The opinion
SMITH, Senior Circuit Judge,
dissenting:
I respectfully dissent because the majority improperly applies Daubert v. Merrell Dow Pharmaceuticals, 509 U.S. 579 , 113 S.Ct. 2786 , 125 L.Ed.2d 469 (1993), and does not adequately clarify the roles of the expert, the trial court and the appellate court. The following analysis is based on a few basic ideas. As a “gatekeeper,” the trial court must sift through expert testimony to decide not only whether an expert may testify, but what portion of the expert’s testimony is admissible. A single expert may offer several opinions to reach his ultimate conclusion, and each opinion must be admissible under Daubert . Further, an expert’s testimony does not “assist” the trier of fact if the expert does not explain the steps he took to reach his conclusion. We should not require the trier of fact to accept blindly the expert’s word to fill the analytical gap between proffered “scientific knowledge” and the expert’s conclusions. Therefore, the trial court “gatekeeper” has broad discretion to decide whether a leap of faith across the analytical gap is so great that, without further credible grounds, the testimony is inadmissible.
I. Standard of Review
The majority states that, although we review the trial court’s admissibility rulings for abuse of discretion, “we apply a particularly stringent standard of review to the trial judge’s exclusion of expert testimony” and “our review is plenary” over the trial court’s interpretation of evidence rules. Because understanding the scope of appellate review helps define the role of the trial court, I believe we should follow other circuits and present a more precise explanation of the standard of review. See, e.g., Cook v. American Steamship Co., 53 F.3d 733, 738 (6th Cir.1995) (Three standards in reviewing admissibility of expert opinion: (1) trial court’s factfinding is reviewed for clear error; (2) trial court’s ruling whether opinion is scientific knowledge is question of law requiring plenary review; and (3) trial court’s ruling whether opinion assists the trier of fact is reviewed for abuse of discretion); Bradley v. Brown, 42 F.3d 434, 436-37 (7th Cir.1995) (Plenary review of whether trial court applied Daubert ft’amework, but trial court’s findings not disturbed unless manifestly erroneous.).
In applying a “particularly stringent” review, we do not change the threshold of review, but conduct a searching review of the record (i.e., take a “hard look”) while maintaining the proper standard of review. See, In re Paoli R.R. Yard PCB Litigation, 35 F.3d 717, 749-50 (3d Cir.1994) (give a “ ‘hard look’ (more stringent review)” to decide whether the trial court abused its discretion), cer t. denied, — U.S. -, 115 S.Ct. 1253 , 131 L.Ed.2d 134 (1995). This court already suggested such a “hard look” where it remanded a case in light of Daubert and instructed the trial court to make specific factfindings to facilitate appellate review. United States v. Lee, 25 F.3d 997, 998 (11th Cir.1994). Under this “hard look,” I offer for clear guidance review terminology that is firmly established in the jurisprudence of this and other circuits. Whether the trial court properly applied Rule 702 by following
*536 the framework set forth in Daubert is a question of law over which this court exercises complete and independent review. See, Peterson v. Atlanta Housing Authority, 998 F.2d 904, 912 (11th Cir.1993) (“The district court’s conclusion of law is subject to complete and independent review by this court.”) (quoting, In re Sure-Snap Corp., 983 F.2d 1015 , 1017 (11th Cir.1993)); Bradley, 42 F.3d at 436-37 . I suggest the term “complete and independent” as being more precise and accurate than the ubiquitous “de novo” where the review is in fact the first one ever conducted. “De novo” carries a connotation of repetition, as in a “trial de novo” after a matter has previously been tried. To suggest that an appellate court is conducting a “new” review of the trial court’s conclusions of law is less than accurate when in fact those conclusions have never before been reviewed. The trial court’s preliminary fact-finding during a Rule 104(a) hearing to determine the admissibility of expert opinion is reviewed for clear error. See, Elston v. Talladega County Bd. of Ed., 997 F.2d 1394 , 1405 (11th Cir.1993) (“We review the district court’s findings of fact for clear error. A finding is clearly erroneous when although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed.”) (quoting Anderson v. Bessemer City, 470 U.S. 564, 573 , 105 S.Ct. 1504, 1511 , 84 L.Ed.2d 518 (1985)) (internal quotations omitted); Cook, 53 F.3d at 738 . In applying the Daubert framework, the trial court’s ruling on whether the expert opinion is (1) reliable (i.e., scientific knowledge grounded in the methods and procedures of science) and (2) relevant (i.e., “fits” the facts of the case) is reviewed for abuse of discretion. 1 See, Hibiscus Associates Ltd. v. Board of Trustees, 50 F.3d 908 , 917 (11th Cir.1995) (“A judge has broad discretion to exclude expert testimony, and his action will be upheld unless it is manifestly erroneous.”).
