Jennie Mayer parked her car in the lot at the Miller Mall in Lake County, Indiana, and bought some articles at K-Mart, one of the mall's seven stores. Returning to the car, Mayer was the victim of a purse snatching, during which the three thieves hit her on the head. She fell to the ground and suffered a fractured pelvis and other injuries. A jury in this diversity suit ordered K-Mart and Gary Partners, the owners of the mall, to pay Mayer $260,000 on the theory that they negligently omitted precautions that would have protected their customers from crime.

Defendants contend that the evidence does not support the verdict, that the damages are excessive, and that errors in the jury instructions are not harmless. Although state law supplies the substantive rules that govern, Erie R.R. v. Tompkins, 304 U.S. 64, 58 S. Ct. 817, 82 L. Ed. 1188 (1938), each question has a procedural dimension. Who decides whether the evidence is sufficient, the damages excessive, or the errors harmless? A judge, obviously, but under what standards? Once the judge decides, what is the standard of appellate review? We know from Donovan v. Penn Shipping Co., 429 U.S. 648, 649, 97 S. Ct. 835, 836, 51 L. Ed. 2d 112 (1977), that because the seventh amendment and federal rules make the roles of judge and jury a subject of federal law, "[t]he proper role of the trial and appellate courts in the federal system in reviewing the size of jury verdicts [also] is ... a matter of federal law." Accord, Browning-Ferris Industries v. Kelco Disposal, Inc., 492 U.S. 257, 276-80, 109 S. Ct. 2909, 2920-22, 106 L. Ed. 2d 219 (1989). Similarly, we know from cases such as Patel v. Gayes, 984 F.2d 214 (7th Cir.1993); Simmons, Inc. v. Pinkerton's, Inc., 762 F.2d 591, 595 (7th Cir.1985); and Platis v. Stockwell, 630 F.2d 1202, 1206 (7th Cir.1980), that federal law supplies the method of proposing, arguing about, objecting to, and evaluating the probable effect of errors in, jury instructions.

Yet cases in this circuit routinely announce that state law supplies the standard of appellate review of the sufficiency of the evidence if there has been a jury trial. E.g., Dolder v. Martinton, 998 F.2d 499, 501 (7th Cir.1993); Amplicon Leasing v. Coachmen Industries, Inc., 910 F.2d 468, 470 (7th Cir.1990). At the same time, federal law provides the standard of appellate review if there has been a bench trial. Anderson v. Bessemer City, 470 U.S. 564, 105 S. Ct. 1504, 84 L. Ed. 2d 518 (1985); Fed.R.Civ.P. 52(a). Thirty years ago we were confident that federal law governed both questions. Gudgel v. Southern Shippers, Inc., 387 F.2d 723, 725 (7th Cir.1967); Reitan v. Travelers Indemnity Co., 267 F.2d 66, 69 (7th Cir.1959). Opinions that call for the application of state law do not cite the earlier cases specifying a federal standard--and neither set of cases gives reasons. Eleven years ago we noted the contradiction, see Abernathy v. Superior Hardwoods, Inc., 704 F.2d 963, 970-71 (7th Cir.1983), but left matters as we found them. Recently a rumble of discontent with this state of affairs has been heard in our opinions. E.g., Jean v. Dugan, 20 F.3d 255, 263 (7th Cir.1994); Sokol Crystal Products, Inc. v. DSC Communications Corp., 15 F.3d 1427, 1431-32 (7th Cir.1994). The time has come, we believe, to put the matter to rest. Sometimes it is said that a particular state's law is the same as federal law, so that the choice is irrelevant. But Indiana's standard, whether the evidence "points unerringly" to one conclusion, see Huff v. Travelers' Indemnity Co., 266 Ind. 414, 363 N.E.2d 985, 990 (1977), may differ subtly from the federal reasonable-person standard that we discuss below. Ours is a close case, one in which the choice of standard potentially matters, and we therefore want to be sure to get the standard right.

