposition to summary judgment.

Nonetheless, because we are reversing the grant of summary judgment with respect to some plaintiffs and some issues, whether these witnesses can testify at trial may still matter on remand. We could therefore decide whether the district court’s original decision excluding these witnesses was an abuse of discretion, the matter having been properly put before us. However, we decline to do so, preferring to allow the district court to reconsider its decision in light of the additional time that will be available to the defendants to depose the witnesses.

In excluding the witnesses originally, the district court partly relied on the proximity of the trial date to the revelation by plaintiffs of their intent to call these experts. And proximity is indeed one important factor. 6 *741 When the plaintiffs named these experts in July of 1992, two weeks before trial, giving defendants a significant time to depose them would have required delaying a long scheduled trial in a complex case that had already gone on for years. But after remand, the district court will easily b'e able to give the defendants significant time to depose the plaintiffs. In sum, in light of the changed circumstances that will be present upon remand, we choose not to determine whether the district court’s original decision constituted an abuse of discretion.

Y. LEGAL STANDARDS FOR THE ADMISSIBILITY OF EXPERT OPINION

The district court excluded plaintiffs’ remaining experts under Rules 702, 703, and 403. While the district court, the plaintiffs, and the defendants basically agree on the legal standards to be applied in analyzing admissibility, they also have some significant differences which repeatedly surfaced throughout the disputes over the admissibility of particular testimony. Accordingly, before analyzing the admissibility of the testimony of particular experts, we will first provide an overview of the legal standards under which we will evaluate that admissibility. Such an overview is also important in light of the Supreme Court’s recent decision in Dau-bert, which, although it largely adopted the Rule 702 test we announced in Downing, works certain changes in our jurisprudence on the admissibility of expert opinions.

A. Rule 702

Rule 702 provides that:

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 thereto in the form of an opinion or otherwise.

1. Qualifications

Rule 702 has two major requirements. The first is that a witness proffered to testify to specialized knowledge must be an expert. We have interpreted this requirement liberally. See Paoli I, 916 F.2d at 855 . We have held that a broad range of knowledge, skills, and training qualify an expert as such. See id. at 855 . In Paoli I we ruled that the district court had abused its discretion in excluding the opinions of Drs. Barsotti, Za-halsky, and Nisbet. We explained that exclusion was not the proper remedy “simply because the experts did not have the degree or training which the district court apparently thought would be most appropriate.” Id. at 856 .

Defendants submit that when this court held that Rule 702 mandates a policy of liberal admissibility, all that it was doing was eliminating any requirement that a candidate have specific formal qualifications such as a degree in a particular field. They assert that we were simply directing the trial court to focus on the “substance of a proffered expert’s background and its fit with respect to the issue at hand.” Thus, defendants suggest, when a trial court evaluates an expert’s substantive rather than formal qualifications, it does not have to use a liberal policy of admission.

But this is incorrect. Rule 702’s liberal policy of admissibility extends to the substantive as well as the formal qualification of experts. We have eschewed imposing overly rigorous requirements of expertise and have been satisfied with more generalized qualifications. See Hammond v. International Harvester Co., 691 F.2d 646, 652-53 (3d Cir.1982) (holding that an engineer, whose only qualifications were sales experience in the field of automotive and agricultural equipment and teaching high school automobile repair, nevertheless could testify in a products liability action involving tractors); Knight v. Otis Elevator Co., 596 F.2d 84, 87-88 (3d Cir.1979) (holding that an expert could testify that unguarded elevator buttons constituted a design defect despite expert’s lack of specific background in design and manufacture of elevators). As we explain below, however, the level of expertise may affect the reliability of the expert’s opinion.

*742 2. Reliability

The second requirement of Rule 702 is that the expert must testify to “scientific, technical or other specialized knowledge [that] will assist the trier of fact.” Fed.R.Evid. 702. The Supreme Court recently evaluated the meaning of this requirement in Daubert and held that Rule 702 does not incorporate the common law Frye rule in which expert testimony is admissible only insofar as it is based on a technique generally accepted in the scientific community. Instead, the Supreme Court agreed with this Court’s basic holding in Downing that an expert’s testimony is admissible so long as the process or technique the expert used in formulating the opinion is reliable. See Daubert, — U.S. at -, 113 S.Ct. at 2794-95 (“That the Frye test was displaced by the Rules of Evidence does not mean, however, that the Rules themselves place no limits on the admissibility of purportedly scientific evidence. Nor is the trial judge disabled from screening such evidence. To the contrary, under the Rules the trial judge must ensure that any and all scientific testimony or evidence admitted is not only relevant, but reliable.”); Downing 753 F.2d at 1237. Daubert explains that the language of Rule 702 requiring the expert to testify to scientific knowledge means that the expert’s opinion must be based on the “methods and procedures of science” rather than on “subjective belief or unsupported speculation”; the expert must have “good grounds” for his or her belief. Daubert , — U.S. at-•, 113 S.Ct. at 2795 . In sum, Daubert holds that an inquiry into the reliability of scientific evidence under Rule 702 requires a determination as to its scientific validity. Id. at-, 113 S.Ct. at 2795 n. 9.

Daubert suggests several factors that a district court should take into account in evaluating whether a particular scientific methodology is reliable (i.e. scientifically valid), including the testability of the expert’s hypothesis (“whether it can be (and has been) tested”), Daubert , — U.S. at -, 113 S.Ct. at 2796 , whether the methodology has been subjected to peer review and publication, the frequency by which the methodology leads to erroneous results, the existence and maintenance of standards controlling the technique’s operation, and whether the methodology has been generally accepted in the scientific community. 7 See Daubert, — U.S. at-, 113 S.Ct. at 2796-97 . We listed very similar factors a district court should consider in Downing. See Downing, 753 F.2d at 1238-39. Daubert lists two factors not listed in Downing — whether a method produces testable hypotheses and the existence of standards controlling the technique’s operation, and Daubert leaves off its list several factors listed in Downing — the degree to which the expert testifying is qualified, the relationship of a technique to “more established modes of scientific analysis,” and the “non-judicial uses to which the scientific technique are put.” Downing, 753 F.2d at 1238-39; see Daubert, — U.S. at- -, 113 S.Ct. at 2796-97 . But Daubert specifically refuses to disavow any the particular factors we listed in Downing. See Dau-bert, — U.S.-, 113 S.Ct. at 2797 n. 12. And Daubert , like Downing, indicates that the inquiry as to whether a particular scientific technique or method is reliable is a flexible one. See id. at-, 113 S.Ct. at 2796 , Downing, 753 F.2d at 1238-39. We now make clear that a district court should take into account all of the factors listed by either Daubert or Downing as well as any others that are relevant. 8

3. Fit

In addition to reliability, Rule 702 requires that the expert’s testimony must *743 assist the trier of fact. As we put it in Downing, admissibility depends in part on “the proffered connection between the scientific research or test result to be presented and particular disputed factual issues in the case.” Downing, 753 F.2d at 1237. See Daubert, — U.S.-, 113 S.Ct. at 2795-96 (explicitly adopting the “fit” requirement of Downing). For example, animal studies may be methodologically acceptable to show that chemical X increases the risk of cancer in animals, but they may not be methodologically acceptable to show that chemical X increases the risk of cancer in humans. Daubert explains that, “ ‘[flit’ is not always obvious, and scientific validity for one purpose is not necessarily scientific validity for other, unrelated purposes.” Id. at-, 113 S.Ct. at 2796 . Thus, even if an expert’s proposed testimony constitutes scientific knowledge, his or her testimony will be excluded if it is not scientific knowledge for purposes of the case. “Rule 702’s ‘helpfulness’ standard requires a valid scientific connection to the pertinent inquiry as a precondition to admissibility.” Id. (emphasis added). For example, in order for animal studies to be admissible to prove causation in humans, there must be good grounds to extrapolate from animals to humans, just as the methodology of the studies must constitute good grounds to reach conclusions about the animals themselves. Thus, the requirement of reliability, or “good grounds,” extends to each step in an expert’s analysis all the way through the step that connects the work of the expert to the particular case.

4. The roles of judge and jury: how high a threshold for reliability?

Plaintiffs’ basic position in their challenge to each of the district court’s admissibility decisions is that the district court improperly usurped the role of the jury by the manner in which it analyzed admissibility. Plaintiffs first argue that, so long as they have made a prima facie showing that an expert’s testimony is reliable, it is up to the jury to evaluate the expert’s testimony. They assert that for the judge to evaluate competing evidence on reliability before determining admissibility would force plaintiffs to prove their case twice — once to the judge in order to persuade him or her to admit the evidence and once to the jury to persuade it of liability.

But plaintiffs’ position that a party who wants to introduce expert testimony must make only a prima facie showing that a technique is reliable (e.g., the testimony of a single qualified expert) is specifically foreclosed by precedent. Daubert holds that admissibility under Rule 702 is governed by Rule 104(a), which requires the judge to conduct preliminary factfinding, to make “a preliminary assessment of whether the reasoning or methodology underlying the testimony is scientifically valid,” and thus enables the judge to exclude evidence presented in plaintiffs’ prima facie case. 9 Daubert , — U.S. at -, 113 S.Ct. at 2796 . Similarly, we held in Downing that “it is plain that the proponent must make more than a prima facie showing .■.. that a technique is reliable.” Downing, 753 F.2d at 1240 n. 21. We added that, in contrast to the view that the admission of scientific testimony is a matter of conditional relevancy governed by Rule 104(b), “novel scientific evidence carries with it concerns over trustworthiness and reliability akin to those raised by offers of hearsay evidence. When there is a serious question of reliability of evidence, it is appropriate for the court to exercise to some degree an evidentiary screening function.” Id. Cf. In re Japanese Elec. Prod., 723 F.2d 238, 276-77 (3d Cir.1983) (implying that Rule 104(a) requires *744 judges to make factual evaluations in making Rule 703 determinations). 10

This does not megn that plaintiffs have to prove their ease twice — they do not have to demonstrate to the judge by a preponderance of the evidence that the assessments of their experts are correct, they only have to demonstrate by a preponderance of evidence that their opinions are reliable. 11 As the Supreme Court has explained in describing the effect of the preponderance standard of Rule 104(a) generally, “[t]he inquiry made by a court ... is not whether the proponent of the evidence wins or loses his case on the merits, but whether the evidentiary Rules have been satisfied.” Bourjaily v. United States, 483 U.S. 171, 175 , 107 S.Ct. 2775, 2778 , 97 L.Ed.2d 144 (1987).

The evidentiary requirement of reliability is lower than the merits standard of correctness. Daubert states that a judge ■ should find an expert opinion reliable under Rule 702 if it is based on “good grounds,” i.e., if it is based on the methods and procedures of science. A judge will often think that an expert has good grounds to hold the opinion that he or she does even though the judge thinks that the opinion is incorrect. As Dau-bert indicates, “[t]he focus ... must be solely on principles and methodology, not on the conclusions that they generate.” Daubert , — U.S. at -, 113 S.Ct. at 2797 . The grounds for the expert’s opinion merely have to be good, they do not have to be perfect. The judge might think that there are good grounds for an expert’s conclusion even if the judge thinks that there are better grounds for some alternative conclusion, and even if the judge thinks that a scientist’s methodology has some flaws such that if they had been corrected, the scientist would have reached a different result.

As we explained in Paoli I, “the reliability requirement must not be used as a tool by which the court excludes all questionably reliable evidence.” Paoli I, 916 F.2d at 857 . The “ultimate touchstone is helpfulness to the trier of fact, and with regard to reliability, helpfulness turns on whether the expert’s ‘technique or principle [is] sufficiently reliable so that it will aid the jury in reaching accurate results.’ ” DeLuca, 911 F.2d at 956 (quoting 3 J. Weinstein & M. Berger, Weinstein’s Evidence 702[03], at 702-35 (1988)). 12 A judge frequently should find an expert’s *745 methodology helpful even when the judge thinks that the expert’s technique has flaws sufficient to render the conclusions inaccurate. He or she will often still believe that hearing the expert’s testimony and assessing its flaws was an important part of assessing what conclusion was correct and may certainly still believe that a jury attempting to reach an accurate result should consider the evidence. See Paoli I, 916 F.2d at 857 (helpfulness requires more than bare logical relevance, but there is a strong preference for admission) (citing Downing, 753 F.2d at 1235).

The same standard of reliability extends to the step in the expert’s analysis that “fits” his or her conclusions to the case at hand. Once again, we emphasize that the standard is not that high. For example, in Paoli I, we held that testimony that PCBs cause liver cancer “fit” the case even in the absence of plaintiffs who had liver cancer, because an expert’s affidavit suggested that increased risk of liver cancer was probative of increased risk of other forms of cancer. See Paoli I, 916 F.2d at 858 . Nonetheless, the standard is higher than bare relevance. 13

In addition to arguing that the district court usurped the role of the jury by requiring more than a prima facie showing of reliability, plaintiffs submit that the district court mistakenly applied the reliability inquiry to expert testimony that should not have been subject to that inquiry. Both Daubert and Downing require that expert testimony be based on the methods and procedures, the processes and techniques of science. In Pao-li I we suggested that so long as any deviation the expert made from a rehable method merely constituted a change in the application of that method, the expert’s testimony remained based on a rehable scientific method. We though that ferreting out misapplication fell within the province of the jury, and that only if the expert so altered a rehable methodology as to skew it would the rehabihty inquiry of Downing be apphcable to the altered methodology. See Paoli I, 916 F.2d at 858 . Plaintiffs assert that the district court apphed the Downing inquiry to methodologies that merely constituted shght alterations of rehable methodologies.

