suggesting that requiring proof of a causal connection in retaliation cases is inconsistent with a McDonnell Douglas- type burden-shifting analysis
How later courts described this case
- suggesting that requiring proof of a causal connection in retaliation cases is inconsistent with a McDonnell Douglas- type burden-shifting analysis
- "Someday we'll have to decide what the prima facie case of retaliation is in the Seventh Circuit.”
- affirming summary judgment because plaintiff had “insufficient evidence demonstrating a link between reporting his supervisor’s conduct and his termination”
- “Illinois law allows claims for retaliatory discharge when an employee is terminated for filing a workers’ compensation claim or because the employee has reported the employer’s criminal conduct, either to law enforcement personnel or to the company itself.” (citations omitted)
Written by the judges who cited it.
The opinion
POSNER, Circuit Judge,
concurring.
I join the panel’s opinion but write separately to flag for future consideration an issue (ignored by the parties, hence waived) that is implicit in the following sentence in the opinion: “Illinois retaliatory discharge cases brought in federal court may be analyzed using the burden-shifting method presented in McDonnell Douglas [Corp. v. Green, 411 U.S. 792 , 93 S.Ct. 1817 , 36 L.Ed.2d 668 (1973)]” (emphasis added). In support of this proposition, the opinion cites Hiatt v. Rockwell Int’l Corp., 26 F.3d 761, 767 (7th Cir.1994), which in turn cites McEwen v. Delta Air Lines, Inc., 919 F.2d 58 (7th Cir.1990). Neither case cites any authority for the proposition other than Hiatt’s citation of McEwen . The only reason given for the result is a statement in McEwen that since the McDonnell Douglas standard regulates merely the order of proof, it is procedural *474 rather than substantive within the meaning of the Erie doctrine and it therefore governs diversity cases as well as federal-question cases. So even though the Supreme Court of Illinois expressly rejected the application of McDonnell Douglas to Illinois retaliatory-discharge cases in Clemons v. Mechanical Devices Co., 184 Ill.2d 328 , 235 Ill.Dec. 54 , 704 N.E.2d 403, 407-08 (1998), should such a case wander into federal court by virtue of the diversity jurisdiction the court should apply McDonnell Douglas to it.
Our sister circuits have assumed, to the contrary, that state law, not federal law, governs the burden-shifting standard applicable to cases in which the rule of decision is state rather than federal. E.g., Perry v. Woodward, 199 F.3d 1126, 1141-42 (10th Cir.1999); Norville v. Staten Island University Hospital, 196 F.3d 89, 95 (2d Cir.1999); Payne v. Norwest Corp., 185 F.3d 1068, 1073-74 (9th Cir.1999); Carpenter v. Federal Nat’l Mortgage Ass’n, 165 F.3d 69, 72 (D.C.Cir.1999); Mullin v. Raytheon Co., 164 F.3d 696, 699 (1st Cir.1999); King v. Herbert J. Thomas Memorial Hospital, 159 F.3d 192, 198 (4th Cir.1998); Lee v. State of Minnesota, Dept. of Commerce, 157 F.3d 1130, 1133 (8th Cir.1998); Nichols v. Lewis Grocer, 138 F.3d 563, 565-66 (5th Cir.1998); Olson v. General Electric Astrospace, 101 F.3d 947, 956 (3d Cir.1996); Pierce v. Commonwealth Life Ins. Co., 40 F.3d 796 , 802 n. 8 (6th Cir.1994). These cases do not discuss the choice of law issue; they merely take for granted that the McDonnell Douglas standard is substantive for Erie purposes; nor have I found any illuminating scholarly discussions of the issue. But I am inclined to think that the position (or better perhaps the instinct) of these other circuits is correct.
To describe the McDonnell Douglas standard as merely prescribing the order of proof and the allocation of burdens of production, though common, see, e.g., Reeves v. Sanderson Plumbing Products, Inc., - U.S. -, 120 S.Ct. 2097, 2106 , 147 L.Ed.2d 105 (2000), and literally correct, is an invitation to misunderstanding. Ordinarily, if all a plaintiff in a discrimination case could show was that he was qualified for a job (or promotion, or retention) but was passed over in favor of a person of a different race (sex, religion, etc.), his opponent would be entitled to summary judgment, because no reasonable jury could infer from these bare facts that it was more likely than not that the employer’s action was invidious. Under McDonnell Douglas, this bare showing compels summary judgment for the plaintiff unless the defendant produces evidence of a noninvidious reason for the action complained of. Applied to retaliation, the logic of the McDonnell Douglas standard would compel summary judgment for the plaintiff who showed that after lodging a complaint about discrimination, he (but not any otherwise similarly situated employee who did not complain) was fired or otherwise subjected to an adverse employment action even though he was performing his job in a satisfactory manner, unless, again, the defendant presented a noninvidious reason for the action. Hiatt v. Rockwell Int’l Corp., supra, 26 F.3d at 767 ; McEwen v. Delta Air Lines, Inc., supra, 919 F.2d at 59 . Without the boost given by McDonnell Douglas, the plaintiff in such a case would have to establish, as part of his prima facie case and thus before the defendant had any burden of explanation, a causal connection between the lodging of the complaint and his being fired.
