concurring in part and dissenting in part.
I join part IID of the opinion of the court. I also agree with the court’s disposition of the plaintiffs evidentiary argument. See Maj. Op. at 1075-77. I dissent, however, from parts IIA — C of the court’s opinion.
I.
My primary disagreement with the majority concerns the test for determining whether a defense motion for summary judgment or judgment as a matter of law should be granted in an employment discrimination case governed by the procedural scheme sanctioned in McDonnell Douglas Corp. v. Green, 411 U.S. 792, 93 S.Ct. 1817, 36 L.Ed.2d 668 (1973). Like several other courts of appeals {see Maj. Op. at 1066-67), the majority here holds that when the plaintiff has made out a prima facie case and has offered enough evidence to support a finding that the explanation was pretextual, a defense motion for summary judgment or judgment as a matter of law must always be denied. However, there is strong contrary authority, which the majority does not acknowledge. The in banc Fifth Circuit, by a vote of 16 to 1, has rejected the majority’s position. Rhodes v. Guiberson Oil Tools, 75 F.3d 989, 993 (5th Cir.1996) (in banc). So have the First and Eleventh Circuits. Isenbergh v. Knight-Ridder Newspaper Sales, Inc., 97 F.3d 436, 442-43 (11th Cir.1996); Woods v. Friction Materials, Inc., 30 F.3d 255, 260-61 n. 3 (1st Cir.1994); LeBlanc v. Great American Ins. Co., 6 F.3d 836, 842-43 (1st Cir.1993), cert. denied, — U.S. -, 114 S.Ct. 1398, 128 L.Ed.2d 72 (1994). I believe that these other courts have analyzed the question more accurately than the majority has here.
If the majority had merely said that, under the circumstances described above, a defense motion for summary judgment or judgment as a matter of law must generally be denied, I would agree. When a plaintiff makes out a prima facie case and there is sufficient evidence in the record to permit a rational trier of fact to find that the employer’s explanation is untrue, a defense motion for summary judgment or judgment as a matter of law should usually be denied. But not always, as the majority contends.
In my view, the correct test is the following; a defense motion for summary judgment or judgment as a matter of law should be granted when the evidence in the record could not persuade a rational trier of fact that intentional discrimination on the ground alleged by the plaintiff was a determinative cause of the challenged employment action. This does not mean that a plaintiff, in order to reach the trier of fact, must always prove “pretext plus,” i.e., that the plaintiff must always produce some evidence in addition to what is necessary to establish a prima facie ease and to show that the employer’s explanation is pretextual.1 On the contrary, in most cases, such additional proof is not needed. But I disagree with the majority that *1079proof of the elements of the prima facie ease and proof of pretext are always enough.
II.
A. The key to the question at issue lies in the nature of the “presumption” of discrimination that arises when the plaintiff establishes the elements of a prima facie case. In Texas Dept. of Community Affairs v. Burdine, 450 U.S. 248, 254, 101 S.Ct. 1089, 1094, 67 L.Ed.2d 207 (1981), the Court explained that the “[establishment of the prima facie case in effect creates a presumption that the employer discriminated against the employee.” This presumption “places upon the defendant the burden of producing an explanation to rebut the prima facie ease — i.e., the burden of producing evidence that the adverse employment actions were taken ‘for a legitimate nondiseriminatory reason.’” St. Mary’s Honor Ctr. v. Hicks, 509 U.S. 502, 506, 113 S.Ct. 2742, 2747, 125 L.Ed.2d 407 (1993), (quoting Burdine, 450 U.S. at 254, 101 S.Ct. at 1094). But while this presumption shifts the burden of production to the defendant, “‘[t]he ultimate burden of persuading the trier of fact that the defendant intentionally discriminated against the plaintiff remains at all times with the plaintiff.’ ” Hicks, 509 U.S. at 507, 113 S.Ct. at 2747 (quoting Burdine, 450 U.S. at 253, 101 S.Ct. at 1093-94). “In this regard,” Hicks observed, “it operates like all presumptions, as described in Federal Rule of Evidence 301.” 509 U.S. at 507, 113 S.Ct. at 2747.
Critical for present purposes is what happens when the defendant satisfies its production burden. Burdine addressed this question in the following passage:
If the defendant carries this burden of production, the presumption raised by the prima facie case is rebutted,10 and the factual inquiry proceeds to a new level of specificity.
10 See generally J. Thayer, Preliminary Treatise on Evidence 346 (1898). In saying that the presumption drops from the case, we do not imply that the trier of fact no longer may consider evidence previously introduced by the plaintiff to establish a prima facie case. A satisfactory explanation by the defendant destroys the legally mandatory inference of discrimination arising from the plaintiff's initial evidence. Nonetheless, this evidence and inferences properly drawn therefrom may be considered by the trier of fact on the issue of whether the defendant’s explanation is pretextual....
450 U.S. at 255 & n. 10, 101 S.Ct. at 1095 & n. 10.
Hicks similarly explained that, if the defendant meets its production burden, “the McDonnell Douglas framework — with its presumptions and burdens — is no longer relevant” and should not be “resurreet[ed.]” 509 U.S. at 510, 113 S.Ct. at 2748. “The presumption, having fulfilled its role of forcing the defendant to come forward with some response, simply drops out of the picture,” and “the trier of fact proceeds to decide the ultimate question: whether plaintiff has proven that the defendant intentionally discriminated against him [on the ground alleged.]” Hicks, 509 U.S. at 510-11, 113 S.Ct. at 2749 (quoting Burdine, 450 U.S. at 253, 101 S.Ct. at 1093-94).