However, after Daubert , we no longer think that the distinction between a methodology and its apphcation is viable. To begin with, it is extremely elusive to attempt to ascertain which of an expert’s steps constitute parts of a “basic” methodology and which constitute changes from that methodology. If a laboratory consistently fails to use certain quality controls so that its results are rendered unreliable, attempting to ascertain whether the lack of quahty controls constitutes a failure of methodology or a failure of apphcation of methodology may be an exercise in metaphysics. Moreover, any misapplication of a methodology that is significant enough to render it unreliable is likely to also be significant enough to skew the methodology.

As suggested, Daubert inters any need for us to make such a distinction, for Daubert’s requirement that the expert testify to scientific knowledge—conclusions supported by good grounds for each step in the analysis— means that any step that renders the analysis unrehable under the Daubert factors renders the expert’s testimony inadmissible. This is true whether the step completely changes a reliable methodology or merely misapplies that methodology. 14

*746 Finally, plaintiffs contend that the district court usurped the role of the jury by making admissibility decisions based on its disagreement with the conclusions of plaintiffs’ experts. Plaintiffs are correct, of course, that Daubert requires the judge’s admissibility decision to focus not on the expert’s conclusions but on his or her principles and methodology. See Daubert, — U.S. at-, 113 S.Ct. at 2797 . But we think that this distinction has only limited practical import. When a judge disagrees with the conclusions of an expert, it will generally be because he or she thinks that there is a mistake at some step in the investigative or reasoning process of that expert. If the judge thinks that the conclusions of some other expert are correct, it will likely be because the judge thinks that the methodology and reasoning process of the other expert are superior to those of the first expert. This is especially true given that the expert’s view that a particular conclusion “fits” a particular case must itself constitute scientific knowledge — a challenge to “fit” is very close to a challenge to the expert’s ultimate conclusion about the particular case, and yet it is part of the judge’s admissibility calculus under Daubert, 15

Thus, as we explained above, we think that the primary limitation on the judge’s admissibility determinations is that the judge should not exclude evidence simply because he or she thinks that there is a flaw in the expert’s investigative process which renders the expert’s conclusions incorrect. The judge should only exclude the evidence if the flaw is large enough that the expert lacks “good grounds” for his or her conclusions.

5. Confusion — Rule 403

In Downing we explained that under Rule 702, admissibility of scientific testimony turns not only on reliability but also on the possibility that admitting the evidence would overwhelm, confuse, or mislead the jury. See Downing, 753 F.2d at 1240. We held that in conducting this balancing inquiry, there is a presumption of helpfulness. See id. at 1241 (citing Japanese Elec. 723 F.2d at 279 ). The extent to which an adverse party has had notice and the opportunity to present his or her own experts is also relevant. Moreover, a district court cannot exclude a scientific technique as too confusing and overwhelming simply based on its conclusion that scientific techniques by their very nature confuse and overwhelm the jury. There must be something about the particular scientific technique such as its posture of mythic infallibility that makes it especially overwhelming. See id. at 1239 (overturning the district court’s exclusion of Dr. Nicholson’s meta-analysis because its conclusion that his testimony was overwhelming was based merely on its scientific nature and his credentials).

We explained in Downing that Rule 702 analysis partly incorporates Rule 403 analysis but leaves some room for Rule 403 to operate independently. See Downing, 753 F.2d at 1242. For example, we noted that a judge might use Rule 403 to exclude an expert’s critique of eyewitness testimony even though the critique met the requirements of Rule 702 if there was evidence of defendant’s guilt other than eyewitness testimony which would make efforts to criticize eyewitness testimony a waste of time.

In Daubert , the Supreme Court seems to have inverted our view that much of Rule 403 analysis conflates into Rule 702; rather the Court seems to have conflated the confusion/overwhelming impact prong of our Rule *747 702 analysis into its Rule 403 analysis. The Daubert Court did not mention the confusion/overwhelming prong when discussing Rule 702 but did provide support for application of essentially similar analysis under the rubric of Rule 403. The Court noted that Rule 403’s balancing test of the probative against the prejudicial value of evidence has a special role in cases involving expert witnesses. It stated that because expert evidence is often more misleading than other evidence, Rule 403 gives a judge more power over experts than over lay witnesses. See Daubert, — U.S. at-, 113 S.Ct. at 2798 . This, however, does not change our opinion that in order for a district court to exclude scientific evidence, there must be something particularly confusing about the scientific evidence at issue — something other than the general complexity of scientific evidence. 16

6. Procedural concerns regarding the Rule 403/702 balancing test

In Paoli I we held that Rule 403 is rarely appropriate as a basis of pre-trial exclusion, because a judge cannot ascertain potential relevance -until that judge has a virtual surrogate for a trial record. See Paoli I, 916 F.2d at 859-60 . There, many of the district court’s admissibility determinations were based in part on Rule 403; thus, it is important to determine whether the in limine hearing in the district court created the “virtual surrogate for a trial record” that we have required before exclusion is permissible. Plaintiffs argue that the in limine hearing did not create such a record because they did not have an opportunity to depose defendant’s experts ahead of time. As compared with a trial, plaintiffs assert, this made it much more difficult to impeach defendants’ experts and to bolster the claims of their own experts. We think that the in limine hearing conducted here was more than adequate as a basis for a Rule 403 determination — to hold otherwise given the extensive proceedings in the district court would be to require not a “virtual surrogate for a trial record” but a trial itself.

B. Rule 703

While Rule 702 focuses on an expert’s methodology, Rule 703 focuses on the data underlying the expert’s opinion. Under Rule 703:

The facts or data in the particular case upon which an expert bases an opinion or inference may be those perceived by or made known to the expert at or before the hearing. If of a type reasonably relied upon by experts in the particular field in forming opinions or inferences upon the subject, the facts or data need not be admissible in evidence.

Under our case law on Rule 703, “the proper inquiry is not what the court deems reliable, but what experts in the relevant discipline deem it to be.” In re Japanese Elect Prod., 723 F.2d at 277 . See also DeLuca, 911 F.2d at 952; Paoli I, 916 F.2d at 853 . We have held that the district judge must make a factual finding as to what data experts find reliable, see id. at 853 , and that if an expert avers that his testimony is based on a type of data on which experts reasonably rely, that is generally enough to survive the Rule 703 inquiry. See DeLuca, 911 F.2d at 952.

This view has been rejected by other courts in favor of the view that the judge has to make an independent evaluation of the reliability of the data. See, e.g., Christophersen v. Allied-Signal Corp., 939 F.2d 1106, 1113-14 (5th Cir.1991), cert. denied, — U.S. —, 112 S.Ct. 1280 , 117 L.Ed.2d 506 (1992); Ealy v. Richardson-Merrell, Inc., 897 F.2d 1159, 1161-62 (D.C.Cir.), cert. denied, 498 U.S. 950 , 111 S.Ct. 370 , 112 L.Ed.2d 332 (1990). As Judge Weinstein wrote in In re “Agent Orange” Prod. Liab. Litig., 611 F.Supp. 1223, 1245 (E.D.N.Y.1985):

*748 Rule 703 permits experts to rely upon hearsay. The guarantee of trustworthiness is that it be of the kind normally employed by experts in the field.... Nevertheless, the court may not abdicate its independent responsibilities to decide if the bases meet minimum standards of reliability as a condition of admissibility. See Fed. R. Ev. 104(a). If the underlying data are so lacking in probative force and reliability that no reasonable expert could base an opinion on them, an opinion which rests entirely upon them must be excluded.

Judge Weinstein’s view is extremely persuasive, and we are free to express our agreement with it because we think that our former view is no longer tenable in light of Daubert. See Stana v. School Dist. of Pittsburgh, 775 F.2d 122 , 130 n. 8 (3d Cir.1985) (noting that “[o]ur authority to align this court’s jurisprudence with Supreme Court teaching is clear.”). Daubert makes clear for the first time at the Supreme Court level that courts have to play a gatekeeping role with regard to experts. In stating that Rule 702 is the primary locus of the gatekeeping role, the Court implies that there are at least some secondary loci in other Rules. See Daubert, — U.S. at-, 113 S.Ct. at 2795 . By requiring the judge to look to the views of other experts rather than allowing the judge to exercise independent judgment, current Third Circuit case law eviscerates the judge’s gatekeeping role with respect to an expert’s data and instead gives that role to other experts. The gatekeeping role is reduced even further by our DeLuca holding that the opinion of one expert that a type of data is reliable will generally be enough to render that data reliable.

We now make clear that it is the judge who makes the determination of reasonable reliance, and that for the judge to make the factual determination under Rule 104(a) that an expert is basing his or her opinion on a type of data reasonably relied upon by experts, the judge must conduct an independent evaluation into reasonableness. The judge can of course take into account the particular expert’s opinion that experts reasonably rely on that type of data, as well as the opinions of other experts as to its reliability, but the judge can also take into account other factors he or she deems relevant.

We think that the standard is equivalent to Rule 702’s reliability requirement — there must be good grounds on which to find the data reliable. It makes sense that the standards are the same, because there will often be times when both Rule 702 and Rule 703 apply. 17 For example, expert testimony that uses animal studies must, to be admissible for the purpose of drawing conclusions about humans, meet Rule 702’s requirement of “fit.” This requires the court to assess whether there are good grounds for concluding that the animal studies demonstrate causation in humans. But for an expert to rely on animal studies (if their admissibility in evidence is not independently established), these studies must also meet the requirements of Rule 703 — the conclusions of these studies must be data which must be of a type reasonably relied upon by experts to analyze causation in humans. If the expert’s reliance on animal studies to draw conclusions about people constitutes a methodological flaw, it is a methodological “flaw” consisting of relying on data not of a type reasonably relied upon by experts. 18 It would be *749 problematic to conclude that this testimony was rehable enough to meet Rule 702 but not Rule 703—or vice versa. Testimony that is admissible as rehable under one Rule should not be inadmissible under another when the pohcy considerations underlying the two rules are the same. 19

We therefore think that when a trial judge analyzes whether an expert’s data is of a type reasonably rehed on by experts in the field, he or she should assess whether there are good grounds to rely on this data to draw the conclusion reached by the expert. Whether experts in the field rely on this type of data will simply continue to be a part of the judge’s analysis. After all, the only purpose of Rule 703 was to supplant the (former) common law rule that experts could not (reasonably) rely on inadmissible evidence, a notion demonstrated by the Rule 703 advisory committee to be contrary to real life.

C. Standard of Review

A district court’s ruling on admissibility of evidence is reviewed for abuse of discretion, “but to the extent the district court’s ruling turns on an interpretation of. a Federal Rule of Evidence our review is plenary.” DeLuca, 911 F.2d at 944. The threshold rule is, of course, one of deference. However:

the justifications for committing decisions to the discretion of the court are not uniform, and may vary with the specific type of decisions. Although the standard of review in such instances is generally framed as “abuse of discretion,” in fact the scope of review will be directly related to the reason why that category or type of decision is committed to the trial court’s discretion in the first instance.

United States v. Criden, 648 F.2d 814, 817 (3d Cir.1981). In several other areas, we have applied a heightened abuse of discretion review. See, e.g., Brody v. Spang, 957 F.2d 1108, 1115 (3d Cir.1992) (“We review a denial of a motion to intervene as of right for abuse of discretion, although this review is ‘more stringent’ than the abuse of discretion review we apply to a denial of a motion for permissive intervention”); Harris v. Pernsley, 820 F.2d 592, 597 (3d Cir.), cert. denied, 484 U.S. 947 , 108 S.Ct. 336 , 98 L.Ed.2d 363 (1987). Under the Harris standard, a denial of intervention as of right should be reversed if the district court “applied an improper legal standard or reached a decision that we are confident is incorrect.” (internal quotation omitted); United States v. Miele, 989 F.2d 659, 667 (3d Cir.1993) (looking rigorously to see whether the district court had closely scrutinized drug quantity when the court based its decision on the statements of an addict-informant); cf. Malchman v. Davis, 706 F.2d 426, 434 (2d Cir.1983) (holding that while a court of appeals normally reviews a class action settlement for abuse of discretion “this principle [of deference] applies with less force when, as here, settlement of a class action has been negotiated before the class has been certified.”).

