Michael K. DePAEPE, Executor of the Estate of Kenneth J. DePaepe, Plaintiff-Appellee, v. GENERAL MOTORS CORPORATION, Defendant-Appellant

EASTERBROOK, Circuit Judge. An automobile accident in 1984 left Kenneth DePaepe a quadriplegic: his neck broke during the deceleration. DePaepe says that a better design of the auto’s sun visor would have averted the loss; General Motors contends that the car (a 1984 Buick Regal) was crashworthy and that only DePaepe’s failure to use his seatbelt accounts for the injury. This tort suit under the diversity jurisdiction has been tried twice. After a three-week trial, the first jury absolved gm, but we reversed because the judge should not have instructed the jury about the concept of enhanced injury. 33 F.3d 737 (1994). After a seven-week trial, the second jury found in DePaepe’s favor and awarded $12.6 million in damages. DePaepe died while the case was on appeal; the litigation has been carried on by the executor of his estate. We must, alas, reverse a second time. DePaepe’s lawyer told the jury during his opening statement that the visor was defective because it did not comply with a federal motor vehicle safety standard requiring sun visors for the front seats to be “constructed of or covered with energy-absorbing material”. 49 C.F.R. § 571.201 S3.4.1 (“fmvss 201 s3.4.1”). For 7 trial days (during 13 calendar days) the jury heard evidence designed to persuade it that gm violated this regulation. Counsel sought to convince the jury (i) that “energy-absorbing” means “absorbing a lot of energy” rather than “absorbing some energy”, and (ii) that the visor did not dissipate the requisite amount of energy and thus exposed DePaepe’s neck to too much force. gm contested step one of that argument by a motion in limine, reminding the judge that the meaning of a regulation is a question for the court, not the jury. See Bammerlin v. Navistar International Transportation Corp., 30 F.3d 898, 900 (7th Cir.1994); Tyus v. Urban Search Management, 102 F.3d 256, 266 (7th Cir.1996). Yet instead of telling the jury what fmvss 201 s3.4.1 means, the district judge first allowed DePaepe to argue his interpretation of the regulation to the jury for two weeks and then stunned everyone by declaring fmvss 201 s3.4.1 invalid and forbidding either side to mention it again in the jury’s presence. Both sides were flummoxed: DePaepe wanted to use fmvss 201 s3.4.1 to bolster his case, while gm wanted to answer DePaepe’s presentation (in lieu of which it wanted a mistrial) and to argue on its own behalf that the visor’s compliance *718 with fmvss 201 s3.4.1 is a reason to conclude that the design was not defective. Motor vehicle safety standards come in two flavors: design and performance. Design standards specify what a component of a vehicle must be; performance standards specify what the component must accomplish. A requirement that every car have safety belts is a design standard; a requirement that each anchorage of each safety belt withstand a 5,000 pound force (see fmvss 210 s4.2.1) is a performance standard, fmvss 201 s3.4.1 is written as a design standard: A sun visor that is constructed of or covered with energy-absorbing material shall be provided for each front outboard designated seating position. The National Highway Transportation Safety Administration, which issued fmvss 201 s3.4.1, has confirmed that this regulation creates a design standard. 60 Fed.Reg. 60,404 (1995); 61 Fed.Reg. 23,413, 62,934 (1996). gm covered the sun visors in 1984 Buick Regal ears with nylon backed by polyurethane foam. No one doubts that polyurethane foam dissipates energy when compressed. It is therefore an “energy-absorbing material”. There was not very much foam, and so not very much energy was absorbed. The parties debate the prudence of gm’s engineering decision. But as a design standard fmvss 201 s3.4.1 is straightforward—and it was satisfied as a matter of law by the sun visors in the 1984 Buick Regal. The district court should have granted gm’s motion in limine and prevented DePaepe from arguing to the jury that gm failed to comply with fmvss 201 s3.4.1. What led the district court to declare in mid-trial that the regulation is invalid was a belief that a federal rule must do more to promote safety than fmvss 201 s3.4.1 does as written. The only way to make it do more is to treat it as a performance standard, which is essentially how DePaepe treated it before the district judge barred further reference to the subject. But as a performance regulation fmvss 201 s3.4.1 is deficient. It does not answer the critical question: how much energy must the visor absorb? The district judge saw this as a faffing that required him to deem the standard invalid. But it is a shortcoming only if fmvss 201 s3.4.1 is required to answer the question “how much.” It must do so only if it is a performance standard—which it isn’t. The regulation is valid and was complied with. Many regulations accomplish little because they require little, but this does not impugn their validity. Regulations can come in bite sizes as well as jumbo economy