II. Admissibility of Expert Testimony
After presenting a thorough review of the Daubert standard, the majority errs by first applying the reliability prong of Daubert to the experts’ opinions as a whole, and then applying the relevancy prong. This approach treats all the experts as offering only one opinion leading to the ultimate conclusion that transformer dielectric fluids promoted Mr. Joiner’s small cell lung cancer. However, each expert is actually offering several *537 opinions leading to that ultimate conclusion. For example, the experts offer opinions that (1) furans and dioxins were present and (2) furans and dioxins promoted Mr. Joiner’s cancer. Each of these assertions is a separate opinion which must meet the Daubert standard, regardless of whether the assertions are given by the same or different experts. As the Paoli court stated,
[T]he 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 ... [A]ny step that renders the analysis unreliable under the Daubert factors renders the expert’s testimony inadmissible.
Paoli, 35 F.3d at 743, 745 (emphasis omitted).
The majority admonishes the trial court for not “viewing the bases of an expert’s opinion as a whole.” However, sifting through the expert’s testimony is a crucial “gatekeeping” function that not only requires the trial court to decide which experts may testify, but also requires the trial court to decide what the experts may testify about (i.e., the trial court must separate the wheat from the chaff). Litigants may not offer all of an expert’s testimony so long as they can search and find some portion that is admissible. Similarly, an expert may not bombard the court with innumerable studies and then, with blue smoke and sleight of hand, leap to the conclusion. Instead, the expert must explain how the opinion drawn from each study is acceptable under Daubert (i.e., how the study is methodologically grounded and “fits” the facts of the case), else the expert cannot testify about that particular study.
A Exposure to PCBs, Furans and Dioxins
Although finding there is a genuine dispute whether Mr. Joiner was exposed to PCBs, the trial court found insufficient evidence that Mr. Joiner was exposed to furans or dioxins. The trial court dismissed Mr. Joiner’s assertion that furans were created from PCBs in fire conditions because, although there was evidence of fire and other “hot” conditions, Mr. Joiner failed to show that conditions reached the requisite temperatures in this case (i.e., “fit”). Joiner v. General Electric Co., 864 F.Supp. 1310, 1317-18 (N.D.Ga.1994).
The majority concludes the trial court committed reversible error by overlooking a minor passage from Dr. Teitelbaum’s affidavit that provides specific evidence of “fit”: (1) the transformer’s were smoking which requires temperatures of 700 to 800 degrees centigrade and (2) some transformers were struck by lightning which inevitably produces furans. The majority further suggests the trial court’s ruling was erroneous because the defendants presented no evidence that the fires did not reach the requisite temperature. However, I disagree and I am not prepared to reverse the trial court on this issue because it is Mr. Joiner who has the burden of proving admissibility. Daubert at -, 113 S.Ct. at 2796 n. 10 (citing Bourjaily v. United States, 483 U.S. 171, 175-76 , 107 S.Ct. 2775, 2778-79 , 97 L.Ed.2d 144 (1987)); see also, Deimer, 58 F.3d at 345 (The expert “had the responsibility to apply his analysis to the facts of this case.”); American & Foreign Insurance Co., 45 F.3d at 139 (“[T]he burden is on the [party seeking to admit expert testimony] to persuade this court that the testing was reliable and supported by raw data.”). In making its ruling, the trial court sifted through such overwhelming evidence that it inevitably overlooked the passage from Dr. Teitelbaum’s affidavit. More importantly, Mr. Joiner himself failed to disclose this passage notwithstanding his burden of proving admissibility or his knowing the case hinged on such evidence. Mr. Joiner failed to cite this or any similar passage on appeal. Indeed, this passage would have been forever lost had it not been for the diligent, searching eye of the majority. I am not prepared to place such a burden on either the trial or appellate courts. Similarly, I am not prepared to encourage litigants to inundate the courts with raw data and force the courts to process the data to determine why certain evidence is admissible. The litigants and their experts should know their evidence better than anyone— they should be their own advocates for its admission.