Erie holds that under the Rules of Decision Act, 28 U.S.C. Sec. 1652, state law (including common law) governs the conduct of cases under the diversity jurisdiction unless federal law speaks to the question. The "unless" clause permits the federal forum to apply rules of procedure found in federal statutes (and, equivalently, rules promulgated under the Rules Enabling Act, 28 U.S.C. Secs. 2071-77), even though these procedures may affect the outcome. See Stewart Organization, Inc. v. Ricoh Corp., 487 U.S. 22, 108 S. Ct. 2239, 101 L. Ed. 2d 22 (1988); Burlington Northern R.R. v. Woods, 480 U.S. 1, 107 S. Ct. 967, 94 L. Ed. 2d 1 (1987); Walker v. Armco Steel Corp., 446 U.S. 740, 100 S. Ct. 1978, 64 L. Ed. 2d 659 (1980); Hanna v. Plumer, 380 U.S. 460, 85 S. Ct. 1136, 14 L. Ed. 2d 8 (1965); cf. Barron v. Ford Motor Co., 965 F.2d 195 (7th Cir.1992) (distinguishing between substance and procedure under the Federal Rules of Evidence). The burden of persuasion is tied to the definition of the right, so state law determines whether the plaintiff must prove the case by a preponderance, by clear and convincing evidence, or by some other standard. Stoner v. New York Life Insurance Co., 311 U.S. 464, 61 S. Ct. 336, 85 L. Ed. 284 (1940); see also American Dredging Co. v. Miller, --- U.S. ----, ---- - ----, 114 S. Ct. 981, 988-89, 127 L. Ed. 2d 285 (1994). But who decides an issue is a procedural question. Byrd v. Blue Ridge Rural Electric Cooperative, Inc., 356 U.S. 525, 78 S. Ct. 893, 2 L. Ed. 2d 953 (1958). Whether the trier of fact is a jury, a judge, or a magistrate judge (one presided in this case by consent under 28 U.S.C. Sec. 636(c)), is a subject for the forum's own law.

In some states juries are entitled to decide all questions of fact and law, yet this does not compel a federal court to disregard provisions for summary judgment (Fed.R.Civ.P. 56) and verdict as a matter of law (Fed.R.Civ.P. 50), sending every case to a jury. Herron v. Southern Pacific Co., 283 U.S. 91, 51 S. Ct. 383, 75 L. Ed. 857 (1931). Similarly, although some states curtail judges' review of juries' awards of damages, see Honda Motor Co. v. Oberg, --- U.S. ----, 114 S. Ct. 2331, 129 L.Ed.2d ---- (1994), in federal court judges assess the reasonableness of awards. Donovan; Browning-Ferris. The states are not bound by the seventh amendment and therefore may dispense with civil juries altogether; a federal court in the same state would remain obliged to submit cases to juries whenever courts sitting at common law in 1791 would have done so. Sometimes, however, the federal government is stingier with jury trials than are states: a suit against the United States under the Federal Tort Claims Act uses state substantive law, but there will be no jury, even though state courts would use juries to resolve similar disputes. Thus there are bound to be differences between state and federal practice: some states permit juries to resolve questions that are settled by judges in federal court, some states commit to judges questions that the seventh amendment sends to juries in federal litigation, and under the FTCA judges will decide questions under state law that states would allow juries to decide. Federal juries are drawn from wider districts than are state juries, panels serve longer, jurors are compensated differently, and the procedure for voir dire and challenges differs from that used in state courts. These many distinctions, which must on occasion affect the outcome of litigation, imply that each system of courts should be thoroughgoing in using its own rules to determine which questions go to juries and which to judges--and what is the proper relation between trial and appellate judges. Absorbing bits and pieces of some other procedural system cannot eliminate effects on the outcome, but it can cause confusion and uncertainty in a federal system with more than 50 distinct jurisdictions. A court has a hard enough time keeping track of one set of procedural rules.

From beginning to end, diversity litigation is conducted under federal rules of procedure. The filing and content of the complaint, motions for judgment on the pleadings, the introduction of evidence, these and other elements of federal practice suffuse the proceedings. Federal rules dominate even when the subject is evaluation of the sufficiency of the evidence.

. Summary judgment is granted in diversity cases when the non-moving party lacks enough evidence to sustain a jury verdict according to the federal standard: whether reasonable minds could deem the evidence adequate under the governing substantive rule. Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 249-51, 106 S. Ct. 2505, 2510-11, 91 L. Ed. 2d 202 (1986). See also Celotex Corp. v. Catrett, 477 U.S. 317, 106 S. Ct. 2548, 91 L. Ed. 2d 265 (1986); McEwen v. Delta Air Lines, Inc., 919 F.2d 58, 60 (7th Cir.1990) ("Erie does not require a federal court to employ the state's rules on the allocation of issues between judge and jury.... Federal courts may grant summary judgment under Rule 56 on concluding that no reasonable jury could return a verdict for the party opposing the motion, even if the state would require the judge to submit an identical case to the jury.").

. When federal judges act as triers of fact in diversity cases, all questions concerning the standard of appellate review are governed by federal law. This means principally the "clearly erroneous" standard specified by Fed.R.Civ.P. 52(a), but it includes related questions as well. See Steven Alan Childress, Judicial Review and Diversity Jurisdiction: Solving an Irrepressible Erie Mystery?, 47 SMU L.Rev. 271, 281-82 (1994) (collecting authority).

. When acting on motions for new trials in diversity cases, district judges use the same approach they apply in cases arising under federal law. Blumenfeld v. Stuppi, 921 F.2d 116, 118 (7th Cir.1990); Pincus v. Pabst Brewing Co., 893 F.2d 1544, 1549 (7th Cir.1990). States may prohibit discretionary awards of new trials or use their own standards, without effect on the conduct of federal litigation.