B. I interpret these passages to mean that the McDonnell Douglas presumption is governed by the “bursting bubble” theory associated with Professor James Bradley Thayer. See 2 McCormick on Evidence § 344 at 462 (4th ed. 1992). Under this theory, “the only effect of a presumption is to shift the burden of producing evidence with regard to the presumed fact. If that evidence is produced by the adversary, the presumption is spent and disappears.” Id. The case then proceeds “as though there had never been a presumption at all.” 1 Wein-stein’s Evidence ¶ 300[01] at 300-4 (1996)(footnote omitted).
That Burdine and Hicks regarded the McDonnell Douglas presumption as governed by this theory is suggested by the following. First, Burdine states that the term “presumption” properly ‘“refers only to a. device for allocating the production burden,’ ” 450 U.S. at 255 n. 8, 101 S.Ct. at 1094 n. 8 (emphasis added) (citation omitted) — the orthodox “bursting bubble” view. Second, both Burdine and Hicks employ classic “bursting bubble” language to describe what happens to a presumption if the defendant satisfies its production burden: the presump*1080tion then “drops from the case,”2 “drops out of the picture,”3 is “no longer relevant,”4 and should not be “resurrect[ed].”5 Third, Burdine’s footnote 10, which was quoted above, appears to set out the pure “bursting bubble” theory: once the presumption is burst, all that remains is “the plaintiffs initial evidence” and “inferences properly drawn therefrom.” 450 U.S. at 255 n. 10, 101 S.Ct. at 1095 n. 10. In addition, this footnote begins with a citation to Thayer, who stated that once the opponent of a presumption offers sufficient counterproof, “[a]ll is then turned into an ordinary question of evidence, and the two or three general facts presupposed in the rule of presumption take their place with the rest, and operate, with their own natural force, as a part of the total mass of probative matter.” James B. Thayer, Preliminary Treatise on Evidence 346 (1898). Finally, both Burdine and Hicks invoke Federal Rule of Evidence 301, which has generally been interpreted as embodying the “bursting bubble” theory. See e.g., McKenna v. Pacific Rail Serv., 32 F.3d 820, 829-30 (3d Cir.1994); id. at 841 (Mansmann, J., dissenting); AC. Aukerman Co. v. R.L. Chaides Constr., 960 F.2d 1020, 1037-1038 (Fed.Cir.1992); In re Yoder Co., 758 F.2d 1114, 1120 & n. 13 (6th Cir.1985); Legille v. Dann, 544 F.2d 1, 5-7 (D.C.Cir.1976); 10 Moore’s Federal Practice § 301.04 [4.-1] at III-22 (1995-96 Supp.); 1 Weinstein’s Evidence 301-9; 9 Wigmore on Evidence § 2491(2) (3d ed. 1940).
In McKenna, supra, all of the panel members agreed that the McDonnell Douglas presumption and Federal Rule of Evidence 301 embody the “bursting bubble” theory. Writing for the majority, Judge Lewis concluded that the McDonnell Douglas presumption is governed by Federal Rule of Evidence 301 and that under this rule “the introduction of evidence to rebut a presumption destroys that presumption, leaving only that evidence and its inferences to be judged against the competing evidence and its infer-enees to determine the ultimate question at issue.” 32 F.3d at 830. Judge Lewis then went on to hold that the same rule applies under the New Jersey Law Against Discrimination. Id. In dissent, Judge Mansmann agreed that the “bursting bubble” theory applies to federal claims, but contended that the New Jersey presumption has a more durable effect. Specifically, she wrote that “the federal rule ‘bursts the bubble’ of the presumption, while the New Jersey rule creates an issue for the jury.” Id. at 841 (Mansmann, J., dissenting). With respect to their analysis of federal law, I think that both the McKenna majority and dissent were right.
C. The version of Rule 301 that was proposed by the Advisory Committee and promulgated by the Supreme Court rejected the “bursting bubble” theory in favor of the theory advocated by Professor Edmund M. Morgan and others.6 Proposed Rule of Evidence 301 and Advisory Committee Note, 56 F.R.D. 183, 208 (1973). Under this proposed rule, a presumption would have shifted the burden of persuasion. Id.
Congress, however, rejected this proposal. The House of Representatives instead adopted a rule that represented an “intermediate position” between the “Thayer” and “Morgan” theories. H.R.Rep. No. 93-650, 93d Cong., 1st Sess. 7 (1973), reprinted in 1974 U.S.C.C.A.N. 7075, 7081. The House rule provided that “a presumption imposes on the party against whom it is directed the burden of going forward with the evidence, and, even though met with contradicting evidence, is sufficient proof of the fact presumed to be considered by the trier of fact.” 120 Cong.Rec. 2370 (1974).