While evidentiary rulings are generally subject to a particularly high level of deference because the trial court has a superior vantage point to assess the evidence, see Criden, 648 F.2d at 818 , evaluating the reliability of scientific methodologies and data does not generally involve assessing the truthfulness of the expert witnesses and thus is often not significantly more difficult on a cold record. Moreover, here there are factors that counsel in favor of a hard look at *750 (more stringent review of) the district court’s exercise of discretion. For example, because the reliability standard of Rules 702 and 703 is somewhat amorphous, there is a significant risk that district judges will set the threshold too high and will in fact force plaintiffs to prove their case twice. Reducing this risk is particularly important because the Federal Rules of Evidence display a preference for admissibility. 20 See Daubert, — U.S. at -, 113 S.Ct. at 2794 .

Factors like those just discussed influence the scope of our review of a district court’s discretion. For example, where rules display a preference for a particular outcome, our review of decisions under those rules is sometimes more searching. See United States v. Pennsylvania Dept. of Env’t Resources, 923 F.2d 1071, 1074 (3d Cir.1991) (“[T]he traditional discretion of the federal courts to decide whether to hear declaratory judgment cases is not limited by Colorado River [Water Conservation District v. U.S., 424 U.S. 800 , 96 S.Ct. 1236 , 47 L.Ed.2d 483 ] [ (1976) ] and Moses H. Cone [Memorial Hospital v. Mercury Construction Corporation, 460 U.S. 1 , 103 S.Ct. 927 , 74 L.Ed.2d 765 ], [(1983)] but will be subject to the liberal interpretation to be accorded the Declaratory Judgment Act. Given our correspondingly enlarged scope of review ..., the district court’s declination to exercise jurisdiction ... is not consistent with a sound exercise of judicial discretion.” (internal quotation omitted)).- Moreover, the likelihood of finding an abuse of discretion is affected by the importance of the district court’s decision to the outcome of the ease and the effect it will have on important rights. See Marroquin-Manriquez v. Immigration and Natur. Serv., 699 F.2d 129, 134 (3d Cir.1983) (abuse of discretion will only be found in discovery if there has been interference with a substantial right or fundamental unfairness at the trial has resulted.); cf. Mathews v. Eldridge, 424 U.S. 319, 334 , 96 S.Ct. 893, 902 , 47 L.Ed.2d 18 (1976) (the procedural safeguards required by due process increase as the importance of the decision being made increases).

We acknowledge that there is arguably a tension between the substantial deference normally accorded to rulings where the trial court has a superior vantage point and the preference for admissibility of the Federal Rules of Evidence. We resolve any such tension by holding that when the district court’s exclusionary evidentiary rulings with respect to scientific opinion testimony will result in a summary or directed judgment, we will give them a “hard look” (more stringent review, cf. Brody v. Spang, 957 F.2d at 1115) to determine if a district court has abused its discretion in excluding evidence as unreliable. 21

D. The Role of Pennsylvania’s Requirement of Reasonable Medical Certainty

Pennsylvania requires experts to testify that defendant’s actions caused plaintiff’s illness with a reasonable degree of medical certainty, see Cohen v. Albert Einstein Medical Ctr., 405 Pa.Super. 392 , 592 A.2d 720, 724 (1991), appeal-denied, 529 Pa. 644 , 602 A.2d 855 (1992) (“Expert testimony is admissible when, taken in its entirety, it expresses reasonable certainty that the accident was a substantial factor in bringing about that injury. The expert need not express his opinion in precisely the same language we use to enunciate the legal standard.”) (quoting Kravinsky v. Glover, 263 Pa.Super. 8 , 396 A.2d 1349 (1979) (citation within Kravinsky omitted)). The issues presented by this appeal require us to determine whether this rule governs in federal court. Because it will facilitate our discussion of the individual evidentiary rulings if we do so in the earlier portions of the opinion, we turn our attention to the question now.

*751 If the rule conflicts with federal rules and is “rationally capable of classification” as procedural rather than substantive, then, as a federal court, the district court ignores the rule and applies federal rules instead. See Hanna v. Plumer, 380 U.S. 460, 472 , 85 S.Ct. 1136, 1144 , 14 L.Ed.2d 8 (1965). But “the determination of whether a particular evidentiary ruling involves federal procedural law or state substantive law, can be difficult. Often admissibility issues overlap with substantive concerns such as standards of proof.” Schulz v. Celotex Corp., 942 F.2d 204, 207 (3d Cir.1991).

In Schulz , we refrained from deciding whether New Jersey’s requirement that a medical expert testify with “reasonable medical certainty” was procedural or substantive, because we held that the federal rules, like the New Jersey rules, also required that, at a minimum, a physician be able to “form an opinion with sufficient certainty so as to make a medical judgment,” id. at 209 , and that this was enough to decide the particular case. Thus, while Schulz did not require the specific words “reasonable degree of medical certainty,” testimony about “possibilities” or even “strong possibilities” was not admissible (though testimony made without qualifications was admissible). Id. at 208 . The issue is framed here primarily by an instance in which Dr. Sherman stated her conclusion with more certainty than a “strong possibility” (she stated that PCBs were a “significant factor” in Amber Burrell’s diminished ankle reflexes) but was unwilling to testify that she had reached that conclusion with a reasonable degree of medical certainty. (The issue also arises with respect to other pieces of evidence.) Thus, we must decide whether experts were required to testify with a reasonable degree of medical certainty.

We conclude that they were. Although if it were purely a rule of admissibility Pennsylvania’s standard would not apply in federal court, Pennsylvania’s rule also constitutes part of the plaintiffs burden of proof. As a rule of admissibility, Pennsylvania’s standard is in conflict with Rules 702 and 703 which require a reliable methodology and reliable data but nowhere require a reasonable degree of medical certainty. See United States v. Cyphers, 553 F.2d 1064, 1072 (7th Cir.), cert. denied, 434 U.S. 843 , 98 S.Ct. 142 , 54 L.Ed.2d 107 (1977) (upholding a district court’s admission of expert testimony that was not based on reasonable scientific certainty); cf. Daubert, — U.S. at—, 113 S.Ct. at 2794 (holding that the Frye rule requiring a methodology to be generally accepted was not part of the Federal Rules of Evidence); but see Grant v. Farnsworth, 869 F.2d 1149 (8th Cir.1989), cert. denied, 493 U.S. 898 , 110 S.Ct. 252 , 107 L.Ed.2d 202 (1989) (upholding the district court’s exclusion of expert testimony where a chiropractor could not state the causes of plaintiffs disability with a reasonable degree of medical certainty).

If Pennsylvania’s rule were only a standard of admissibility in conflict with Federal Rules of Evidence Pennsylvania’s rule would be “rationally capable of classification as procedural” and the Federal Rules of Evidence would govern. See, e.g., Salas v. Wang, 846 F.2d 897 , 904-06 (3d Cir.1988) (holding that under the Federal Rules of Evidence, an expert could testify to aggregate damages and that this rule was procedural, thus trumping New Jersey’s opposite rule). However, we think that Pennsylvania’s rule is more than a requirement of admissibility, but rather it is an element of plaintiffs burden of proof. In Cohen, the Pennsylvania Superior Court stated that:

When a party must prove causation through expert testimony the expert must testify with reasonable certainty that in his professional opinion, the result in question did come from the cause alleged ... [I]f the plaintiffs ... expert cannot form an opinion with sufficient certainty so as to make a [professional] judgment, there is nothing on the record with which a [fact-finder] can make a decision with sufficient certainty so as to make a legal judgment.

Cohen, 592 A.2d at 723 (quoting Kravinsky v. Glover, 263 Pa.Super. 8 , 396 A.2d 1349 (1979)) (internal quotations omitted). Thus, the Cohen court strongly implied that, even if admissible, testimony with less than a reasonable degree of medical certainty was insufficient to prove causation. See also id. *752 (stating that this was “[t]he degree of medical certainty necessary to prove causation.”).

As part of the burden of proof, then, Pennsylvania’s rule is a substantive one, not in conflict with Federal Rules of Evidence, and thus governs in federal court. See, e.g., Herber v. Johns-Manville Corp., 785 F.2d 79, 81-82 (3d Cir.1986) (holding that New Jersey’s requirement that a plaintiff must show a “reasonable medical probability” of future injury in order to recover for that injury governed in federal court). As a result, in analyzing the testimony of Dr. Sherman and other witnesses with respect to particular plaintiffs, we will uphold the district court’s grant of summary judgment for the defendants in those instances when the plaintiffs’ expert states the he or she could not testify that a plaintiffs illness was caused by PCBs with a reasonable degree of medical certainty. We turn to application of the foregoing principles to the record in this case.

VI. THE RELATIONSHIP BETWEEN PCBs AND PLAINTIFFS’ ILLNESSES

In order to sustain their traditional tort claims, i.e. other than for medical monitoring, the personal injury plaintiffs must show that they were exposed to the chemicals released by the defendants, that these chemicals can cause the types of harm they suffered, and that the chemicals in fact did cause them harm. We begin our analysis at the last step, because this is the step at which it is easier to winnow the claims on which the plaintiffs have presented admissible testimony.

Eleven of the Kohn/Klehr plaintiffs offered Janette D. Sherman, M.D. as an expert in internal medicine and toxicology., They proposed to have Dr. Sherman testify: (1) that they had been exposed to PCBs and other toxic chemicals; (2) that the toxic chemicals had caused adverse health effects in the various plaintiffs including high blood pressure, lung disease, asthma, elevated triglycerides, aortic sclerosis and eventual death, arthritis, respiratory infections and a history of miscarriages; and (3) that exposure to these chemicals poses an increased risk for various plaintiffs of developing cancer or other illnesses in the future. These plaintiffs also proffered Dr. Sherman to show that they require regular medical monitoring. The district court excluded Dr. Sherman’s testimony holding that she lacked the requisite expertise, that her methodology was unreliable, and that the probative value of her testimony was substantially outweighed by unfair prejudice and undue delay.

Nine plaintiffs, those represented by Bruce Hanes, called Dr. G. John DiGregorio to testify that the plaintiffs’ present injuries were caused by PCB exposure, that they have an increased and reasonable fear of future injuries, and that a medical monitoring program, which Dr. DiGregorio would design, should be implemented for plaintiffs. (A2634-35). Dr. DiGregorio received both a Ph.D. in pharmacology and an M.D. from Hahnemann Medical College and is a professor in pharmacology and medicine at Hahne-mann. The district court accepted Dr. Di-Gregorio as a qualified expert but excluded his testimony regarding personal injuries as unreliable.

We will begin our analysis by evaluating Dr. Sherman’s qualifications and will reverse the district court’s exclusion of Dr. Sherman as an unqualified expert. We will next briefly examine the district court’s exclusion of the immunological tests relied' on by Dr. Sherman for some of her diagnoses and will uphold that exclusion. We will then discuss the defendants’ claims that Drs. Sherman and DiGregorio engaged in too few of the standard diagnostic techniques of “differential diagnosis,” and failed seriously to consider the possibility of alternative causes of plaintiffs’ illnesses other than exposure to PCBs. After describing the critique, we will apply Daubert to analyze generally what makes a doctor’s method of differential diagnosis so unreliable as to be within the district court’s power to exclude it. We will then apply this framework to the assessment of individual plaintiffs that Dr. DiGregorio and Dr. Sherman made. We will uphold the district court’s exclusion of the testimony of Dr. Sherman and Dr. DiGregorio for the assessment of present injuries with the exception of Dr. Sherman’s testimony regarding Bessie Cunningham and Amber Burrell.

*753 A. Dr. Sherman’s Qualifications

Dr. Sherman is a 1964 graduate of the Wayne State University School of Medicine. Following a one year internship at Wayne State University Hospital and a two year fellowship funded by the National Institutes of Health to study diabetes, she practiced internal medicine full time from 1970-76. Since that time she has apparently been engaged in the full time practice of forensic medicine. Dr. Sherman is the author of 22 articles and of the book Chemical Exposure and Disease—Diagnostic and Investigative Techniques. She served for six years on a 16 person advisory committee to the EPA that was convened in connection with the adoption of the Toxic Substance Control Act—an Act largely concerned with PCBs. She was on the consulting staff of Harper Hospital in Detroit until 1988 and is still on its correspondence staff. A. 5086-87.

The district court held that Dr. Sherman was not qualified to testify as an expert. It explained that Dr. Sherman no longer practices general internal medicine—she does not have admitting privileges at any hospital and lacks an office where she can examine patients. Since 1976 most of her work has consisted of consulting for litigation; she has been involved in 800 cases and has been on the plaintiffs side in all but two. According to the district court, out of fifty patients she has examined who have been exposed to PCBs, she cannot remember one instance in which she found that PCBs had not caused adverse health effects. A. 5124.

Moreover, the district court noted, Dr. Sherman lacks board certification in any medical specialty including internal medicine and toxicology, and admits that she is not an expert in immunology, oncology, neurology, psychiatry, dermatology, epidemiology or exposure assessment. Although Dr. Sherman represents that she became an expert in toxicology through experience, reading, writing, teaching, evaluating patients and course work, she took no courses on toxicology in medical school, and the district court concluded that it did not observe anything in her testimony to indicate that her reading was of a serious nature.