*538 I would also affirm the trial court on the issue of exposure to dioxins. The trial court properly discarded treatise excerpts as inadmissible hearsay because they were not offered through expert testimony. The trial court did not abuse its discretion in discarding testimony that dioxins can be formed from Pyranol because there was no evidence that Pyranol was or may have been present in this case (i.e., “fit”). Nor did the trial court abuse its discretion in excluding testimony that burning PCBs produces dioxins where the testimony did not reference any supporting studies (i.e., grounded in science). Finally, the trial court did not abuse its discretion in finding that expert testimony concerning a specific incident “has little probative value given the evidentiary deficits in this ease.” Joiner at 1319 .
B. Causation — Promotion of Cancer
The trial court gave two alternative grounds for granting summary judgment on the issue of causation (i.e., whether Mr. Joiner’s exposure to dielectric fluid promoted his cancer): (1) the experts’ testimony did not “fit” because they assumed Mr. Joiner was exposed to furans and dioxins and (2) the experts did not show how the studies they relied on “fit” this case. Regarding the former ground, I am not prepared to reverse the trial court due to Mr. Joiner’s failing to disclose the critical passage regarding the temperature of the transformers which would have provided the “fit” required to admit evidence about furan and dioxin exposure. Moreover, I would affirm the trial court on the latter ground because it did not abuse its discretion in finding the experts failed to show how the proffered studies “fit” this ease.
1. Mice Studies. — The trial court found the experts’ reliance on mice studies was questionable because (1) there were only two studies; (2) the studies used massive doses; and (3) the studies yielded only preliminary results. Joiner at 1323 . The trial court excluded the studies because Mr. Joiner did not respond to these concerns, but merely “proceed[ed] as if the only issue is whether animal studies can ever be [proper].” Joiner at 1324 (emphasis added). The majority opinion apparently adopts Mr. Joiner’s argument, stating that “it is improper to find research unreliable solely because it uses animal subjects.” However, this ignores the trial court’s concern that the experts have not demonstrated how these mice studies “fit” this particular case.
In discussing “fit,” the Supreme Court stated,
The study of the phases of the moon ... may provide valid scientific “knowledge” about whether a certain night was dark, and if darkness is a fact in issue, the knowledge will assist the trier of fact. However (absent creditable grounds supporting such a link), evidence that the moon was full on a certain night will not assist the trier of fact in determining whether an individual was unusually likely to have behaved irrationally on that night.
Daubert at — —, 113 S.Ct. at 2796 (emphasis added). In explaining the concept of “fit,” the Paoli court stated,
[Expert] testimony will be excluded if it is not scientific knowledge for the purposes of this case.... [I]n 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.
Paoli, 35 F.3d at 743 (emphasis in original).
The trial court’s ruling was not that animal studies are inadmissible per se, but that Mr. Joiner’s general response that experts generally rely on animal studies fails to show the reliability and “fit” of these particular animal studies. Joiner at 1324 n. 25. The trial court’s concern is that the proffered studies (1) were on mice, not humans; (2) were of substantially higher doses of PCBs than Mr. Joiner’s exposure; (3) resulted in a different form of cancer than Mr. Joiner’s; (4) yielded only preliminary results and (5) were not accompanied by other studies (there were only two studies). Because Mr. Joiner failed to address the latter two concerns, the trial court found the studies were unreliable. Regarding the other concerns about “fit”, the trial court found that Mr. Joiner did not present “creditable grounds for supporting” *539 the link between these mice studies and Mr. Joiner’s cancer. 2
It is incumbent on the proponent of scientific evidence to fill the analytical gap between a proffered study and the particular facts of the ease (i.e., “fit”). Daubert at-n. 10, 113 S.Ct. at 2796 n. 10 (citing Bourjaily v. United States, 483 U.S. 171, 175-76 , 107 S.Ct. 2775, 2778-79 , 97 L.Ed.2d 144 (1987)); see also, Deimer, 58 F.3d at 345 (The expert “had the responsibility to apply his analysis to the facts of this case.”); American & Foreign Insurance Co., 45 F.3d at 139 (“[T]he burden is on the [party seeking to admit expert testimony] to persuade this court that the testing was reliable and supported by raw data.”). The trial court exercises its discretion to determine whether such a showing has been made, weighing several factors including the “liberal thrust” toward admitting expert evidence, the adversarial system’s ability to scrutinize admitted evidence, and the powerful influence of expert opinion. 