. When deciding whether verdicts in diversity cases are excessive--and whether the remedies of remittitur and additur are available in the first place--federal courts use their own rules. Donovan; Browning-Ferris; American Business Interiors, Inc. v. Haworth, Inc., 798 F.2d 1135, 1146 (8th Cir.1986).

If evaluating the sufficiency of the evidence before the trial begins occurs under federal standards, as Anderson, Celotex, and McEwen hold, it is more than a little strange to say that evaluating the sufficiency of the evidence once trial gets under way (for purposes of judgment as a matter of law under Fed.R.Civ.P. 50) is within the domain of state standards--especially because under Anderson and Celotex the pre-trial, mid-trial, and post-trial standards are supposed to be identical! Yet mid-trial and post-trial the state standard governs, at least in this circuit. In addition to Dolder and Amplicon, see, e.g., Eastman v. Chicago, Central & Pacific R.R., 930 F.2d 1173, 1176-77 (7th Cir.1991); Luson International Distributors, Inc. v. Mitchell, 939 F.2d 493, 495 (7th Cir.1991). But see Bose Corp. v. Consumers Union of United States, Inc., 466 U.S. 485, 498-501, 104 S. Ct. 1949, 1958-1959, 80 L. Ed. 2d 502 (1984). Things look even stranger when we realize that, despite resort to state law for mid-trial and post-trial rulings on the sufficiency of the evidence on the merits, we use federal law under Donovan and Browning-Ferris to evaluate the sufficiency of the evidence on damages, and federal law to evaluate the propriety of setting aside the first jury's decision in favor of another decision at a second trial.

Although the Supreme Court has twice ducked the question whether the federal reasonable-person standard applies to the mid-trial and post-trial evaluation of evidence in diversity cases, Mercer v. Theriot, 377 U.S. 152, 84 S. Ct. 1157, 12 L. Ed. 2d 206 (1964); Dick v. New York Life Insurance Co., 359 U.S. 437, 444-45, 79 S. Ct. 921, 926, 3 L. Ed. 2d 935 (1959), courts of appeals other than this one use the federal standard. See, e.g., Wratchford v. S.J. Groves & Sons Co., 405 F.2d 1061, 1064-66 (4th Cir.1969); Planters Manufacturing Co. v. Protection Mutual Insurance Co., 380 F.2d 869, 870-71 (5th Cir.1967). Although at one time or another the sixth, eighth, ninth, and tenth circuits have applied the state standard, all (with the potential exception of the sixth) have changed their minds. Professor Childress summarizes the circuits' current positions. 47 SMU L.Rev. at 295-308. Scholars likewise believe that the use of federal norms flows directly from the fact that the division of tasks between judge and jury in federal court is a subject of national law. E.g., Charles Alan Wright & Arthur R. Miller, 9 Federal Practice and Procedure Sec. 2525, 19 Id. Sec. 4511 at 177-78; Charles Alan Wright, The Law of Federal Courts 607-09 (4th ed. 1983); 5A Moore's Federal Practice p 50.06 (1992 rev.); Edward H. Cooper, Directions for Directed Verdicts: A Compass for Federal Courts, 55 Minn.L.Rev. 903, 972-89 (1971); Childress, 47 SMU L.Rev. at 308-27.

The attraction of the contrary position depends on a confusion between the who and the how of litigation. Take a jurisdiction in which even a scintilla of evidence favoring the plaintiff requires submission of the case to a jury. The judge does not instruct this jury that a scintilla is enough for the plaintiff to prevail; instead the judge most likely will tell the jury to decide according to the preponderance of the evidence. The "scintilla of evidence" standard determines who resolves the factual dispute, and the "preponderance" standard tells that body how to evaluate the evidence presented. In acting on post-trial motions, and on appeal, the state judiciary will apply the "scintilla of evidence" standard in order to respect the decision committing to the jury the ultimate evaluation of the evidence. Erie commands a federal court to respect the state's decision not only on the elements of a claim for relief but also on how evidence should be evaluated (by a preponderance or by some other standard). But, as we have stressed, nothing in Erie or the Rules of Civil Procedure commands a federal court to accept the state's views about who should evaluate the evidence. Thus it is entirely consistent to say that although state law defines the elements of a claim and the burden of persuasion, federal law defines the standard for evaluating the sufficiency of the evidence. If reasonable persons could not find that the evidence justifies a decision for a party on each essential element, the court should grant judgment as a matter of law--before trial under Rule 56, later under Rule 50, and using the same federal standard each time. By linking the standard for summary judgment to the standard for overturning a verdict, Anderson and Celotex leave no other option. We now adopt the federal reasonable-person standard across the board: pre-trial, mid-trial, post-trial, and on appeal, for evaluating both the merits and the quantum of relief. Contrary decisions are no longer authoritative.d