Finally, the district court credited defendants’ witnesses who testified that, during the course of her testimony, Dr. Sherman made numerous basic medical errors in the fields of immunology, internal medicine and dermatology. The court referred to several examples cited by defendants’ experts. Dr. Jordan Fink, Professor of Medicine and Chief of the Allergy Immunology Division at the Medical College of Wisconsin, testified that Dr. Sherman incorrectly excluded essential hypertension, the name for the 80% of hypertension with an unknown cause, from her list of the causes of hypertension, incorrectly stated that cells in the immune system transport skin pigment cells, and referred to “myelin cells” which do not exist. Dr. James Leyden, a professor at the University of Pennsylvania Medical School, who has twice won the gold medal from the American Academy of Dermatology, stated that Dr. Sherman incorrectly asserted that one could not distinguish between ehloracne and acne based on their appearance—a distinction which, in his view, any qualified dermatologist could easily make. Drs. Michael Wei-tekamp, Robert Morgan and Paul Roediger pointed out similar errors in Dr. Sherman’s testimony. And Dr. Fink concluded that these gaps in knowledge were “major deficits. They—these items should be right in the forefront of knowledge of any internist, any resident, any medical student.” The district court concluded that these “major deficits” removed Dr. Sherman from the category of expert.

We disagree. We hold that Dr. Sherman, while arguably a relatively poor clinician and less than fully credible witness, qualifies as an expert. She was on the consulting staff of a hospital until she moved from Michigan in 1988. Since then she has had extremely broad experience in the field of toxic substances, has written extensively in the field and has consulted for the American Legion Science Panel—a group of experts analyzing scientific issues, such as dioxin poisoning, that affect veterans. The fact that most of her work since 1976 has been for plaintiffs in litigation may undermine her credibility but does not eradicate her expertise. For litigants to have access to experts, it may be *754 necessary for some experts to concentrate on litigation.

Dr. Sherman apparently made substantive mistakes in her testimony, but that testimony also demonstrated significant familiarity with the literature on PCBs; indeed her testimony in that area, which is central to this case, was not substantially challenged. Of course, Dr. Sherman was not merely testifying about PCBs but about the relation of PCBs to diseases in particular plaintiffs. In order to testify about this relationship, she needed to know about the various diseases, and it is mostly in this area that she made mistakes. But in a deposition lasting four days, it is hardly surprising that there were a few technical errors. Indeed, many of Dr. Sherman’s mistakes seem minor and unlikely to affect her ability to express an opinion (e.g. use of the word “thymus” instead of “thyroid,” use of the terms “myelin cells” instead of “myelin containing cells”).

Others of Dr. Sherman’s “mistakes” are hard to judge. For example, her reference to the movement of skin pigment referred to pigment that developed in nails and skin as a result of exposure to PCBs rather than to general. skin pigment (Á791, A5114) — and whether this type of pigment moves is unclear. However, some of the mistakes Dr. Sherman made, such as her contention that acne and ehloracne look the same, seem significant for diagnostic purposes. 22 Moreover, the number of mistakes she made was quite significant.

Nonetheless, the mistakes Dr. Sherman made do not render her unqualified as an expert for purposes of Rule 702. If the liberal standard of Rule 702 allows an engineer who teaches auto mechanics to testify in a products liability action about tractors, see Hammond v. International Harvester Co., 691 F.2d 646 (3d Cir.1982), supra at pp. 741-42, it surely allows a trained internist-who has spent significant time reviewing the literature on PCBs to testify as to whether PCBs caused illness in plaintiffs. In light of Hammond , Rule 702 does not require plaintiffs to hire separate experts on the relationship between PCBs and diabetes, PCBs and chlorac-ne, PCBs and cancer etc.; it is likely that none such exist, and the cost would be prohibitive. Under our liberal standards for qualifications of experts, the district court’s decision that Dr. Sherman was not a qualified expert was an abuse of discretion.

B. Rule 70S and Dr. Sherman’s Immunological Tests

The district court also held that some of the data on which Dr. Sherman relied, immunological test results from a chemical injury kit, were not the type of data reasonably relied-upon by experts in the field under Rule 703. The district court based its holding on the fact that the lab work was performed by an unaccredited laboratory and the fact that two of defendants’ experts testified that the tests were not important, would probably be more misleading than helpful, and had been misinterpreted. Plaintiffs cite Dr. Sherman’s testimony for the proposition that she did not rely on these tests. Defendants respond that these tests were often the only basis she cited for her conclusion that PCBs had caused plaintiffs’ injuries.

We will analyze the importance of the tests to Dr. Sherman’s testimony when we analyze her testimony with respect to the particular plaintiffs. See supra p. 749 n. 19 (explaining what a court should do when an expert testifies based in part on data deemed unreliable under Rule 703). However, since the plaintiffs do not defend the tests themselves, there is no real dispute that they may not be the basis for any of Dr. Sherman’s conclusions. If Dr. Sherman is allowed to testify, she will not be allowed to rely on these tests.

*755 C. Adequacy of Dr. Sherman’s and Dr. DiGregorio’s Use of Differential Diagnosis in Evaluating Causation of Present Illnesses

1. The district court’s critique

The district court also excluded Dr. Sherman’s testimony under the helpfulness prong of Rule 702. It believed defendants’ seven witnesses who opined that Dr. Sherman did not employ differential diagnosis, the basic method of internal medicine. Dr. Fink testified that an internist should inquire of the patient about his or her complaint(s), the history of the complaint(s), any other illnesses the patient might have, the patient’s occupation, and habits such as smoking. In his view, the internist should then use differential diagnosis to rule out possible alternative diseases and causes of these diseases by outlining the possibilities and then narrowing them down based on the doctor’s experience. For example, even if plaintiffs were exposed to PCBs and PCBs are dangerous, a doctor employing proper methodology could not conclude that PCBs were a substantial factor in the cause of an illness without thinking about other potential causes.

More particularly, according to Dr. Fink, before a doctor could reasonably conclude that PCBs contributed to an individual’s lung cancer, she would have to ask whether the individual was a smoker, and if that individual was a smoker, she would have to take steps to determine whether the individual’s lung cancer was more consistent with exposure to smoking or PCB exposure or both. She would also have to consider the history of the person’s exposure to both PCBs and cigarettes, and would have to perform laboratory tests and physical examinations to determine the likely cause(s).

According to the district court, Dr. Sherman failed to employ any of the ordinary methods doctors use to narrow down the cause of disease. She did not order standard laboratory tests to be performed on plaintiffs (ordering only unhelpful immunological tests from an unaccredited laboratory), physically examined only two of the thirteen plaintiffs about whom she offered opinions, 23 and took medical histories only from the two plaintiffs she examined physically. 24 The district court noted that Dr. Fink testified that it is standard medical practice to take medical histories and conduct physical examinations, and that it would not be appropriate to base a patient’s care solely on medical records from the patient’s prior physicians. Similarly, Dr. Weitekamp testified that without going through the entire process outlined above, it is difficult to make a complete diagnosis, and that Dr. Sherman did not perform a differential diagnosis.

The district court also relied on the fact that Dr. Sherman did not even consider alternative causes for various diseases. The court pointed to the testimony of Dr. Fink and Dr. Roediger, both of whom implied that *756 the real problem with Dr. Sherman’s methodology was not that she was unaware of possible alternative causes but that she did not really consider them. 25 Defendants note that, despite her testimony that she had employed differential diagnosis and had considered alternative causes, when asked about specific findings of causation, Dr. Sherman acknowledged that she had not looked at nor ruled out alternative causes for plaintiffs’ illnesses including high blood pressure, stomach cancer, asthma, episcleritis or anemia. Moreover, Dr. Roediger testified that, “I do not feel that the concept of ruling out the commonest causes of diseases was ever displayed in her testimony in a fashion that would portray her as a competent internist.” This was because she did not give evidence of having considered the common causes of many of the diseases she diagnosed. On this basis, the district court held that Dr. Sherman’s methodology was unreliable.

For similar reasons, the district court excluded the testimony of Dr. DiGregorio. In some ways Dr. DiGregorio’s methodology was more complete than was Dr. Sherman’s. He took a limited medical history, of each plaintiff — asking about their family medical history. He also had a blood test and a fat biopsy performed on each plaintiff to test for PCBs. Nonetheless, in other ways Dr. Di-Gregorio’s analysis was more limited than that of Dr. Sherman. Dr. DiGregorio did not even look at medical records of the plaintiffs much less examine them. With selective exceptions, Dr. DiGregorio’s methodology was to have each plaintiff mark their symptoms on a checklist, answer questions about their family medical history, discuss the answers they had given on the checklist, and obtain blood and fat samples to be tested for PCBs.

The district court applied the testimony offered by defendants about Dr. Sherman to the methodology of Dr. DiGregorio. 26 It found that differential diagnosis which is used to rule in or out alternative causes is the “hallmark” of internal medicine and that Dr. DiGregorio had not employed it. The court found that Dr. DiGregorio had intended to use differential diagnosis but had not performed the examinations and tests that he as well as defendants’ experts thought were necessary to render “a correct, final diagnosis.” The court found that Dr. DiGregorio admitted that, until he physically examined plaintiffs, he would not even be able to know if they suffered any injuries caused by PCBs. Yet Dr. DiGregorio did not perform these tests or examinations in the four years between his 1988 deposition and his 1992 deposition, although he had testified in 1988 that he planned to perform these tests and exams and that they' could be completed in six months.

According to the district court, Dr. DiGre-gorio’s failure to perform these tests not only left him with little information about plaintiffs’ actual injuries, it left him with little information about causation. The court cited examples of Dr. DiGregorio being unaware of activities engaged in by plaintiffs that were *757 possible alternative causes of their injuries, such as excessive alcoholic drinking as an alternate cause of Craig Brown’s elevated LDH, and taking the birth control pill Or-tho-Novum as an alternate cause of Cathlene Brown’s bruising. The court observed that Dr. DiGregorio had simply concluded that all of the plaintiffs’ symptoms were caused by exposure to PCBs “until proven otherwise.”

As a result of the foregoing, the court found that Dr. DiGregorio’s methodology was incomplete and therefore would not be helpful to the jury under Rule 702. In the court’s view, by failing to consider alternative causes, Dr. DiGregorio’s testimony would force the jury arbitrarily to choose one cause from many possible causes and, because Dr. DiGregorio did not examine plaintiffs’ to determine that they actually had injuries, his methodology was rendered incomplete and unreliable. Finally, because Dr. DiGregorio failed to perform physical exams or many tests, the court determined that the type of data he relied on was not a type experts reasonably rely on under Rule 703. It noted in conclusion that “[t]he only opinion that Dr. DiGregorio is consistent in holding throughout his deposition is that the Plaintiffs’ anxiety, fear of cancer, depression and abnormal lab values were caused by exposure to PCBs.”

2. The plaintiffs’ response

The plaintiffs attempt to revive the methodologies of Drs. Sherman and DiGregorio by arguing that the requirements of differential diagnosis are not nearly as rigorous as the district court found them to be. Plaintiffs point out that Dr. Sherman testified at the in limine hearing that she had used differential diagnosis since she was in medical school, and that she had done so here even with regard to those plaintiffs whom she did not examine. More particularly, she looked at medical records, which contained the results of physical exams performed by other physicians (and medical histories taken by other physicians), and at illnesses across the population, and then applied the medical literature to narrow down the cause. Dr. Sherman also testified that she did consider possible alternative causes before reaching her conclusions.

Similarly, while it is true that Dr. DiGrego-rio testified that in order to render a final opinion he would have to perform more tests, he also testified that, absent further evidence, he had enough information to conclude with a reasonable degree of medical certainty that PCBs were the cause of plaintiffs illnesses. He explained:

When a doctor does a history and there’s a cause of a problem, whether it’s pneumonia, pulmonary tract infection, the differential diagnosis — the leading differential diagnosis at the time is what the doctor suspects is the immediate cause of the problem. And it’s up to the doctor to proceed after that point and to prove either by physical examination or more tests whether his diagnosis is correct. So, as far as I’m concerned the symptoms that these patients gave me in the history are due to PCB exposure. Until I have the opportunity to continue on with my investigation they all will have, within a reasonable degree of medical certainty, that type of symptom.

Thus, he concluded, at least as a preliminary manner, that until performance of physical exams or more tests, PCBs were the cause of plaintiffs’ illnesses. And in reaching this conclusion, he stated that he followed normal methodology — “[t]he differential diagnosis you make with whatever information you have at the time. What you do now is to prove your diagnosis.”