3 Daubert at -, -, 113 S.Ct. at 2794, 2798 . Where no other scientific evidence is offered to fill the analytical gap, the trier of fact is required to take the expert simply on his word, placing blind faith in his expertise. However, if the trial court finds the expert testimony requires too great a leap of faith across the analytical gap, it may properly request good grounds to bridge the gap before admitting the testimony. See, Turpin v. Merrell Dow Pharmaceuticals, Inc., 959 F.2d 1349, 1360-61 (6th Cir.1992) (Regarding animal studies used to show the cause of birth defects, the court found “[t]he analytical gap between the evidence presented and the inferences to be drawn on the ultimate issue ... is too wide. Under such circumstances, a jury should not be asked to speculate on the issue of causation.”), cert. denied, 506 U.S. 826 , 113 S.Ct. 84 , 121 L.Ed.2d 47 (1992). This is not too onerous a request because the expert should certainly have reasons for drawing his conclusions from the study, else his testimony is inadmissible as the “subjective belief or unsupported speculation” that Daubert requires the trial court “gatekeeper” to screen out. 4 Daubert at -, 113 S.Ct. at 2795 . Therefore, the trial court did not abuse its discretion in ruling the mice studies testimony inadmissible where Mr. Joiner completely failed to respond to the trial court’s concerns.
2. Epidemiological Studies. — The trial court disregarded the experts’ reliance on epidemiological studies because “in every case ... the studies are either equivocal or not helpful” and “simply do not support the experts’ position that PCBs more probably than not promoted Joiner’s lung cancer.” Joiner at 1324, 1326 (emphasis in original). The majority reverses the trial court on this issue, alleging the trial court improperly decided whether the experts’ conclusions were correct instead of limiting its analysis to whether the studies were reliable. I respectfully disagree; the trial court’s concern is with “fit,” not whether the experts are correct.
*540 The Paoli court recognized that the distinction between focusing on an expert’s methodology instead of his conclusion “has only limited practical import.” Paoli at 746. The court explained,
When a judge disagrees with the conclusions of an expert, it will generally be because he or she thinks there is a mistake at some step in the investigative or reasoning process of that expert.... [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 .
Paoli at 746. By directing attention away from the trial court’s choice of terminology and toward its actual analysis, I conclude that the trial court did not abuse its discretion in ruling each study inadmissible.
The trial court found the Bertazzi capacitor manufacturers study inadmissible because its results showed “no grounds” for linking exposure to lung cancer, and the specific excerpts relied on by the experts merely show the “plausibility,” not probability, that exposure could cause cancer. Joiner at 1324 n. 26. These concerns alone are not dispositive because an expert may analyze a study and draw different conclusions than the study. However, an expert should have reasons for differing with the study or for finding that the study supports his conclusion notwithstanding language in the study to the contrary. Because Mr. Joiner failed to respond and provide supporting grounds, the trial court did not abuse its discretion in ruling this evidence inadmissible.
The trial court ruled the Zaek & Musch Monsanto study inadmissable where the study itself stated that the results were not “statistically significant.” Joiner at 1325 . The trial court ruled the Norwegian cable manufacturers study inadmissible because it “never mentions PCBs,” involves mineral oil exposure, and the study itself concludes that “[fjurther follow up ... studies ... are needed before any firm conclusions may be drawn.” Joiner at 1325 . The trial court also ruled the Yusho accidental toxic exposure study inadmissible because the study was a “preliminary report,” the study involves persons exposed to furans and dioxins, and Mr. Joiner’s own expert testified that the study “is not very convincing as the Japanese lifestyle is different ... [it is] suggestive but not convincing.” Joiner at 1326 (quoting Deposition of Dr. Teitelbaum). As with the Bertazzi study, the trial court did not abuse its discretion where Mr. Joiner failed to respond to the trial court’s concerns and provide further grounds for relying on these studies.
III. Conclusion
The trial court properly applied Daubert and did not abuse its discretion in ruling certain expert testimony inadmissible. Based on these rulings, there is insufficient evidence on the issue of causation. Therefore, I would affirm the trial court’s granting summary judgment in favor of defendants. Moreover, I caution against using the majority’s approach that applies each DaubeH prong to the testimony as a whole. I would approve the trial court’s step-by-step approach which properly anticipates a single expert as offering more than one opinion to support his ultimate conclusion.