And so the Illinois court concluded in Clemons that to follow McDonnell Douglas in an Illinois retaliatory-discharge case “would, in essence, expand the tort of retaliatory discharge by reducing plaintiffs burden of proving the elements of the tort. Because we refuse to expand the tort of retaliatory discharge, we decline plaintiffs invitation to adopt the three-tier allocation of proof method in retaliatory discharge cases.” 235 Ill.Dec. 54 , 704 N.E.2d at 408 . In like vein the U.S. Supreme Court recently found “substantive” a New York *475 state law regulating review of damages awards because while the statute “con-tainted] a procedural instruction ... the State’s objective [wa]s manifestly substantive.” Gasperini v. Center for Humanities, Inc., 518 U.S. 415, 429 , 116 S.Ct. 2211 , 135 L.Ed.2d 659 (1996). In rejecting McDonnell Douglas for retaliation cases, the objective of the Illinois court likewise was “manifestly substantive,” namely to avoid “expand[ing] the tort of retaliatory discharge.” Or as we put it in S.A. Healy Co. v. Milwaukee Metropolitan Sewerage Dist., 60 F.3d 305, 310 (7th Cir.1995), “the state’s goals are substantive — designed to shape conduct outside the courtroom and not just improve the accuracy or lower the cost of the judicial process — though the means are procedural.”
A practical way to decide whether a rule of state law is substantive or procedural for purposes of the Erie doctrine is to ask whether the rule is limited to a particular substantive area or applies across the board. If the former, it is likely to reflect substantive policy, to which the federal court should defer, and if the latter to be a product of purely procedural concerns that properly may differ between federal and state courts. E.g., Herremans v. Carrera Designs, Inc., 157 F.3d 1118, 1123 (7th Cir.1998); S.A. Healy Co. v. Milwaukee Metropolitan Sewerage Dist., supra, 60 F.3d at 310 . It was on this basis that we suggested in Harbor Ins. Co. v. Continental Bank Corp., 922 F.2d 357, 364 (7th Cir.1990), that the doctrine of “mend the hold,” a type of estoppel, was substantive for Erie purposes insofar as it was limited to contract cases, just like the parol evidence rule, which might otherwise be thought a rule of procedure. “Rules of contract interpretation, such as the parol evidence and four-corners rules, are deemed substantive, because of their effect on the conduct of contracting parties outside the courtroom, even though the rules operate through limiting the kinds of evidence that are admissible.” AM Int'l, Inc. v. Graphic Management Associates, Inc., 44 F.3d 572, 576 (7th Cir.1995). The McDonnell Douglas rule, unlike for example the rule that a motion for summary judgment need not be supported by evidence, Celotex Corp. v. Catrett, 477 U.S. 317, 322-24 , 106 S.Ct. 2548 , 91 L.Ed.2d 265 (1986), is not a general rule of federal procedure; it is tailored for and limited to discrimination cases, as we emphasized in Diaz v. Fort Wayne Foundry Corp., 131 F.3d 711 (7th Cir.1997). It is part of the law of discrimination, which is substantive.
But there is a rather considerable complication. The McDonnell Douglas standard that we apply in most of our retaliation cases is not really the McDonnell Douglas standard. The logical adaptation of McDonnell Douglas to retaliation, the adaptation we assumed in Hiatt and McEwen and that may have led the Supreme Court of Illinois to reject its use in cases under the Illinois law of retaliatory discharge, is, as I have suggested, to entitle the plaintiff to summary judgment if he shows that after lodging a complaint about discrimination, only he, not otherwise similarly situated employees who did not complain, was subjected to an adverse employment action even though he was performing his job in a satisfactory manner, unless the defendant presents evidence of a noninvidious reason for the adverse action. But most cases in this and other circuits hold that the prima fa-cie case of retaliation under McDonnell Douglas requires more — requires proof of a “causal link” between the protected expression in which the plaintiff engaged (as by filing a complaint about an unlawful act by his employer) and the adverse employment action of which he is complaining. E.g., Miller v. American Family Mutual Ins. Co., 203 F.3d 997, 1007 (7th Cir.2000); Sauzek v. Exxon Coal USA, Inc., 202 F.3d 913, 918 (7th Cir.2000); Byers v. Dallas Morning News, Inc., 209 F.3d 419, 427 (5th Cir.2000); Jones v. Washington Metropolitan Area Transit Authority, 205 F.3d 428, 433 (D.C.Cir.2000).