In essence, plaintiffs’ argument is that what Dr. DiGregorio was saying was that differential diagnosis is an ongoing process of making judgments about causation and then adapting those judgments as new information is acquired. Dr. DiGregorio then, was asserting that, when he made his “preliminary” differential diagnosis, he had sufficient information (his knowledge of plaintiffs’ exposure to PCBs, his knowledge of the effects of PCBs, and the medical histories he took) reliably to conclude that PCBs were more likely than not a cause of plaintiffs’ illnesses even after considering other possibilities. Thus, the question becomes whether the district court abused its discretion in concluding *758 that Dr. Sherman and Dr. DiGregorio’s methodology of assessing causation was unreliable because of the limited information they had gathered.

3. Rule 702 Analysis

The defense experts’ criticisms of the methodologies of Drs. Sherman and DiGre-gorio were specific criticisms of the manner in which they each engaged in differential diagnosis (to the point where they asserted that Drs. Sherman and DiGregorio had not even employed differential diagnosis). Unfortunately, most of the Daubert factors— testability, general acceptance, peer review, and degree of production of errors, see supra at 742 n. 8, are of only limited help in assessing whether the methodologies of Drs. Sherman and DiGregorio are reliable (i.e., scientifically valid). We find that the final Daubert factor, the existence of standards controlling the technique’s operation, is slightly more helpful in deciding whether the methodologies of the plaintiffs’ doctors are reliable.

While applying the first Daubert factor, differential diagnosis can be considered to involve the testing of a falsifiable hypothesis (e.g. that PCBs caused a plaintiffs cancer) through an attempt to rule out alternative causes, that methodology involves far more elements of judgment than does a scientific study attempting to test a more general scientific proposition. While an important aspect of assessing scientific validity (and therefore evidentiary reliability) is the ability of other scientists to test or retest a proponent’s theory, differential diagnosis involves assessing causation with respect to a particular individual. This merely makes it a different type of science than science designed to produce general theories; it does not make it unreliable science. Although in terms of the remaining Daubert factors, differential diagnosis generally is a technique that has widespread acceptance in the medical community, has been subject to peer review, and does not frequently lead to incorrect results, it is a method that involves assessing causation with respect to a particular individual. As a result, the steps a doctor has to take to make that (differential) diagnosis reliable are likely to vary from case to case; the information a doctor needs in order to reliably assess the cause of a patient’s lung cancer is often very difficult from the information needed to assess the cause of a patient’s back or heart trouble.

Thus, although differential diagnosis is a generally accepted technique, no particular combination of techniques chosen by a doctor to assess an individual patient is likely to have been generally accepted. But unlike a methodology used in conducting a scientific study, lack of general acceptance is not a sign of unreliability, it is merely a result of the fact that the medical community will rarely have considered the reliability of a particular process of differential diagnosis used in an individual case. Nor is it likely that the particular combination will have been published and subject to peer review, because a particular version of differential diagnosis will rarely be of general interest to the medical community. However, to the extent that a doctor utilizes standard diagnostic techniques in gathering this information, the more likely we are to find that the doctor’s methodology is reliable. For these reasons, we must be flexible in conducting our Daubert inquiry.

We first note that defendants presented evidence that the methodologies of Drs. Sherman and DiGregorio were not generally accepted. Defendants’ experts explained that a reliable differential diagnosis generally requires a physical examination of the patient, a review of medical records, taking a medical history and conducting of laboratory tests, and always requires careful consideration of alternative causes. We agree with the defendants that performance of physical examinations, taking of medical histories, and employment of reliable laboratory tests all provide significant evidence of a reliable differential diagnosis, and that their absence makes it much less likely that a differential diagnosis is reliable. 27 We reach *759 this conclusion in part based on the persuasive testimony of defendants’ experts that these techniques are generally accepted as a standard part of differential diagnosis, and that such techniques significantly reduce the likelihood of erroneous results. We also base our conclusion on the fact that all of the experts agree that at the core of differential diagnosis is a requirement that experts at least consider alternative causes — this almost has to be true of any technique that tries to find a cause of something. Moreover, performance of standard diagnostic techniques provides prima facie evidence that a doctor has considered such causes and has attempted to test his or her initial hypothesis as to cause.

But, we agree with the plaintiffs that a doctor does not always have to employ all of these techniques in order for the doctor’s differential diagnosis to be reliable. Dr. Di-Gregorio testified that in reaching his conclusions, he followed normal methodology— “[t]he differential diagnosis you make with whatever information you have at the time.” What Dr. DiGregorio was saying was that differential diagnosis is an ongoing process of making judgments about causation and then adapting those judgments as new information is acquired. He was asserting that even without examining the plaintiffs or conducting laboratory tests, he had sufficient information (his knowledge of plaintiffs’ exposure to PCBs, his knowledge of the effects of PCBs, and the medical histories he took) to reliably conclude that PCBs were more likely than not a cause of plaintiffs’ illnesses even after considering other possibilities. 28

Put differently, we agree with the view of Drs. Sherman and DiGregorio that sometimes differential diagnosis can be reliable with less than full information, and, to the extent that the district court concluded otherwise, we hold that it abused its discretion. To provide a simplistic example, imagine a patient who comes in with medical records that include x-rays showing a fractured arm and who tells the doctor that he hurt the arm in a biking accident; the doctor could reliably conclude that the patient had a fractured arm caused by a biking accident even without physically examining the patient or taking a *760 medical history. 29 The biking accident is so much more likely to have been the cause of the fracture than anything else that there is no need to examine alternatives. See Peteet v. Dow Chem. Co., 868 F.2d 1428, 1431-32 (5th Cir.), cert. denied, 493 U.S. 935 , 110 S.Ct. 328 , 107 L.Ed.2d 318 (1989) (allowing a doctor to testify as to causation based on an examination of medical records, of the plaintiffs deposition, and of the medical literature). 30

In light of the foregoing discussion, we conclude that where Dr. Sherman or Di-Gregorio offered an opinion as to the source of a party’s illness, the district court abused its discretion in excluding that opinion under Rule 702 unless either (1) Dr. Sherman or DiGregorio engaged in very few standard diagnostic techniques by which doctors normally rule out alternative causes and the doctor offered no good explanation as to why his or her conclusion remained reliable, or (2) the defendants pointed to some likely cause of the plaintiffs illness other than the defendants’ actions and Dr. Sherman or DiGrego-rio offered no reasonable explanation as to why he or she still believed that the defendants’ actions were a substantial factor in bringing about that illness.

In response, of course, the doctor could offer an explanation based on the plaintiffs’ medical history, laboratory tests or any other reliable source. 31 Cf. Cantrell v. GAF *761 Corp., 999 F.2d 1007 , 1012-13 (6th Cir.1993) (applying Daubert to uphold the admissibility of the testimony of a doctor that asbestos had caused the plaintiffs’ laryngeal cancer based on the unusually high incidence of this cancer among the plant workers despite the fact that the defendants had pointed to smoking as a possible alternative cause of the cancer). Thus, to avoid exclusion of his or her opinion, Dr. Sherman or DiGregorio did not have to conduct every possible test to assess whether his or her view was correct so long as he or she employed sufficient diagnostic techniques to have good grounds for his or her conclusion that the plaintiffs illness was caused by PCBs.

The conclusion that the opinion of a doctor who has engaged in few standard diagnostic techniques should be excluded unless the doctor offers a good justification for his or her conclusion is supported by the case law. Cf. In re Agent Orange Prod. Liab. Litig., 611 F.Supp. 1223, 1234-38, 1246-47 (E.D.N.Y.1985) (excluding testimony by a doctor, based on his evaluation of form statements signed by the plaintiffs in which they indicated that they had been exposed to Agent Orange and then checked off their symptoms, that Agent Orange had caused plaintiffs’ injuries at least absent evidence of an alternative cause); Arnold v. Loose, 352 F.2d 959 (3d Cir.1965) (holding that Pennsylvania law precludes an expert from testifying based on guess or conjecture, and thus that the district court had properly excluded a doctor’s opinion that an individual had lapsed into a diabetic coma before an accident when the doctor had failed to rule out the alternative possibility that the brain injury suffered during the accident had caused the coma); Viterbo v. Dow Chem. Co., 826 F.2d 420 (5th Cir.1987) (excluding doctor’s testimony that the defendant had caused plaintiffs illness, because the doctor had relied on a medical history that omitted important information, the doctor had admitted the ailments could have a number of causes, and there was no good evidence that the" defendant had caused these types of illnesses). 32

We acknowledge that some of these cases can be read to indicate that an expert’s opinion on causation should always be excluded absent employment of standard diagnostic techniques to rule out alternative causes. See also Carroll v. Litton Sys., Inc., 1990 WL 312969 , *18, 19-20, 20,1990 U.S.Dist. LEXIS 16833, *49, 53, 55 (W.D.N.C.1990) (excluding testimony of medical doctors where they failed to employ differential diagnosis and *762 rule out alternative causes). However, for the reasons explained above, we think that there will be some cases in which a physician can offer a reliable differential diagnosis without examining the patient, looking at medical records, taking a medical history, and performing laboratory tests.

We also think that the view of defendants’ experts that a physical examination and review of medical records are integral parts of differential diagnosis is not based solely on their claim that these diagnostic techniques are necessary to assess causation but also on a view that these techniques are necessary to determine what illness a patient has contracted in the first place. In In re “Agent Orange”, 611 F.Supp. at 1246 (E.D.N.Y. 1985), Judge Weinstein faced a similar situation. In the end, he excluded a doctor’s testimony as to causation, where the evidence of the patients’ illnesses stemmed solely from a checklist they had filled out. He explained, “here we have statements of problems ranging from baldness to the most serious cancers.... While courts allow reliance on patient statements, they are based upon a personal history corroborated by medical records, a physical examination and medical tests.” Id. He added, “[h]ere, the usual inducement to candor with a physician — the hope of successful treatment or diagnosis— was wholly lacking. Cf. Fed.R.Evid. 803(4). Instead, plaintiffs had every incentive to be overinclusive in describing their symptoms....” Id.

Judge Weinstein was correct. We do not doubt the propriety of a medical report prepared just for litigation purposes, but a physician who evaluates a patient in preparation for litigation should seek more than a patient’s self-report of symptoms or illness and hence should either examine the patient or review the patient’s medical records simply in order to determine that a patient is ill and what illness the patient has contracted. Thus, we are satisfied "that where Drs. DiGregorio or Sherman based their conclusion as to a plaintiffs symptoms solely on the plaintiffs self-report of illness in preparation for litigation, the district court acted within its discretion in excluding the testimony as based on an unreliable source of information.

On the other hand, we think that evaluation of the patient’s medical records, like performance of a physical examination, is a reliable method of concluding that a patient is ill even in the absence of a physical examination. It is hornbook law that medical records are admissible for such a purpose — “[a]s to hearsay symptoms told by third persons, ..., where the information is that of an attending nurse or physician having personal observation and an interest in learning and describing accurately, there seems to be every reason for admitting testimony based in part on this.” 3 John Henry Wigmore, Wig-more on Evidence § 688(4) (Chadbourn ed. 1970). See also Cohen v. Albert Einstein Medical Ctr., 405 Pa.Super. 392 , 592 A.2d 720 (1991) (holding that the testimony of a doctor who relied on medical records without conducting a physical exam was admissible proof that the plaintiff suffered from a psychiatric disorder); Peteet v. Dow Chem. Co., 868 F.2d 1428, 1432 (5th Cir.1989) (allowing testimony of a doctor who had not physically examined the plaintiff and noting that, “[a] personal examination of the person or object of the expert’s testimony is not required under Fed. R.Evid. 703”). Hence, we think that generally, a doctor only needs one reliable source of information showing that the plaintiff is ill and either a physical examination or medical records will suffice — but the doctor does need at least one of these sources. .

Moreover, as we explained above, in order to establish causation sufficiently reliable for us to overturn the district court’s decision as an abuse of discretion, where Drs. Sherman or DiGregorio engaged in few of the standard procedures of differential diagnosis, they had to offer a good explanation as to why their conclusion remained reliable. Where they did employ such standard techniques, they still had to offer such an explanation if the defendants pointed to some likely cause of the plaintiffs illness other than the defendants’ actions.

We now apply these principles to the opinions of Drs. DiGregorio and Sherman regarding particular plaintiffs.

*763 a. The opinion of Dr. DiGregorio

As we have noted, Dr. DiGregorio did not examine the plaintiffs or review their medical records; he simply relied on their answers to a questionnaire he gave them. Under the rule we have articulated above, see supra pp. 763-64, Dr. DiGregorio could not even reliably conclude that the plaintiffs had any illness. In fact, Dr. DiGregorio admitted that without examining the plaintiffs he could not state to a reasonable degree of medical certainty that they had been injured as a result of exposure to PCBs. We now uphold the district court’s decision to exclude his testimony as within that court’s discretion. Dr. DiGregorio simply lacked the foundation to make the judgments he did. For that reason alone, the district court was correct to grant summary judgment against the Hanes plaintiffs on their personal injury claims.