In the Senate, the Advisory Committee and the Judicial Conference’s Standing Committee on Practice and Procedure advocated a return to the originally proposed version of Rule 301. Dismissing the House compromise as conceptually unsound, the committee argued that “[t]he basic choice is between *1081the so-called ‘bursting bubble’ theory and one shifting the burden of persuasion.” Reporter’s Memorandum, for the Advisory Committee on Evidence and the Standing Committee, to the Senate Judiciary Committee, reprinted in 1 Weinstein’s Evidence at 301-3. The Senate rejected the House approach and passed the present version of Rule 301. S.Rep. No. 93-1277, 93d Cong., 2d Sess. reprinted in 1974 U.S.C.C.A.N. 7051, 7056; 120 Cong.Ree. 37,085 (1974).
The Advisory Committee and the Committee on Rules of Practice and Procedure then turned to the Conference Committee and again urged the adoption of the originally proposed version of Rule 301, arguing once more that “[bjasically the choice to be made in treating the effect to be given presumptions lies between giving them only the effect of a ‘bursting bubble’ and giving them the greater effect of imposing a burden of disproof once evidence has established the conditions that call the presumption into operation.” See 1 Weinstein’s Evidence at 301-7. However, the Conference Committee recommended adoption of the Senate version, Conf. Rep. No. 93-1597, 93d Cong., 2d Sess. (1974), reprinted in 1974 U.S.C.C.A.N. 7098, 7099, and this recommendation was enacted. 120 Cong.Ree. 40,070, 40,897 (1974).
Rule 301 states:
In all civil actions and proceedings not otherwise provided for by Act of Congress or by these rules, a presumption imposes on the party against whom it is directed the burden of going forward with evidence to rebut or meet the presumption, but does not shift to such party the burden of proof in the sense of the risk of nonpersuasion, which remains throughout the trial upon the party on whom it was originally cast.
Like the members of the McKenna panel and the other previously mentioned authorities, I think that the most reasonable interpretation of Rule 301 is that it incorporates the “bursting bubble” theory. The text of the rule supports this conclusion since it does not even hint that a presumption does anything but shift the burden of production. If Congress had intended for a presumption to have any further effect, such as guaranteeing that the presumed fact would not be rejected at the summary-judgment or judgment-as-a-matter-of-law stage, I think that Congress would have said so in the text of the rule. I do not think that Congress would have left it for the courts to divine, without any clue in the language of the rule, that a presumption should have such an important additional effect.
Furthermore, the legislative history suggests that Congress, advised that it was compelled to choose between the Thayer and Morgan approaches, adopted “the pure Thayer rule.” 1 Weinstein’s Evidence at 301-9. Professors Wright and Graham disagree with this position based on a passage in the Conference Report. See 21 Charles A. Wright and Kenneth W. Graham, Jr., Federal Practice and Procedure §§ 5120 at 547 & n. 27, 5126 at 609-10 & nn. 23-24 (1977). Discussing the consequences that follow when the adverse party meets its production burden, the Conference Report states:
If the adverse party does offer evidence contradicting the presumed fact, the court cannot instruct the jury that it may presume the existence of the presumed fact from proof of the basic facts. The court may, however, instruct the jury that it may infer the existence of the presumed fact from proof of the basic facts.
Conf.Rep. 93-1597, supra, at -, reprinted in 1974 U.S.C.C.A.N. at 7099 (emphasis deleted; emphasis added). Because the highlighted sentence discusses jury instructions, Professors Wright and Graham view it as showing that the Conference Committee contemplated that a presumption, even if met with enough counterproof to satisfy the adverse party’s production burden, would nevertheless serve to take the factual question to the jury. 21 Wright and Graham, supra, § 5126 at 610 n. 23.
Like the Sixth Circuit, In re Yoder Co., 758 F.2d at 1119-20 n. 13, I do not find this interpretation compelling. The Sixth Circuit wrote:
The emphasized portion of this quotation is not inconsistent with the bursting bubble theory.... The statement that the jury may be instructed to consider an inference is most naturally read as permitting such *1082an instruction when called for by the existence of a logical inference.
Id. (emphasis in original). This reading is at least as reasonable as the alternative offered by Professors Wright and Graham, and I am therefore unwilling, based solely on the statement in question, to conclude that Rule 301 was intended to give presumptions an important effect not mentioned in the text of the rule.
I must acknowledge that a passage in Hicks may be read as supporting Professors Wright and Graham’s argument, at least with respect to the McDonnell Douglas presumptions. Hicks stated:
The factfinder’s disbelief of the reasons put forward by the defendant (particularly if disbelief is accompanied by a suspicion of mendacity) may, together with the elements of the prima facie case, suffice to show intentional discrimination. Thus, rejection of the defendant’s proffered reasons will permit the trier of fact to infer the ultimate fact of intentional discrimination,4 and the Court of Appeals was correct when it noted that, upon such rejection, “[n]o additional proof of discrimination is required.”
4 Contrary to the dissent's confusion-producing analysis, post, at 535-536, there is nothing whatever inconsistent between this statement and our later statements that (1) the plaintiff must show "both that the reason was false, and that discrimination was the real reason," infra, at 515, and (2) ‘‘it is not enough ... to disbelieve the employer,” infra, at 519. Even though (as we say here) rejection of the defendant's proffered reasons is enough at law to sustain a finding of discrimination, there must be a finding of discrimination.