*476 If this means that the plaintiff must present evidence that his protected expression caused-the defendant to retaliate, then McDonnell Douglas really does regulate just the order of proof and the allocation of burdens of production and has no substantive implications, since the plaintiff would have to prove as much in a retaliation case if McDonnell Douglas had never been heard of. But it would be better then to drop reference to McDonnell Douglas altogether, because cases such as Miller and Sauzek go on to say, very confusingly as it seems to me, that after the plaintiff has made his prima facie case, the defendant can defeat it by producing evidence that the motive for the adverse employment action was not retaliatory, unless the plaintiff is able to come back and show that the alleged nonretaliatory motive was actually pretextual. That is McDonnell Douglas- speak, all right, but it is out of place.' If the plaintiff has produced evidence that he was fired because of his protected expression, he has gone beyond McDonnell Douglas by producing actual evidence of unlawful conduct, and that should be enough to get him to a jury. McDonnell Douglas is designed to give the plaintiff a boost when he has no actual evidence of discrimination (or retaliation) but just some suspicious circumstances. If he can prove retaliation with evidence that his protected expression was in fact the cause of his being fired, he doesn’t need McDonnell Douglas and it gives him nothing.
Further complicating the picture, however, is disagreement — or maybe sheer muddle — as to what “causal link” means. If it means, as I have assumed thus far, and as we held in King v. Preferred Technical Group, 166 F.3d 887, 892 (7th Cir.1999); Johnson v. City of Fort Wayne, 91 F.3d 922, 939 (7th Cir.1996), and Klein v. Trustees of Indiana University, 766 F.2d 275, 280 (7th Cir.1985), that the plaintiff must present evidence that had it not been for his protected expression, he would not have been fired (or suffered whatever other adverse action of which he complains), then the situation is as I have described it: McDonnell Douglas has no proper role to play in a federal retaliation suit. But the Fifth and Eleventh Circuits have held that all that “causal link” signifies in this context is that the protected expression and the adverse action “were not wholly unrelated,” and that this means that something less than proof that the plaintiff wouldn’t have been fired (or suffered other adverse action) had he not engaged in the protected expression — less than proof, that is, of “but for” causation (what philosophers call a necessary condition) — will suffice to complete the prima facie case. Long v. Eastfield College, 88 F.3d 300 , 305 n. 4 (5th Cir.1996); Simmons v. Camden County Board of Education, 757 F.2d 1187, 1189 (11th Cir.1985). And some of our cases accept this definition, e.g., Sauzek v. Exxon Coal USA, Inc., supra, 202 F.3d at 918 ; Hunt-Golliday v. Metropolitan Water Reclamation District, 104 F.3d 1004, 1014 (7th Cir.1997) — yet without attempting to distinguish King, Johnson, and Klein or to explain what “not wholly unrelated” means. In Sauzek and Hunh-Golliday, we actually required (despite our “not wholly unrelated” language) proof of a causal relation, 202 F.3d at 918-19 ; 104 F.3d at 1014-15 , while Johnson , a case that recites the stricter test, accepts a mere coincidence in timing — expressly rejected by Sauzek as sufficient to prove that the protected expression and the adverse employment action were “not wholly unrelated,” 202 F.3d at 919 — -as proof of the required causal link! 91 F.3d at 939 .
Someday we’ll have to decide what the prima facie case of retaliation is in the Seventh Circuit. But whatever it is has no relation to McDonnell Douglas, which is not about the meaning of “causal link.” The Erie issue, however, may turn on that meaning. If the requirement of proving cause is so attenuated as to give the plaintiff a boost toward winning his case that he would not have under ordinary rules of pleading and production, then there is a conflict with substantive state law, and what the federal courts inaptly call the *477 McDonnell Douglas standard for proving retaliation must give way in any retaliation ease governed by state law. If the prima facie case of retaliation under federal law is instead what I am calling the logical adaptation of McDonnell Douglas, requiring no evidence of causation at all, then even more clearly it must give way in a case governed by state law. In either event Hiatt and McEwen would have to be reexamined. Only if the prima facie case of retaliation under federal law requires proof of a causal relation in the usual sense, without attenuation, is there no conflict between state and federal law, at least in Illinois, which requires such proof. Only then would the “McDonnell Douglas” standard, though thoroughly misnamed when interpreted to require proof of causation to make out a prima facie case, properly apply to state retaliation cases litigated in federal courts.