Moreover, inasmuch as he did not engage in standard diagnostic techniques, we presume that Dr. DiGregorio’s conclusions as to causation were unreliable unless he offered good grounds for concluding that they were correct. See supra pp. 761-62. But Dr. DiGregorio stated that his preliminary assumption was that all of the plaintiffs’ injuries that they checked off in the questionnaire were caused by PCBs, and he did not explain what justified this assumption. For example, Dr. DiGregorio testified that PCBs caused Cathlene Brown’s bruising but admitted that the birth control pills she took could also have caused the bruising. He explained that in order to confirm his diagnosis he would have to “investigate with her when did she start the bruising, how long has she been bruising, does she have any bruising now, and how long has she been on the contraceptives,” but noted that his diagnosis would remain unchanged until he could conduct this investigation. Dr. DiGregorio nonetheless provided no reason to explain why he thought that PCBs were more likely to have caused Brown’s bruising than contraceptives. 33 His analysis of other Hanes plaintiffs was similarly flawed, as explained in the margin.

Presumably, Dr. DiGregorio’s general reasoning was that- all of the plaintiffs had significant exposure to PCBs, that PCBs are a known cause of the illnesses of the various plaintiffs, and that, in the absence of any indication that any plaintiff had been exposed to something more likely to cause his or her illness than PCBs, it was reasonable to conclude that PCBs were a likely cause. But Dr. DiGregorio did not point to any evidence showing that PCB exposure was so likely to produce the type of illnesses the plaintiffs *764 had in comparison to other possible causes to which plaintiffs had likely been exposed that it was rehable to conclude that PCBs were the cause without further analysis. We think that the district court essentially, and properly, read Dr. .DiGregorio’s testimony as showing that his opinion that PCBs caused plaintiffs’ illnesses was only a hypothesis which he had yet to attempt to verify or disprove by subjecting it to the rigors of scientific testing. Thus, the district court properly excluded Dr. DiGregorio’s testimony with respect to the Hanes plaintiffs’ personal injury claims and correctly granted summary judgment against them on these claims.

b. The opinion of Dr. Sherman

Unlike Dr. DiGregorio, Dr. Sherman reviewed the medical records of all of the plaintiffs on whose behalf she testified (the Kohn/ Klehr plaintiffs). Moreover, she physically examined two plaintiffs (and took their medical history). Thus, she had a rehable source of information that each of the plaintiffs was actually ill. Dr. Sherman also had information on the history of PCBs in the Paoli area, the level of PCBs in plaintiffs’ soil, some evidence of the level in their blood (for many plaintiffs), and the medical histories of the plaintiffs that were contained in their medical records. Nonetheless, with respect to those plaintiffs whom Dr. Sherman did not examine, she had little, if any data with which to rule out alternative causes of their illnesses. She did not take medical histories herself. There is no evidence in the record showing the extent to which medical histories were present in the plaintiffs’ medical records, and she performed no rehable laboratory tests.

The district court excluded Dr. Sherman’s testimony as a whole, on the grounds that her testimony demonstrated that she never considered alternative causes and always found PCBs to be the cause. If the district court was correct, then we would likely uphold its exclusion even if the defendants did not point to specific alternative causes that Dr. Sherman failed to consider for those plaintiffs whom she did examine. Given the number of plaintiffs in this case, we think that if there were no evidence that Dr. Sherman ever considered alternative causes, it would be reasonable for the district court to have presumed that she did not consider alternative causes even for those plaintiffs whom she examined and for whom she had significant data about alternative possibilities.

However, as we read Dr. Sherman’s testimony, when she said that she did not rule out other causes, that does not necessarily mean that she failed to consider them, but rather only that she did not rule them out as a possible cause or a partial cause in combination with PCBs, which she concluded were one significant factor in plaintiffs’ illnesses. More particularly, where defendants point to evidence that Dr. Sherman acknowledged failing to rule out alternative causes, Dr. Sherman was merely admitting that there might have been other causes while stating that PCBs were a significant cause.

Plaintiffs point to several instances in which Dr. Sherman specifically considered alternative causes. She specifically considered Charles Stanbach’s smoking as a possible cause of his esophageal cancer before ruling it out based on when he had stopped smoking and the types of changes he had in his cells. Moreover, she declined to say with a reasonable degree of medical certainty that the redness around George Burrell’s eyes had been caused by PCB exposure even though PCBs were associated with such redness in the literature — she explained that she could not rule out alternative causes. For similar reasons she refused to say that Priscilla Burrell’s headaches were caused by PCBs, and she opined that Monica Hilton’s skin reaction was caused by poison ivy rather than exposure to PCBs. Dr. Sherman further testified that she could not state with a reasonable degree of medical certainty that PCBs caused ringing in the ears even though three of the plaintiffs complained of such ringing.

In view of these facts, we will not uphold the district court’s blanket exclusion of Dr. Sherman’s testimony without examining her testimony concerning particular plaintiffs. Applying the Daubert analysis we conducted above, see supra at p. 761, unless Dr. Sherman presented a good explanation for why *765 she could reasonably testify that the illnesses of the plaintiffs whom she did not examine were caused by PCBs, the district court was within its discretion in excluding Dr. Sherman’s testimony. With respect to those plaintiffs whom Dr. Sherman did examine, we conclude that she employed a sufficient number of standard diagnostic techniques that the district court should have presumed that her testimony was reliable. Thus, Dr. Sherman’s testimony is admissible with respect to these plaintiffs unless the defendants pointed to particular potential alternative causes and she was unable to explain why she thought these alternatives had not caused the plaintiffs’ illnesses. See supra at pp. 760-61. In analyzing the adequacy of Dr. Sherman’s explanations, we will weigh in the balance Dr. Sherman’s somewhat dubious expertise — a factor we have deemed important under the Supreme Court’s flexible Daubert inquiry. See supra at pp. 743-44.

Defendants also contend that even if Dr. Sherman’s testimony is admissible, it is insufficient to allow plaintiffs to survive summary judgment. Because plaintiffs did not rule out alternative causes, defendants submit, her testimony cannot establish that defendants’ PCBs more likely than not caused plaintiffs’ injuries. We disagree, for assuming that Dr. Sherman’s testimony is admissible, her assertion that PCBs caused plaintiffs’ medical injuries is sufficient to establish a genuine issue of material fact on causation. However, for reasons we have explained supra pp. 751-52, defendants are correct that under Pennsylvania law they are entitled to summary judgment where Dr. Sherman refused to state her conclusions with a reasonable degree of medical certainty or its substantive equivalent, and for reasons we will explain infra, they are entitled to summary judgment where the illness to which Dr. Sherman attested was asymptomatic. See infra pp. 785-86 & n. 51 (concluding as part of an analysis of Pennsylvania’s law on medical monitoring, that Pennsylvania does not recognize a cause of action for physical changes which are asymptomatic). Thus, in analyzing Dr. Sherman’s testimony with respect to the particular plaintiffs, we will affirm the district court’s grant of summary judgment with respect to some of the plaintiffs’ claims without reaching the issue of the reliability of Dr. Sherman’s methodology.

i. Bessie Cunningham

Dr. Sherman examined Bessie Cunningham, took her medical history and reviewed her medical records. Sherman testified that PCBs significantly contributed to Cunningham’s headaches, sinusitis, blurred vision, arthritis, hypertension, peripheral swelling (edema) and nerve damage, and hysterectomy following three pregnancy losses. However, defendants offer a number of reasons why Dr. Sherman’s testimony is not admissible and we turn to these contentions now.

First, Dr. Sherman’s view that PCBs cause headaches partly relied on studies of the Yusho victims, studies which the district court properly excluded under Rule 403. See infra p. 783. But the fact that these studies would be inadmissible as substantive evidence before a jury does not render them unreliable as a source of data for experts who are in better position to evaluate (and discount them), and we have not deemed them to be unreliable as a source of such data. See infra p. 783. Thus, we do not think that Dr. Sherman’s reliance on this data renders her entire methodology as to the source of Bessie Cunningham’s headaches unreliable. Nor did the defendants point to an alternative source of these headaches that Dr. Sherman failed to consider. Thus, defendants did not put forth a basis for excluding Dr. Sherman’s testimony regarding these headaches.

Dr. Sherman also testified that PCBs caused Bessie Cunningham’s arthritis/arthralgia. She stated that Cunningham probably had both arthritis and arthralgia and that, of the types of arthritis, Bessie Cunningham probably had osteoarthritis. Defendants submit that Dr. Sherman’s testimony is fatally inconsistent insofar as she opined that the only known cause of osteoarthritis is wear and tear. But defendants did not ask Dr. Sherman why she concluded that PCBs caused Mrs. Cunningham’s arthritis if the only known cause of osteoarthritis is wear and tear; nor did they ask her how she eliminated wear and tear as a likely cause. *766 They also did not ask about possible alternative causes of arthralgia. Thus, they did not establish a basis for excluding Dr. Sherman’s testimony that PCBs caused Bessie Cunningham’s arthritis/arthralgia.

Dr. Sherman also opined that PCBs caused Cunningham’s hypertension by elevating her lipid levels or through some other mechanism. Although she acknowledged that tests did not show elevated lipid levels, she stated that the tests were not necessarily an accurate indicator because lipid levels change rapidly and added that the PCBs might have caused the hypertension through a mechanism other than raised lipid levels. In any case, we have no reason to believe that, for differential diagnosis to be a reliable methodology, all of the data a doctor acquires must point to the same conclusion. The lipid tests were unreliable at all events, see supra p. 756, and thus we do not think that the fact that the lipid tests failed to support Dr. Sherman’s conclusions inherently renders those conclusions unreliable. Finally, defendants did not point to any alternative causes of Cunningham’s hypertension that Dr. Sherman failed to rule out. Thus, we think that her testimony that PCBs caused Bessie Cunningham’s hypertension should not have been excluded.

Although Dr. Sherman initially testified that PCBs contributed to Mrs. Cunningham’s pregnancy losses and caused her to have a hysterectomy, she later opined that PCBs were only a possible cause. She admitted she did not know the circumstances of Mrs. Cunningham’s hysterectomy; nor did she know when Mrs. Cunningham’s miscarriages had occurred in relationship to the time that PCBs began to be used at Paoli. Thus, when defendants pointed to possible alternative causes, Dr. Sherman could provide no justification for her elimination of these alternative causes thus rendering her testimony inadmissible. Even if her testimony were admissible, Dr. Sherman’s statement that PCB’s were a possible cause did not have sufficient scientific certainty to survive summary judgment. See Schulz, 942 F.2d at 208 , supra p. 751.

Dr. Sherman’s entire testimony explaining her conclusion that PCBs caused Mrs. Cunningham’s sinusitis was as follows:

I said that I thought that the chronic exposure to PCBs, dibenzofurans, dioxins and the associated chemicals from the Pao-li yard interfered with her immune function, and secondarily, she had long time multiple complaints with respiratory upper respiratory problems. I think it’s a fact. I don’t think it’s the sole cause.

This testimony seems to place heavy reliance on the unreliable immunological data, see supra p. 756, and therefore we hold that the district court acted within its discretion in concluding that Dr. Sherman’s entire methodology with respect to Bessie Cunningham’s sinusitis was unreliable and thus inadmissible.

In sum, we conclude that while the district court properly excluded Dr. Sherman’s testimony with respect to Bessie Cunningham’s pregnancy losses, hysterectomy, and sinusitis, the court abused its discretion in excluding Dr. Sherman’s testimony that PCBs caused Cunningham’s headaches, arthritis/ar-thralgia and hypertension.

ii. Amber Burrell

.Dr. Sherman examined Amber Burrell, took a medical history of her from Amber’s mother, and reviewed Amber’s medical records. Dr. Sherman concluded that Amber’s skin reactions were prodromes of chloracne caused by exposure to PCBs, and that Amber’s recurring respiratory ailments and infections, immune system damage, and absence of appropriate ankle reflexes were also caused by exposure to PCBs.

The defendants’ experts pointed to a possible alternative cause of Amber’s respiratory ailments and infection, suggesting that these ailments might have been normal childhood problems. But Dr. Sherman testified that, although she did not know whether Amber had contracted more respiratory infections than most children, the unusual nature of Amber’s skin abnormalities and premature breast development combined with Amber’s elevated leukocyte counts shown in hospital records and current immunological *767 tests and with the fact that Amber’s infections largely went away when she moved away from Paoli, demonstrated that PCBs had caused Amber’s respiratory ailments and infections.

Hence, when asked about a possible alternative cause, Dr. Sherman pointed to a variety of data as justification for her conclusion that PCBs had caused Amber’s respiratory infections. Regardless of whether Dr. Sherman’s conclusion was correct, her methodology was reasonably reliable. Although part of Dr. Sherman’s differential diagnosis relied on current immunological tests that the district court excluded as unreliable, this was only a small part of the basis for her opinion. Thus, the district court abused its discretion in excluding Dr. Sherman’s testimony regarding Amber’s respiratory ailments.