509 U.S. at 511, 113 S.Ct. at 2749 (all emphases in original) (ellipsis in original).
Read in isolation, this passage may seem to suggest that the facts underlying the pri-ma facie case plus the rejection of the employer’s explanation are always sufficient to take the case to the jury. This is the interpretation adopted in Fuentes v. Perskie, 32 F.3d 759, 763 (3d Cir.1994), and by the majority here. See Maj. Op. at 1065-66. However, like the First, Fifth, and Eleventh Circuits, I do not think that this reading is compelled. See Isenbergh, 97 F.3d at 442-45; Rhodes, 75 F.3d at 993-95; Woods, 30 F.3d at 260-61 n. 3. Instead, I believe that this passage can reasonably be interpreted as simply rejecting the “pretext plus” approach that Justice Souter attributed to the Court in his dissent (509 U.S. at 535, 113 S.Ct. at 2762 (Souter, J., dissenting)) to which the Court referred. 509 U.S. at 511 n. 4, 113 S.Ct. at 2749 n. 4. Under the “pretext plus” approach, as summarized in the law review article that Justice Souter cited, “the plaintiff must produce some additional evidence other than the evidence supporting the prima facie case and other than the fact of the defendant’s deception.” Catherine J. Lanctot, The Defendant Lies and the Plaintiff Loses: The Fallacy of the Pretext-Plus’ Rule in Employment Discrimination Cases, 43 Hastings L.J. 57, 87-88 (1991). The previously quoted passage from Hicks may be interpreted to mean that such additional proof (i.e., the “plus” in “pretext plus”) is not always required. Accordingly, when the Court wrote (509 U.S. at 511, 113 S.Ct. at 2749) that “rejection of the defendant’s proffered reasons will permit the trier of fact to infer the ultimate fact of intentional discrimination,” what the Court may have meant is that the trier of fact is permitted to draw such an inference — in the sense that the trier of fact is not precluded by any legal rule, such as “pretext plus,” from doing so — if such an inference is factually warranted. The same interpretation may be given to the Court’s subsequent statement that “upon [the] rejection [of the defendant’s proffered reason] ‘[n]o additional proof of discrimination is required,’ ” 509 U.S. at 511, 113 S.Ct. at 2749. ‘“No additional proof is required”’ in that there is no blanket legal requirement of such proof, as there would be under the “pretext plus” approach. Similarly, the “rejection of the defendant’s proffered reasons is enough at law to sustain a finding of discrimination” (id. at 511 n. 4, 113 S.Ct. at 2749 n. 4) in the sense that no other blanket legal requirement, such as proving the “plus,” is needed. I favor this interpretation because it is consistent with my conclusion, discussed above, that Rule 301 presumptions in general and the McDonnell Douglas presumption in particular are governed by the “bursting bubble” theory.
*1083D. If it is trae, as I believe, that the McDonnell Douglas presumption is governed by the pure “bursting bubble” theory, it follows that the majority’s blanket rule barring summary judgment or judgment as a matter of law in favor of the defendant is unsound. It is well recognized that under the pure “bursting bubble” theory, once a presumption is destroyed, the proponent of the presumption has no guarantee that its case will go to the trier of fact. One treatise states that the previously presumed fact may then be found “only if the natural probative force of the basic facts that brought the presumption into play is sufficient to support such a finding (or the evidence as a whole supports it). Otherwise, the presumed fact may not be found, and the presumption does not protect this possibility.” 1 Christopher B. Mueller and Laird C. Kirkpatrick, Federal Evidence § 70 at 332 (1994) (emphasis added). Another observes: “The opponent of the presumption may still not be entitled to a directed verdict, but if its motion is denied, the ruling will have nothing to do with the existence of a presumption.” 2 McCormick on Evidence, § 344 at 462. See also 21 Wright and Graham, Federal Practice § 5122 at 564. In Thayer’s words, after the presumption is spent, “[a]ll is then turned into an ordinary question of evidence.” Thayer, Preliminary Treatise at 346.
III.
Once it is recognized that the McDonnell Douglas presumption disappears in the face of sufficient counterproof, the test to be applied in ruling on a defense motion for summary judgment or judgment as a matter of law is clear: the motion should be granted if all of the evidence in the record and the inferences that may be drawn from that evidence could not persuade a rational trier of fact that intentional discrimination on the ground alleged by the plaintiff was a determinative cause of the challenged action. This test follows from the following three rules:
1.In a McDonnell Douglas case, “[t]he ultimate burden of persuading the trier of fact that the defendant intentionally discriminated against the plaintiff remains at all times with the plaintiff.” Burdine, 450 U.S. at 253, 101 S.Ct. at 1093; see also Hicks, 509 U.S. at 507, 113 S.Ct. at 2747.
2. This burden requires proof that discrimination on the ground alleged was a determinative cause of the challenged action. Miller v. CIGNA Corp., 47 F.3d 586 (3d Cir.1995) (in banc).
3. Summary judgment and judgment as a matter of law are proper where a trier of fact could not rationally return a verdict in favor of the opposing party. Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 250-51, 106 S.Ct. 2505, 2511-12, 91 L.Ed.2d 202 (1986); Celotex Corp. v. Catrett, 477 U.S. 317, 322-23, 106 S.Ct. 2548, 2552-53, 91 L.Ed.2d 265 (1986).
Putting these rules together, it is apparent that a defense motion for summary judgment or judgment as a matter of law should be granted where a trier of fact could not rationally find based on the record (and giving no further consideration to the presumption as such) that discrimination on the ground alleged was a determinative cause of the challenged action.