. Those circuits addressing Daubert have shown similar deference to the trial court's admissibility determinations. See, e.g., Pedraza v. Jones, 71 F.3d 194, 197 (5th Cir.1995) (trial court's ruling drug addict's expert testimony inadmissible is reviewed for abuse of discretion); Gier v. Educational Service Unit No. 16, 66 F.3d 940 , 942 (8th Cir.1995) (trial court’s ruling psychologist testimony inadmissible reviewed for "clear abuse of discretion”); Deimer v. Cincinnati Sub-Zero Products, Inc., 58 F.3d 341, 344 (7th Cir.1995) ("[W]e apply a deferential standard of review ... A decision to allow expert testimony is within the broad discretion of the trial judge and is to be sustained ... unless manifestly erroneous.”) (internal quotations omitted); Cook v. American Steamship Co., 53 F.3d 733, 738 (6th Cir.1995) ("[Wjhether the proffered expert opinion 'will assist the trier of fact to understand the evidence or to determine a fact in issue,' is a relevancy determination and therefore one we review for abuse of discretion.”); United States v. Dorsey, 45 F.3d 809, 814 (4th Cir.1995) ("[Ejven under the Daubert analysis, a trial judge has a great deal of discretion in deciding whether to admit or exclude expert testimony.”) (citing United States v. Bynum, 3 F.3d 769, 773 (4th Cir.1993) ("The [Daubert ] Court emphasized that it was prescribing a 'flexible' rule, one committed, as are most questions of admissibility of evidence, to the discretion of the district courts.”), cert. denied, - U.S. -, 114 S.Ct. 1105 , 127 L.Ed.2d 416 (1994)), cert. denied, - U.S. -, 115 S.Ct. 2631 , 132 L.Ed.2d 871 (1995); American & Foreign Insurance Co. v. General Electric Co., 45 F.3d 135, 137 (6th Cir.1995) ("A trial court has broad discretion in the matter of the admission or exclusion of expert evidence, and ... is to be sustained unless manifestly erroneous.”) (internal quotations omitted); Bradley v. Brown, 42 F.3d 434, 436-37 (7th Cir.1995) ("We first undertake a de novo review of whether the district court followed the framework set forth in Daubert [, and if so,] we will not disturb the district court’s findings unless they are manifestly erroneous.”); In re Paoli, 35 F.3d 717, 749-50 (3d Cir.1994) (a "hard look” at trial court's exercising its discretion); United States v. Rincon, 28 F.3d 921, 923 (9th Cir.1994) (admissibility of expert opinion on eyewitness identification reviewed for abuse of discretion), cert. denied, - U.S. -, 115 S.Ct. 605 , 130 L.Ed.2d 516 (1994); and United States v. Muldrow, 19 F.3d 1332, 1337 , (10th Cir.1994) ("We review a trial court's admission of evidence under an abuse of discretion standard.”), cert. denied, - U.S. -, 115 S.Ct. 175 , 130 L.Ed.2d 110 (1994).
. Had this law suit involved mice exposed to high doses of PCBs who developed some type of lung cancer, the “fit" would have been self-evident. However, the relationship between the studies and the facts of this case is much more tenuous.
. In this regard, the Daubert Court stated,
Vigorous cross-examination, presentation of contrary evidence, and careful instruction on the burden of proof are the traditional and appropriate means of attacking shaky but admissible evidence____ Expert evidence can be both powerful and quite misleading because of the difficulty in evaluating it. Because of this risk, the judge in weighing possible prejudice against probative force under Rule 403 ... exercises more control over experts than over lay witnesses.
Daubert at-, 113 S.Ct. at 2798 .
.Common law precluded an expert from testifying at all about an ultimate fact in issue, relegating his role to guiding the trier of fact up to the ultimate fact without taking the final step. Although an expert may now testify to an ultimate fact, this permissiveness certainly does not permit an expert to testify solely to an ultimate fact without guiding the trier of fact to that conclusion. For example, an expert could not give a one sentence testimony, "Mr. Joiner’s lung cancer was promoted by his exposure to dielectric fluid, you can take my word for it.” Nor would he save his testimony by adding, “I’ve heard of studies that show saccharine causes cancer in laboratory animals.” In order to "assist” the trier of fact, the expert must further explain his reasoning by testifying about what studies he relies on to form his opinion, how rehable are the studies, and how the studies relate to this particular case.