Dr. Sherman also testified that PCBs caused Amber’s skin abnormalities which she asserted were prodromes of chloracne. She stated that the only reason Amber did not develop full blown chloracne was that she moved away from Paoli. But the defendants showed that Amber’s skin condition might have been normal teenage acne with no relationship to PCBs. Dr. Sherman offered no explanation as to why she eliminated this possibility — at least no explanation reliable enough to be helpful to the jury. Dr. Sherman admitted she was not an expert in dermatology, and she demonstrated little knowledge about chloracne. She incorrectly testified that “we don’t know whether ordinary teenage acne isn’t really chloracne.” See supra p. 753. Moreover, she acknowledged that she was not sure if she had ever seen acne on anyone who had not been exposed to PCBs, and opined that the only way to determine whether a patient’s acne was caused by PCBs was to do a history of their exposure to PCBs (presumably, if the patient had been exposed to PCBs then ipso facto his or her acne was caused by PCBs). This suggests that Dr. Sherman made no effort to employ differential diagnosis to eliminate possible alternative causes of Amber’s skin abnormalities; we therefore uphold the. district court’s determination that Dr. Sherman’s testimony on Amber’s skin condition was inadmissible. 34

Dr. Sherman also testified that PCBs were a significant factor in Amber’s reduced ankle reflexes. However, she was not willing to answer yes or no to the question whether she could say this with a reasonable degree of medical certainty. Thus, the district court properly granted summary judgment on this claim regardless of whether Dr. Sherman’s testimony was admissible. See supra p. 753.

In sum, the district court properly excluded Dr. Sherman’s testimony regarding Amber Burrell’s skin condition and granted summary judgment on her claim regarding ankle reflexes, but abused its discretion in excluding her testimony regarding Amber’s respiratory ailments.

iii. Priscilla Burrell

Dr. Sherman did not examine any plaintiffs other than Bessie Cunningham or Amber Burrell, nor did she take their medical histories. The district court properly excluded Dr. Sherman’s testimony with respect to all of these plaintiffs because she did not advance a reliable explanation as to why she ruled out alternative causes with respect to any of them. We will delineate Dr. Sherman’s failure to offer such an explanation beginning with her testimony regarding Priscilla Burrell.

Dr. Sherman did not examine Priscilla Burrell, nor did she take her medical history. She did talk to Priscilla Burrell and reviewed her medical records formulated by doctors who had examined her. Based on these medical records, Dr. Sherman testified that Priscilla Burrell suffered skin problems, two pregnancy losses, dizziness, ringing in her ears, and kidney damage, all as a result of her exposure to PCBs. Subsequent to Dr. Sherman’s deposition and entry of a district court order that she should not supplement her expert report, the thirty-three year old *768 Mrs. Burrell was diagnosed as having kidney cancer.

Dr. Sherman’s assessment of the cause of Priscilla Burrell’s skin problems suffers from the same methodological flaws as her opinion with respect to Amber’s skin problems. Similarly, in linking PCBs to Priscilla Burrell’s pregnancy losses, Dr. Sherman failed even to ascertain whether these losses had been due to optional abortions. With respect to dizziness, Dr. Sherman testified that infections cause dizziness and that she was unable to state whether they had done so here. Dr. Sherman took no steps that might have allowed her to conclude that PCBs rather than infections caused Mrs. Burrell’s dizziness. Thus, she did not offer any explanation as to why she had eliminated even the most obvious possible alternative causes of skin problems, pregnancy losses or dizziness.

Dr. Sherman also testified that PCBs caused kidney damage which resulted in the blood in Mrs. Burrell’s urine. But the only basis for Dr. Sherman’s belief that there even was blood in Mrs. Burrell’s urine was a conversation with Martin D’Urso, plaintiffs’ counsel. Dr. Sherman did not examine Mrs. Burrell’s medical records with respect to this problem and thus, did not have a reliable basis for concluding that Burrell had kidney problems much less that those problems were caused by PCBs. Dr. Sherman offered no explanation as to what alternative causes she considered for Mrs. Burrell’s kidney problems or how she ruled them out. Thus, Dr. Sherman’s testimony regarding Priscilla Burrell’s injuries was unreliable and hence inadmissible.

Plaintiffs made no effort to ask the district court to allow Dr. Sherman to supplement her opinion in light of the new diagnosis of kidney cancer, nor did they offer the opinion of Mrs. Burrell’s treating physician in opposition to summary judgment. Because plaintiffs presented no admissible evidence on Priscilla Burrell’s personal injury claims, the district court properly granted summary judgment for the defendants on these claims.

iv. Monica Hilton

Dr. Sherman did not examine Monica Hilton. She opined that PCBs contributed to Monica Hilton’s susceptibility to cough and respiratory problems and to her (related) augmented immune response. But Dr. Sherman failed to explain why she thought that PCBs were a cause of these problems in light of possible alternative causes. Thus, the district court properly excluded Dr. Sherman’s testimony with respect to Monica Hilton and granted summary judgment against her on her personal injury claims.

v. Matthew Cunningham

Dr. Sherman did not examine Matthew Cunningham but nonetheless testified that PCBs caused his heart disease including hypertension, aortic sclerosis, myocardial infarction and ultimately his death. She also averred that PCBs caused Cunningham’s fiver damage.

Dr. Sherman theorized that PCBs caused Mr. Cunningham’s heart disease either by elevating his cholesterol and triglyceride fractions or by directly affecting his tissues. While she later admitted that she was not sure whether these two mechanisms were distinct and that there were no tests showing an elevation of cholesterol levels, she pointed out that there was a test showing elevated triglyceride levels. But the district court properly excluded this test as unreliable. See supra pp. 755-56. Absent reliance on this data, Dr. Sherman expressed no reason for her conclusion that PCBs caused Cunningham’s heart disease nor why she had ruled out the possibility of alternative causes for this disease. Thus, the district court acted within its discretion in excluding her opinion.

While Dr. Sherman admitted that hereditary factors may have increased Mr. Cunningham’s risk of contracting hypertension, she did not withdraw her opinion that PCBs were a significant factor in his developing hypertension and other forms of heart disease. But once again she failed to explain why she thought that PCBs rather than alternative causes were responsible and thus the district court properly excluded her testimony.

Dr. Sherman also asserted that PCBs caused Mr. Cunningham’s cellular fiver dam *769 age, but there was no evidence indicating that that damage had any symptoms. Thus, the district court properly granted summary judgment on the claim of liver damage because the damage was asymptomatic, see infra pp. 785-86 & n. 51, and on the claims of hypertension and heart disease because Dr. Sherman’s causation opinion with respect to these diseases was inadmissible.

vi.Patricia Ingram

Dr. Sherman did not examine Patricia Ingram but did review her medical records. She testified that PCBs were a significant factor in Patricia Ingram’s chronic obstructive pulmonary disease and asthma, and her increased cellularity. She added that the steroids that Patricia Ingram took for her asthma caused her diabetes and therefore that PCBs indirectly caused that diabetes. But she could not testify with a reasonable degree of medical certainty as to the particular mechanism by which PCBs caused the asthma and ultimately the diabetes.

Dr. Sherman conceded that she had not ruled out possible alternative causes of Patricia Ingram’s asthma, although this may have just meant that she had not completely eliminated that these alternatives as causal agents. Yet this testimony, when considered in the light of Dr. Sherman’s limited expertise about asthma show that her methodology in evaluating the causation of Patricia Ingram’s asthma was unreliable. This renders her methodology in relating PCBs to diabetes unreliable as well because her methodology depended on the fact that PCBs had caused Ingram’s asthma. Moreover, Dr. Sherman acknowledged that she did not consider any possible causes of Patricia Ingram’s diabetes other than her asthma treatment. As for increased cellularity, there is no indication that there are any symptoms associated with this condition and thus it cannot be the basis of a personal injury claim. See infra pp. 785-86 & n. 51. Because Dr. Sherman’s testimony on the cause of Patricia Ingram’s illnesses is inadmissible and because increased cellularity is asymptomatic, Patricia Ingram’s personal injury claims cannot survive summary judgment.

vii. John Ingram, Sr.

Dr. Sherman did not examine John Ingram, Sr. but testified that PCBs caused his elevated triglycerides, augmented immune response and hypertension. The first two are asymptomatic and non-compensable. See infra at pp. 785-86 & n. 51. As for hypertension, Dr. Sherman indicated that because she was working solely from medical records, she had not examined possible alternative causes. Therefore, Dr. Sherman’s testimony on John Ingram, Sr.’s hypertension must be excluded as unreliable which means that John Ingram, Sr.’s personal injury claims cannot survive summary judgment.

viii. John Ingram, Jr.

Dr. Sherman reviewed John Ingram, Jr.’s medical records but did not examine him. She testified that PCBs caused a suppression of his immune system which significantly contributed to his recurring respiratory infections. However, the only evidence Dr. Sherman cited for the proposition that John Ingram, Jr.’s immune system was in fact suppressed was the AAL immunological test results which are unreliable and inadmissible. See supra p. 756. Thus, Dr. Sherman’s methodology for concluding that John Ingram, Jr.’s personal injuries were caused by PCBs is unreliable. As a result, her opinion is inadmissible and Ingram’s personal injury claims cannot survive summary judgment.

ix. George Burrell

Dr. Sherman stated that George Burrell’s headaches, dizziness, ringing in the ears, and eye irritations were compatible with the pattern of illnesses caused by PCBs but she did not state with a reasonable degree of medical certainty that PCBs caused Mr. Burrell’s medical problems. Without sufficient evidence of causation, these claims cannot survive summary judgment.

x. Wallace Cummins

Dr. Sherman did not examine Wallace Cummins, nor did she read his medical records before submitting her expert report. By the time of the in limine hearing, howev *770 er, Dr. Sherman had read Cummins’ medical records and she listed Cummins’ cancer as evidence for the proposition that Paoli residents had a higher incidence of kidney cancer than was present in the general population. However, she did not testify that PCBs caused Cummins’ cancer. Therefore, her testimony, even if it were admissible, is insufficient to enable Cummins to survive summary judgment.

Plaintiffs represent that Cummins’ treating physician, Dr. Stanley Glauser, will testify that PCBs caused Cummins’ kidney cancer and his hypertension. However, while Cummins listed Dr. Glauser as a fact witness for trial, he did not submit his affidavit in opposition to summary judgment. Indeed, the only reference to Dr. Glauser in the summary judgment record is an unsupported reference in plaintiffs’ opposition brief to the deposition of Cummins’ wife, who testified as to what another physician had told her that Dr. Glauser had told him about Cummins’ condition. Such hearsay testimony is insufficient to survive a motion for summary judgment. Fed.R.Civ.P. 56(e). See Fowle v. C & C Cola, Div. of ITT-Cont. Baking Co., 868 F.2d 59, 67 (3d Cir.1989); Garside v. Osco Drug, 895 F.2d 46, 50 (1st Cir.1990). 35 Thus, Cummins’ personal injury claims cannot survive summary judgment.

xi. "April Ingram

Dr. Sherman testified that April Ingram had skin problems, but she did not testify that PCBs caused these problems. She did allege that PCBs caused April’s immunological abnormalities but these are asymptomatic. See infra at pp: 785-86 & n. 51. Thus, April Ingram’s personal injury claims cannot survive summary judgment.

xii. The Remaining Plaintiffs

Dr. Sherman did not testify that PCBs caused illnesses of any of the other plaintiffs, including those seeking damages for personal injuries. For their personal injury claims, these plaintiffs relied on the testimony of the experts whose testimony the district court excluded for failure to comply with the court’s pre-trial orders. But because the plaintiffs did not point to the testimony of these experts in their opposition to summary judgment, the testimony of these experts does not enable these plaintiffs to survive summary judgment. See supra pp. 740-41.

c. Rule U08

The district court also excluded Dr. Sherman’s testimony under Rule 403, reasoning that the jury would give more probative weight to Dr. Sherman’s testimony than it was entitled to, because Dr. Sherman simply attributed all illnesses to PCBs and was evasive on cross examination. This is not an appropriate Rule 403 rationale. We fail to see what characteristics of Dr. Sherman’s testimony would make it the subject of the type of 403 balance necessary for its exclusion, such as that it is harder for the jury to evaluate this testimony than any other scientific testimony. See Downing, 753 F.2d at 1240. The 403 ruling will be reversed.

d. Summary (DiGregorio and Sherman)

In sum, we will uphold the district court’s exclusion of Dr. DiGregorio’s testimony regarding causation of the Hanes plaintiffs’ personal injuries as based on an unreliable methodology — he failed to examine the plaintiffs, review their medical records, or consider alternative causes of their illnesses. As a result we will affirm the district court’s grant of summary judgment against the Hanes plaintiffs on these claims.

Similarly, we will uphold the district court’s exclusion of Dr. Sherman’s testimony regarding causation of the Kohn/Klehr plain *771 tiffs’ present injuries with the exception of the claims of Bessie Cunningham and Amber Burrell. With respect to most plaintiffs, the district court acted within its discretion in concluding that Dr. Sherman’s methodology was unreliable. She failed to examine plaintiffs or take their medical histories and offered no good explanation as to why she concluded that PCBs were the cause of plaintiffs’ illnesses. While we will reverse the district court’s ruling that Dr. Sherman is not a qualified expert, we will uphold the court’s decision that the immunological tests on the plaintiffs were not of the type reasonably relied on by experts.