The majority, however, takes the position that such a motion must be denied whenever there is enough evidence to support a finding that the explanation offered by the employer is untrue. See Maj. Op. at 1066. This position can be correct only if proof that is sufficient to justify disbelief of the employer’s explanation (the majority’s test) will always be enough to justify a finding by a rational trier of fact that discrimination on the ground alleged was a determinative cause of the challenged action. Unless this relationship invariably holds true — and I will show below that it does not (see part VI, infra) — the majority’s blanket rule is wrong.
IV.
Before turning directly to that question, however, it is helpful to examine the types of evidence of discrimination that may exist in the record of á discriminatory-treatment case after the McDonnell Douglas presumption is burst and sufficient evidence to show pretext is offered. This evidence, may be grouped into four categories.
*1084A. First, there are the facts that the plaintiff proved to make out the prima facie case and the inferences to which those facts naturally give rise. In some cases, the “natural probative force”7 of these facts is substantial, but in others it is relatively weak. “The burden of establishing a prima facie case of disparate treatment is not onerous,” Burdine, 450 U.S. at 253, 101 S.Ct. at 1094, and “evidence sufficient to make out a prima facie case is not always sufficient to support the ultimate finding of intentional discrimination. ... Where the evidence behind the prima facie showing is strong, it may, standing alone, justify a finding of intentional discrimination. ... But where the prima facie case is based on minimal evidence, it cannot.” United States v. Redondo-Lemos, 27 F.3d 439, 442 (9th Cir.1994). An example which the prima facie case gives rise to only a weak inference of discrimination is the age discrimination case in which a somewhat younger person is hired, promoted, retained, etc. instead of a slightly older person. Under our cases, a plaintiff in an age discrimination suit can make out a prima facie case by showing that he or she (1) was over 40 years of age at the time in question, (2) applied for and was minimally qualified for the job, (3) but was rejected (4) in favor of a person who was “ ‘sufficiently younger to permit an inference of age discrimination.’ ” Barber v. CSX Distribution Services, 68 F.3d 694, 698 (3d Cir.1995), quoting Fowle v. C & C Cola, 868 F.2d 59, 61 (3d Cir.1989). With respect to the final element of this formulation, we have held that an eight-year gap is enough and have suggested that a five-year difference may suffice. Barber, 68 F.3d at 695. Thus, I believe that a plaintiff claiming that he or she was denied a promotion due to age discrimination could make out a prima facie case under our circuit law by showing, for example, that he or she (1) was 43 years old at the time in question, (2) had the minimum requisite qualifications, (3) and applied for the job, but that (4) the employer instead chose a 35-year-old.
I agree that these facts permit an inference of age discrimination, but I do not think that a reasonable person could give these facts, standing alone, much weight. A way to test this hypothesis is to ask: if you knew nothing more about this promotion decision, how much would you bet that, when all the facts were revealed, the reason for the decision would turn out to be age discrimination? I think that few people would be willing to wager much , on this bet — -precisely because the natural probative force of the few facts needed to establish the prima facie ease is slight.
A similar situation can arise in cases of alleged national origin or religious discrimination. In order to make out a prima facie case of discrimination on these grounds, a plaintiff must show that he or she (1) is a member of a national origin group or adherent of a religion, (2) applied or was considered for the position or benefit, and (3) was at least minimally qualified, but (4) the position or benefit went to a person of another national origin or religion. See, e.g., Vitug v. Multistate Tax Comm’n, 88 F.3d 506, 515 (7th Cir.1996) (plaintiff established a prima facie case by showing that he was a qualified Filipino Catholic and that promotion went to “a non-Filipino and born-again Christian”). How strong an inference of national origin or religious discrimination naturally arises from such a bare showing?
Suppose that all that you knew about an employment decision was that the employer rejected an applicant who traced his or her national origin to one country and instead chose another person who did not trace his or her national origin to that particular country. How confident would you be that the reason for the decision, when all the facts were revealed, would turn out to be national origin discrimination?
Or suppose that all that you knew about an employment decision was that the employer rejected a minimally qualified Protestant, Catholic, Jew, Muslim, or adherent of another religion in favor of another employee of a different religion. How confident would you be that the reason for the decision, when all the facts were revealed, would be intentional discrimination on the basis of religion? In my view, most rational people, without knowing more, would have little confidence that *1085the real reason would turn out to be national origin or religious discrimination, and I therefore submit that these bare facts naturally give rise to only a slight inference of discrimination. And, although I will not belabor the point by posing further examples, I think that the natural probative' force of the prima facie case is sometimes weak in race and gender eases as well. In short, I submit that the natural probative force of the facts underlying the prima facie case varies from case to case and that in some cases that force will be too weak on its own to sustain a finding of discrimination.
B. Second, there is the inference of discrimination that may often be drawn from the ruling out of the employer’s explanation. The degree to which such an inference is justified, however, is inversely proportional to the degree to which the record contains evidence of a third possible explanation for the challenged employment action. For example, if it is certain that an employee was discharged for either reason “a” or reason “b” and no other, and if reason “b” can be ruled out, then obviously it may be inferred that the real reason for the discharge was “a.” But if an employee was discharged for either reason “a,” reason “b,” or reason “e,” then ruling out reason “b” does not necessarily permit a strong inference that reason “a” was the real reason.