The parties agree that Indiana requires businesses to take reasonable precautions to protect customers from foreseeable aggression on their premises--that a business may not rely on the police plus the deterrent force of the criminal law, but must on occasion supplement public precautions against crime. State law on this subject is skimpy, but the defendants agreed to instructions stating this principle, which is conclusive at least for this case. Rakovich v. Wade, 850 F.2d 1180, 1192 (7th Cir.1988) (en banc); Will v. Comprehensive Accounting Corp., 776 F.2d 665, 675 (7th Cir.1985). Defendants do not argue that the evidence would be insufficient under a properly reformulated legal standard. See Boyle v. United Technologies Corp., 487 U.S. 500, 513-14, 108 S. Ct. 2510, 2519, 101 L. Ed. 2d 442 (1988). (Boyle, which we have read for the proposition that even in a diversity case "a party need not object to jury instructions to urge a j.n.o.v. based on different standards", Hystro Products, Inc. v. MNP Corp., 18 F.3d 1384, 1392 n. 6 (7th Cir.1994); see also Geldermann, Inc. v. Financial Management Consultants, Inc., 27 F.3d 307, 312-313 (7th Cir.1994); illustrates the application of distinctly federal standards of jury control.) The parties' dispute concentrates on the "foreseeability" requirement. Defendants conceded that there had been one other purse snatching from a customer in the parking lot. But it is a large lot, and one crime over an extended period did not require the mall to hire a force of armed guards, defendants submitted. The mall hired a guard, and defendants contend that such a precaution suffices, given the low risk implied by the single prior crime.

If one earlier crime were all Mayer had to go on, we would agree with defendants that reasonable jurors could not find the risk of harm great enough to make additional precautions mandatory. But Mayer sought and obtained a missing-evidence instruction, one we think proper under Indiana law. Every day the guard filled out a report and turned it in to K-Mart. According to K-Mart, these reports were used only to determine whether the guard performed, and should be paid for, a full day's work; treating the forms as payroll records, K-Mart discarded them after 30 days. A reasonable juror could conclude, however, that these forms recorded crimes and other incidents that occurred at the mall. By recording what the guard did, in addition to the fact of his presence at the mall, the forms were potentially useful to show the level of criminal activity, and thus the risk to customers. Mayer asked the jury to infer from the destruction of these records that the mall was hiding knowledge of additional crimes in order to protect itself from tort liability. Managers at K-Mart denied that they had this purpose, but jurors were at liberty to disbelieve them.

What the jury knew, then, is this: the mall is located in a run-down area; the businesses were sufficiently concerned that they hired a guard; the defendants' managers remember two purse snatchings (one from a customer, one from an employee) during the last five years and do not deny that there were other crimes ("occasional car theft attempts", according to an answer to an interrogatory); the possibility of additional crimes can be neither verified nor refuted; records that would have assisted in the process have been destroyed. We think this is enough--barely--to permit reasonable jurors to believe that additional criminal episodes were foreseeable, which is to say that they were sufficiently likely that prudent persons would take precautions against them. The sole precaution taken, the hiring of one guard, not only did not prevent this crime but was unlikely to do so, reasonable jurors could find, because the guard devoted the bulk of his energies to protecting employees (rather than customers) and keeping traffic lanes free. A retired police officer testified that it takes a visible presence to dissuade muggers and other offenders, and that the mall's guard was not visible. We conclude that the evidence supports the verdict on liability. Mayer fell during the attack, suffering a broken pelvis and a concussion. Less than a month later she suffered a stroke, which two physicians linked to these injuries. Today she has restricted mobility and persistent pain. Reasonable jurors could assess damages at $260,000.

A few challenges to the jury instructions remain. Most of defendants' arguments are quibbles. For example, the magistrate judge did not give their proposed instruction that businesses are not "insurers" of customers' safety, but that point was made in other instructions, which directed the jury through the usual steps of a negligence claim. Only one of these instructional disputes calls for comment. Although defendants accept the premise that Indiana requires businesses to take reasonable precautions against crime, see Welch v. Railroad Crossing, 488 N.E.2d 383 (Ind.App.1986), they asked the judge to instruct the jury that they are not liable if the victim's injury has an intervening cause--here, the purse snatching and battery. An intervening cause usually relieves a negligent person of liability. But when state law creates a duty to protect customers from crime, the crime itself cannot be an "intervening cause"; if it were, the state would be taking with the left hand what it gives with the right. When a state creates a duty to protect people from third parties' intentional conduct, the fact that a third party intentionally caused the injury is a reason to impose liability rather than to avoid it. Defendants have not located a case from Indiana (or any other state) treating a criminal episode as an "intervening cause" when the duty in question calls for precautions against crime.

AFFIRMED.