However, we have concluded that the district court abused its discretion in excluding Dr. Sherman’s testimony regarding some of the present illnesses of Bessie Cunningham and Amber Burrell — both of whom Dr. Sherman did examine and whose medical histories she took, and with respect to whom she performed an acceptable differential diagnosis. Their claims will survive summary judgment so long as the plaintiffs have presented sufficient admissible evidence of their exposure to PCBs and the harm caused by PCBs. We turn to those questions now.

VII. EXPOSURE — THE OPINION OF IAN C.T. NISBET, Ph.D.

The defendants submit that the plaintiffs’ personal injury and medical monitoring claims cannot survive summary judgment because they cannot demonstrate that they were exposed to PCBs at levels above the background levels to which members of the general population are exposed. 36 Six plaintiffs (Bessie Cunningham, Matthew Cunningham, Amber Burrell, Priscilla Burrell, and Patricia Ingram), including the two who have potentially viable personal injury claims, relied on Dr. Ian C.T. Nisbet to demonstrate the “extent of their exposure to PCBs.” 37 Dr. Nisbet has a Ph.D. in physics with a specialization in fluid mechanics. Although he has no formal education in toxicology, he has worked exclusively in the field of environmental sciences since 1968, and he has developed and worked on methods for assessing the way chemicals move in the environment and the risks they pose. In 1975, the EPA commissioned Dr. Nisbet to summarize and analyze studies on PCBs. The Electrical Power Research Institute has also commissioned Dr. Nisbet’s organization to review PCB literature; indeed, two of the defendants’ experts, Drs. Whysner and Chase, were subcontractors on the study. In 1991, the Agency for Toxic Substances and Disease Registry selected Dr. Nisbet as one of three peer reviewers for a toxicological profile on PCBs. Dr. Nisbet has also been involved in six or seven litigated cases involving PCBs — half for plaintiffs, half for defendants. Thus, Dr. Nisbet is a well qualified expert.

Dr. Nisbet began his assessment of plaintiffs’ exposure to PCBs by evaluating the history of PCBs at Paoli. He reviewed the plaintiffs’ medical records and employment records and the use of PCBs at the Yard, including the increasing concern at the Yard about the potential harm of PCBs and the safety measures taken as a result. He looked at Monsanto’s escalating warnings to its customers about PCB risks, increasing federal actions on PCBs, improved housekeeping measures by defendant railroads in the early 1980s, emergency measures taken by EPA in the mid 1980s to reduce runoff from the Paoli Railyard, and statements from *772 plaintiffs about steps they took to decrease exposure to PCBs after being warned in 1986. From this information, he concluded that plaintiffs’ exposure to PCBs was greatest before 1986.

Dr. Nisbet combined this general assessment of plaintiffs’ history of exposure to PCBs with three methods of calculating actual PCB levels: (1) recalculation of blood tests taken by the American Medical Laboratories (“AML”), (2) recalculation of background levels of PCBs, and (3) back calculations of PCB levels in 1986 based on the decay rate of PCBs. The district court held all three methods to be unreliable. We agree with the district court with respect to the first and third methods described below, but conclude that the district court erred in finding Dr. Nisbet’s second method unreliable and that it abused its discretion in excluding this opinion.

A. Recalculation of the AML Data

Dr. Nisbet first relied on a recalculation of results of blood tests for PCBs taken by the American Medical Laboratories, Inc. and National Medical Services (“NMS”) in 1986 and in 1987. These tests originally showed that plaintiffs’ PCB blood levels fell within the general population background PCB levels reported by the laboratory. However, Dr. Nisbet concluded that these tests were inaccurate because the laboratory manager of AML advised him that lab practices in 1986 and 1987 had become sloppy. This was in turn because: (1) AML had used the packed column method, an ostensibly obsolete method of PCB assessment; (2) AML no longer had back up data and had not analyzed samples until a month after they were collected; (3) AML had analyzed blood serum instead of plasma which understates PCB concentrations; and (4) AML’s chromatogram showed a PCB pattern that was not consistent with any of the PCB transformer fluids (Aroelors) in the Yard.

Because AML results were two and a half to three times lower than Michigan State Laboratory results for three unrelated samples in March of 1986 and because Dr. Nis-bet thought Michigan State’s laboratory was more reliable than AML, he concluded that AML results were generally too low by a factor of two and a half to three. (N 158, 164-68). He therefore recalculated AML tests for plaintiffs from August of 1987 by multiplying them by two and a half to three. After such recalculation, the level of PCBs ostensibly present in plaintiffs’ blood became far more than background. Defendants challenge this methodology.

The district court found that many of the factual bases from which Dr. Nisbet determined that AML results were inaccurate were themselves inaccurate. It found that the packed column method AML used to assess PCBs was better known than the alternative, that AML did have back up data that Dr. Nisbet could have attained and evaluated, and that AML analyzed the samples a mere two weeks after they were collected. Moreover, tests of AML procedures demonstrated that AML results were generally accurate. The court referenced the testimony of Dr. Whysner who explained that AML had produced results the EPA had found acceptable when EPA spiked samples with PCBs and sent them to AML for analysis in December of 1986; and that AML had produced accurate results for 30 samples sent to it by Washington Occupational Health Associates, Inc. from 1986-92. And, data provided to Dr. Nisbet at the in limine hearing (data obtained by Dr. Whysner from AML and Michigan State) demonstrated that the disagreement between AML results and Michigan State results was smaller than he had previously thought. 38

The district court went on to find that even assuming a large gap between AML and Michigan State results for the three samples and that the Michigan State results were more accurate, it was not scientifically valid *773 to recalculate results for samples from one laboratory based on a comparison of two laboratories’ analysis of three samples a year and a half earlier. Dr. Whysner stated that he had never seen an expert perform such a recalculation. In his view, the problems with this methodology were especially serious given the existence of data showing AML results to be generally accurate, and the existence of data showing that AML and Michigan State had produced similar results for PCB samples sent to each laboratory in both October of 1987 and in 1989. Dr. Whysner also stated that there are a variety of explanations for the difference between AML and Michigan State results for the three samples sent in 1986, including the possibility that something happened to the samples during transit.

In light of this evidence the district court concluded that the recalculation rested on a novel scientific methodology that had not been identified in any literature and so had not been subject to critical scrutiny, that it was based on inaccurate assumptions, and ignored inter-laboratory comparison data. The court found that no reasonable scientist in the field of exposure assessment would perform such a calculation. As a result, the district court held Dr. Nisbet’s testimony on this matter to be unreliable under Rule 702, and the reliance on the 1986 data from the three samples and the recalibrated 1987 AML results to be data not of the type reasonably relied upon by experts under Rule 703.

We defer to the district court. Recalculation might be acceptable methodologically if there were a series of disagreements of two and a half to three times between two labs over a period of time. However, recalculation based on three samples does not appear rehable. Applying the Dcmbert factors, recalculation itself is not a technique identified in the scientific literature; nor is it generally accepted. To the extent it produces falsifiable results enabling scientists to test whether AML consistently understates its results by two and a half to three times, those results have in fact been falsified by data showing that AML and Michigan State generally produce similar results. And recalculation based on a comparison of three samples is likely often to lead to erroneous results. Although plaintiffs argue that because Dr. Nis-bet himself admits that his recalculations are rough estimates he will not confuse the jury, we hold that under Dcmbert, the district court did not err in concluding that the recalculations Dr. Nisbet performed are too rough to be considered reliable at all. Thus, we uphold the district court’s exclusion of these recalculations under Rules 702 and 703.

B. Recalculation of Background

Dr. Nisbet also believes that, because the background range used by AML was too high, AML results were above background even if the AML results were accurate. Dr. Nisbet opined that the 5 to 30 parts per billion background range used by AML was too high because it relied on studies of unrepresentative populations and because it assumed that the background range had not changed from 1950 to 1990. In reality, according to Dr. Nisbet, the PCB background range had declined since 1970 as a result of regulatory efforts and voluntary efforts to reduce PCB use. Apparently, the few longitudinal studies that have been performed agree that this decline has occurred. Dr. Sherman agreed with Dr. Nisbet that background levels of PCBs have dropped over time.

Rather than relying on the background levels used by AML, Dr. Nisbet decided that background use could best be measured by the National Human Adipose Tissue Survey (NHATS), a nation-wide study which appropriately showed declining PCBs in human fat samples from 1970 to 1983. (N 98-99, 142). However, Dr. Nisbet could not directly compare the NHATS data on the background levels of PCBs in fat to the AML data on levels of PCBs in plaintiffs’ blood. Before attempting such a comparison, Dr. Nisbet had to convert background PCB levels in fat into their equivalent in blood by using a ratio of PCBs in blood to PCBs in fat.

However, the district court found that “experts in the field of exposure assessment do not reasonably rely upon NHATS fat data to calculate background PCB blood levels,” and excluded the calculation under Rule 703. *774 The court based its holding on the testimony of Dr. Whysner, who stated that he had never seen experts use NHATS fat data to calculate background levels in blood. Dr. Whysner testified that NHATS reported the background levels in fat in ranges because EPA did not think the data was accurate enough to give average levels; and if it was not accurate enough to give average levels it certainly was not accurate enough to use to convert into blood levels.

To the extent that Dr. Whysner’s testimony was that PCB levels in fat could not be used to determine PCB levels in blood, plaintiffs adequately rebutted it. Plaintiffs cite several articles that convert PCB levels in fat to levels in blood (see, e.g., Luotamo, et al. Assessment of Exposure to PCBs: Analysis of Selected Isomers in Blood and Adipose Tissue, 54 Env. Research 121-34 (1991)) and represent that one of defendants’ experts, Dr. Chase, made such a conversion in a 1984 trial at a rate which would lead one to conclude that plaintiffs’ blood levels were well above background. It thus appears that the plaintiffs are correct that fat levels are a type of data often used by experts to calculate blood levels, and that calculating blood levels from fat levels is a rehable methodology.

Dr. Whysner’s testimony, however, also called into question whether the NHATS fat data is a type of data reasonably relied on by experts to perform such a calculation. 39 Defendants state that while Dr. Chase said that the fat to blood ratio of PCBs could be between 50 to 1 and 500 to 1, he did not say that any estimate was reliable enough to use, and he explicitly stated, like Dr. Whysner, that the NHATS data was not accurate enough to use as a starting point.

Because NHATS data is such a specific type of data, there is little evidence in the literature on whether experts reasonably rely on the particular data. Yet NHATS data is a subset of a type of data reasonably relied on by experts; and thus we are inclined to deem it reliable. See supra pp. 742 & n. 8, 745-46 n. 14. Dr. Nisbet, an extremely well qualified expert, thought that NHATS data was a type of data reasonably relied upon, and he provided strong reasons why this data was at least as accurate a source of PCB background blood levels as other studies that analyzed PCB background blood levels directly. Indeed, defendants cite only one or perhaps two (depending on the meaning of Dr. Chase’s deposition) experts who have contrary opinions on the reliability of NHATS fat data as a source of conversion to blood levels, and they provide no strong reasons why these experts should be believed as opposed to Dr. Nisbet. Although we review the district court’s Rule 703 determination for abuse of discretion, after taking a hard look, the court’s conclusion that the (converted) fat data is not reasonably relied upon does not survive the foregoing analysis. Aided by the preference for admissibility of such evidence under the rules, the NHATS data is rehable enough to prove helpful to the jury when converted into background blood levels, and it passes Daubert muster. The district court therefore abused its discretion in excluding it.

C. Back Calculations

Dr. Nisbet used a third method to, calculate PCB exposure prior to 1986. He began with tests of the PCB levels of six plaintiffs in 1992 by the ELI Eco Logic Laboratories, Inc. (Eco Logic), a Canadian scientific consulting firm that performs original research. These tests showed levels of PCBs in plaintiffs’ blood comparable to those provided by AML tests in 1986. However, because of precautions taken at the Yard since 1986 to decrease leakage of PCBs, Dr. Nisbet reasoned that plaintiffs’ exposure to PCBs had declined since 1986. Thus, he assumed that plaintiffs must have had more PCBs in their blood in 1986 than they had in 1992. By *775 using assumptions about the decay rate of PCBs combined with a model of plaintiffs’ exposure to PCBs from non-yard sources, Dr. Nisbet calculated a range of plaintiffs’ blood level of PCBs in 1986 from the level in their blood in 1992.

The district court excluded Dr. Nisbet’s back calculations as an unreliable methodology for a variety of complicated reasons. One of its reasons was that it concluded under Rule 703, that the Eco Logic data, ostensibly showing the levels of PCBs in plaintiffs’ blood in 1992 was a type of data not reasonably relied upon by experts. Because we think that the district court acted within its

This text is long and has been trimmed here. Open the source document for the complete record.