Cases in which the record contains strong evidence of a third explanation for the challenged action are by no means unknown. See Miller v. CIGNA Corp., 47 F.3d 586, 597 (3d Cir.1995). Perhaps the clearest examples are eases in which the plaintiff challenges a single adverse employment action based on two or more alternative grounds, a rather common tactic. See, e.g., Roxas v. Presentation College, 90 F.3d 310 (8th Cir.1996) (plaintiff alleged race, national origin, gender, and age discrimination); Lawrence v. Natl Westminster Bank of N.J., 98 F.3d 61 (3d Cir.1996); Rabinovitz v. Pena, 89 F.3d 482 (7th Cir.1996) (age, religion, and retaliation); Ford v. Bernard Fineson Dev. Ctr., 81 F.3d 304 (2d Cir.1996) (race, age, and gender); Hartsel v. Keys, 87 F.3d 795 (6th Cir.1996), cert. denied, — U.S. -, 117 S.Ct. 683, — L.Ed.2d -(1997) (age, gender, and retaliation); Evans v. Technologies Applications & Service Co., 80 F.3d 954 (4th Cir.1996), (age and gender); Austin v. Owens-Brockway Glass Container, Inc., 78 F.3d 875 (4th Cir.1996), cert. denied, — U.S. -, 117 S.Ct. 432, 136 L.Ed.2d 330 (1996) (gender and disability); Americanos v. Carter, 74 F.3d 138 (7th Cir.1996), cert. denied, — U.S. -, 116 S.Ct. 1853, 134 L.Ed.2d 953 (1996) (age, gender, and national origin); Castillo v. Frank, 70 F.3d 382 (5th Cir.1995) (age, gender, and national origin); Meinecke v. H & R Block, 66 F.3d 77 (5th Cir.1995) (age and gender); Johnson v. Office of Senate Fair Employment Practices, 35 F.3d 1566 (Fed.Cir.1994) (gender and religion); Dashnaw v. Pena, 12 F.3d 1112, 1114 (D.C.Cir.1994), cert. denied, — U.S. -, 115 S.Ct. 417, 130 L.Ed.2d 333 (1994) (age, national origin, religion, and race); Sarsha v. Sears, Roebuck & Co., 3 F.3d 1035 (7th Cir.1993) (age and gender).
To take one of the many possible combinations of Claims, an employee might contend that he or she did not get a promotion (1) because of gender and (2) because of disability. If the record in such a case contains strong evidence of disability discrimination, rejection of the employer’s explanation (let us say, inferior qualifications) will not by itself permit a strong inference that the employer’s true reason was gender discrimination.
The degree to which a trier of fact can reasonably conclude that there was discrimination on the ground claimed by the plaintiff also depends upon the degree to which the trier of fact can reasonably reject the employer’s reason(s). (It is important to bear in mind that acceptance or rejection of an employer’s reasons need not be an all-or-nothing proposition.) The evidence in a particular case may be such as to justify only a marginal or partial disbelief or belief of the employer’s reason(s). For example, a trier of fact might be justified in believing that it is more probable than not (but barely so) that the employer’s explanation is false. Or, a trier of fact might be justified in believing that it is more probable than not (but barely so) that the employer’s explanation is true. In addition, a trier of fact might be justified *1086in believing that the reason asserted by the employer was not the sole cause but was a partial cause (say a 20%, 40%, 60%, or 80% cause) for the challenged action. Or, if the employer asserts multiple reasons, the evidence might be such as to justify belief (to some degree) of some reasons but not others. All other things being equal, the more strongly and completely the trier of fact can rationally rule out the employer’s reason(s), the more justified it is to conclude that there was discrimination on the ground alleged— and vice versa.
C. Third, disbelief of the employer’s explanation may also give rise to an inference that the employer was trying to conceal discrimination on the ground that the plaintiff claims. But the strength of this inference, too, ■ will vary based on the facts. Its strength will depend on whether there is evidence in the record of some other possible explanation that the employer might not want to disclose (e.g., in my prior hypothetical, disability discrimination).8 In addition, its strength will obviously be proportional to the extent and strength of the trier of fact’s disbelief of the employer’s reason(s).
D. The fourth category consists of any other relevant evidence of discrimination on the ground asserted. , “[Sjtray remarks in the workplace” that are insufficient to make out a Price Waterhouse case are an example. See Price Waterhouse v. Hopkins, 490 U.S. 228, 277, 109 S.Ct. 1776, 1805, 104 L.Ed.2d 268 (1989) (O’Connor, J., concurring). When evidence falling into this category is present, it should always be taken into account in determining whether a defense motion for summary judgment or judgment as a matter of law should be granted — something that the Fuentes test does not accomplish, as I will show below. See infra at 1087-88. But of course, many McDonnell Douglas cases lack evidence of this sort.
V.
Having surveyed the types of evidence of discrimination that are relevant for present purposes, I return to the question whether there will always be enough evidence to support a verdict of discrimination on the ground alleged by the plaintiff after the McDonnell Douglas presumption has burst and sufficient evidence of pretext has been adduced. To ensure that I am not misunderstood, I emphasize again that I think that under these circumstances there will usually be sufficient evidence — but not always. Specifically, where the natural probative force of the facts offered to prove the prima facie case is weak, where there is strong evidence in the record that the challenged employment action was attributable to some factor other than the discriminatory ground alleged by the plaintiff or the explanation offered by the employer, and where there is no other evidence that the action was due to discrimination on the ground alleged, the evidence of discrimination on that ground may be insufficient to sustain a verdict and thus insufficient to defeat summary judgment or judgment as a matter of law.
Here is an example. A man with a disability applies for a promotion, but the promotion is given to a woman without a disability. See, e.g., Antol v. Perry, 82 F.3d 1291 (3d Cir.1996)(man who was denied promotion sued for gender and disability discrimination). The plaintiff establishes a prima facie case of gender discrimination by showing that he is a man, he was qualified for the job, he applied, but it was given to a woman. There is no other evidence of gender discrimination against men; indeed, there is evidence that the decisionmaker was a man and that the great majority of the employees in the job category at issue were men. The employer says that the woman was chosen because her qualifications were better, but the plaintiff shows that this explanation is *1087full of holes. In addition and most important, there is strong, direct evidence that the decisionmaker was biased against the plaintiff because of his disability. Among other things, there is testimony that, when the decisionmaker learned that the plaintiff had applied for the promotion, the decisionmaker made insulting and derogatory remarks about the plaintiffs disability.
Is there enough evidence in this ease to convince a rational trier of fact that the promotion decision was based on gender discrimination? I think not, and if I am right, then the majority’s blanket test is disproven. Consider the evidence in each of the four categories that I previously described. The facts underlying the prima facie case of gender discrimination — basically that the plaintiff was qualified but a woman was given the job instead — have little natural probative force for the purpose of showing gender discrimination; a qualified man might be rejected in favor of a woman for many other reasons. As for the evidence that the employer’s explanation (the woman’s supposedly better qualifications) was untrue, ruling out the employer’s explanation still leaves both gender and disability discrimination as possible explanations, and in light of the strong evidence of disability discrimination, does little to show gender discrimination. Likewise, while the employer’s conduct in proffering a false explanation permits an inference that the employer’s real reason was one that it wished to hide, that inference does little to show that the employer’s real reason was gender discrimination because there is strong evidence of another explanation (disability discrimination) that the employer would also have a motive to conceal. Therefore, I submit that the evidence in this hypothetical case is insufficient to support a reasonable finding of gender discrimination and that that theory should be weeded from the case before the plaintiffs claim of disability discrimination is sent to the trier of fact.
The majority’s test, however, would require that the gender discrimination claim be sent to the trier of fact as well. Indeed, if the plaintiff had added claims of age, race, national origin, and religious discrimination and was able to do the minimum necessary to establish a prima facie case on all of those grounds, all of those claims would have to be sent to the trier of fact as well. I do not think that such results can be squared with the cardinal rules that the plaintiff in a discriminatory treatment ease must prove intentional discrimination on each ground alleged and that summary judgment or judgment as a matter of law must be granted if the record does not rationally support a judgment for the non-moving part on each such ground.9 See, e.g., Deborah C. Malamud, The Last Minuet: Disparate Treatment After Hicks, 93 Mieh.L.Rev. 2229, 2306-2311 (1995).
VI.
Recognition of what I believe is the correct test would clarify the analysis of discriminatory-treatment cases and would avoid analytical difficulties that the majority’s test creates. I will discuss two of these difficulties here.
A. The first concerns cases in which the employer offers a multi-part explanation for the challenged action. The Fuentes opinion, which the majority follows, first states that, when an employer asserts that the challenged action was taken for several rea'sons, the evidence “must allow a factfinder reasonably to infer that each of the employer’s proffered nondiscriminatory reasons ... was either a post hoc fabrication or otherwise did not actually motivate the employment ac-tion_” 32 F.3d at 764 (emphasis in original) (citations omitted). However, Fuentes qualifies this statement by adding that “[i]f the defendant proffers a bagful of legitimate reasons, and the plaintiff manages to cast substantial doubt on a fair number of them, the plaintiff may not need to discredit the remainder.” Id. at 764 n. 7 (emphasis added).
I am not sure what these rules mean. What is the difference between “several rea*1088sons” and “a bagful”? What is a “fair number” of a “bagful”? If the employer offers three reasons, each of which, it says, was equally important in the challenged decision, how many must the plaintiff knock down? All of them? Or will two suffice? (I assume that one would not be a “fair number.”) What if the employer says that reason one was the most important, counting for, say, 40%, and that reasons two and three each counted for 30%? Would it be enough for the plaintiff to refute reason one? If so, would refuting reason two by itself also suffice? Why? I don’t know, and I don’t think that it is possible to provide a satisfactory answer to questions of this sort within the framework of the majority’s test.
By contrast, these problems disappear if what I contend is the correct test is used. No matter how many reasons the employer offers and no matter what combination of reasons the plaintiff succeeds in knocking down, the dispositive question remains the same: taking into account all of the evidence in the record, including whatever inferences and deductions can rationally be drawn from the rejection of some (or all) of the employer’s proffered reasons, is there enough proof to permit a rational trier of fact to find that intentional discrimination on the ground alleged was a determinative cause of the challenged action?
B. The second analytical difficulty concerns the ability of a plaintiff to survive summary judgment or judgment as a matter of law by combining the evidence that necessarily remains after the McDonnell Douglas presumption has burst and sufficient evidence of pretext has been offered (the facts underlying the prima facie ease and the inferences that logically flow from the rejection of the employer’s explanation) with any other direct and circumstantial evidence of discrimination that the plaintiff may be able to find. Under Fuentes, a plaintiff may defeat summary judgment or judgment as a matter of law by “either (i) discrediting the proffered reasons, either circumstantially or directly, or (ii) adducing evidence, whether circumstantial or direct, that discrimination was more likely than not a motivating or determinative cause of the adverse employment action.” Fuentes, 32 F.3d at 764 (emphasis in original).
Why can’t a plaintiff potentially satisfy his or her burden by combining some evidence from each of these categories? For example, if a plaintiff can almost, but not quite, show enough “weaknesses, implausibilities, inconsistencies, incohereneies, or contradictions in the employer’s proffered legitímate reasons” to come within the first part of this test, id. at 765, why can’t the plaintiff make it over the hump by adding a bit of evidence from the second part, i.e., other direct or circumstantial evidence of discrimination? Fuentes doesn’t explain, and I don’t think that any good explanation is possible.
In sum, I submit that the majority’s test is both wrong and unwieldy and that the correct test is simply whether a rational trier of fact could find, based on the record, that discrimination on the ground alleged was a determinative cause of the challenged employment action.
VII.
Applying this test, I agree with the district judge’s analysis of the record. The record shows great personal friction between the plaintiff and her supervisors regarding matters such as grooming, smoking, tardiness, and giving away free food and beverages, but the district judge saw little if any evidence of any kind that could reasonably link this personal animosity to the plaintiffs gender. The district judge wrote that she was “left searching the record for evidence that gender played a determinative role in defendant’s conduct.”
In reaching the opposite conclusion, the majority apparently relies on the following: (a) “no woman had ever held the position of Manager of Restaurants,” (b) “a man replaced Sheridan as Head Captain of the Green Room morning shift,” (c) “Amblard had told Sheridan he would watch her like a ‘hawk’ and a ‘dog’,” and (d) Amblard “ignored] her and [spoke] instead to one of her male supervisors if one was present.” Maj. Op. at 1064. But factor (a) gives rise to only a weak inference of gender discrimination because the plaintiff was found to be unquali*1089fied for the position of Manager of Restaurants. Factor (b) likewise gives rise to only a weak inference since two of the other four head captains were women. Factor (c) does not provide a link to gender; nor does factor (d). Is all of this evidence taken together enough to persuade a rational factfinder that the reason for the animosity between the plaintiff and her supervisors was gender rather than sheer personal antipathy? The district judge concluded that it was not, and I am inclined to agree with her analysis.
I also agree with the district judge’s conditional grant of a new trial. The district judge applied the correct legal standard. She recognized that “[a] new trial cannot be granted ... merely because the court would have weighed the evidence differently and reached a different conclusion.” Sheridan II at 12 (quoting Markovich v. Bell Helicopter Textron, Inc., 805 F.Supp. 1231, 1235 (E.D.Pa.), aff'd, 977 F.2d 568 (3d Cir.1992)). Instead, the court stated, a new trial may be granted on the ground that the verdict was against the weight of the evidence only when the failure to do so would result in injustice or shock the conscience of the court. Id. Moreover, in assessing the evidence, the court recognized that disbelief of defendant’s proffered reason was evidence of discrimination. Accordingly, I believe that the district judge applied the correct legal precepts in ruling on defendant’s new trial motion, and because I do not think that she abused her discretion in granting the motion, I would affirm.
. See St. Mary’s Honor Ctr. v. Hicks, 509 U.S. 502, 535, 113 S.Ct. 2742, 2762, 125 L.Ed.2d 407 (1993) (Souter, J., dissenting); Catherine J. Lanctot, The Defendant Lies and the Plaintiff Loses: The Fallacy of the ‘Pretext-Plus’ Rule in Employment Discrimination Cases, 43 Hastings L.J. 57, 81-83 (1991) (explaining the "pretext plus" rule).
. Burdine, 450 U.S. at 254 n. 10, 101 S.Ct. at 1095 n. 10.
. Hicks, 509 U.S. at 511, 113 S.Ct. at 2749.
. Id. at 510, 113 S.Ct. at 2748-49.
. Id.
. See, e.g., Edmund M. Morgan, Instructing the Jury upon Presumptions and Burden of Proof, 47 Harv.L.Rev. 59, 82 (1933).
. 1 Mueller and Kirkpatrick, Federal Evidence § 70 at 333.
., Or imagine a situation when the employer promotes A (a white male) over B (a white female) because A was related to the employer. The employer may not wish to disclose his real reasons for not promoting B because the news as to his criteria for promoting B over A would likely hurt his reputation and lower employee morale. "[A]n employer may offer a pretextual reason for a personnel decision that is nonetheless nondiscriminatory." Fisher v. Vassar College, 70 F.3d 1420, 1437 (2d Cir.1995); Cf. Michael Selmi, Testing for Equality: Merit, Efficiency, and the Affirmative Action Debate, 42 UCLA L.Rev. 1251, 1303 (1995) (describing a set of economic models that tie worker effort to perceptions of fairness).
. The majority’s test would of course he correct if a plaintiff, in order to establish a prima facie case, were required to prove facts that are sufficient to support a finding of discrimination on the ground alleged. At least under our circuit law, however, the establishment of a prima facie case does not demand such a heightened showing.