sizes. DePaepe argues that these two errors—allowing two weeks worth of evidence and argument that gm failed to comply with fmvss 201 s3.4.1 followed by an enforced quietus on the subject—were harmless. See Fed.R.Civ.P. 61. According to DePaepe, gm was a beneficiary of the second error and can hardly complain that the plaintiff was stopped in his tracks in an effort to elaborate a theory of liability. As for the first, De-Paepe’s lawyer contends that he had introduced the jury to fmvss 201 s3.4.1 and gm’s treatment of that regulation, but was cut off before he could administer the cowp de grace by showing noncompliance. Again gm was the beneficiary of the judge’s timing, the argument concludes. We think this position erroneous both legally and factually. The legal error is DePaepe’s assumption that only a plaintiff is entitled to use a federal safety standard to advantage. Not so. Compliance with a safety standard does not foreclose tort remedies, see 49 U.S.C. § 30103(e), but a manufacturer is entitled to inform a jury that its vehicles pass federal muster and to invite the jury to conclude that a vehicle that complies with all federal rules is safe enough to be on the road. In Illinois, whose law governs this action, compliance with federal safety standards is enough by itself to support a verdict for the manufacturer. Moehle v. Chrysler Motors Corp., 93 Ill.2d 299, 303-04, 66 Ill.Dec. 649, 443 N.E.2d 575, 577 (1982). As for the facts: after a review of the trial transcript, we conclude that much that was said, and more that was implied, during plaintiff’s presentation conveyed the view that gm did not comply with fmvss 201 s3.4.1 and that its certification of compliance was disingenuous. Little purpose would be served by rehearsing the evidence and arguments. Suffice it to say that no plaintiff would take seven trial days to make his *719 opponent’s case before even beginning the work of undermining the defense. When the judge precluded all further reference to fmvss 201 s3.4.1, gm was behind the eight ball and never had an opportunity to present its side of the matter to the jury. Two other issues argued on appeal may recur if the parties are unable to resolve their differences on remand and a third trial must be held. We address them briefly in the hope that the third trial will be the last. 1. DePaepe’s theory of liability has evolved since the suit began. Initially he blamed the car’s brake and occupant restraint systems—the former supposedly contributed to the crash, and the latter aggravated the injury. Later counsel added the windshield header and front-end components to the list of supposedly defective systems. In supplemental interrogatories before the second trial, DePaepe pointed to 17 design defects in almost as many different parts of the car. But by the time the second trial got under way, the list was down to the sun visor, which supposedly trapped DePaepe’s head in an awkward position during the crash, gm wanted to inform the jury about DePaepe’s ambulatory theories, doubtless hoping that the jury would infer that De-Paepe hadn’t found a real villain but was easting about for a motif that might resonate with a lay audience. The district court invoked Fed.R.Evid. 403 to prevent gm from introducing the pleadings, interrogatory answers, and other documents that would have set the stage for this argument. This decision was within the court’s discretion. A court is entitled to keep the jury focused on the claim of liability that requires decision; the judge need not allow the defendant to put the plaintiffs litigation tactics on trial. Inconsistent pleadings are allowable, Fed. R.Civ.P. 8(e)(2), and the use of discovery to winnow or refine theories of liability should not be discouraged. Some things plaintiffs counsel said may have been deceptive—for example, counsel stated to the jury that “it is Kenny DePaepe’s position, it always has been in this trial, his position, that the sun visor broke his neck.” Only the qualifier “in this trial” makes the statement true; the word “always” may have been designed to imply consistency over a longer period. The judge should have corrected this misleading statement with a few words to the jury, but he did not need to allow gm to hijack the trial as it proposed to do. 2. Stephen Syson, a professional expert witness, provided the principal support for DePaepe’s theory that the sun visor was defectively designed, gm contends that Syson offered junk science rather than engineering expertise, see Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993), and that the district court therefore should have excluded his testimony. Appellate review is deferential. General Electric Co. v. Joiner, — U.S.-, 118 S.Ct. 512, 139 L.Ed.2d 508 (1997). gm complains that Syson had not tested or analyzed sun visor designs before this litigation, but a judge does not automatically abuse his discretion in concluding that an expert can offer useful information without having dealt previously with the product at issue in the case. Cummins v. Lyle Industries, 93 F.3d 362, 369 (7th Cir.1996). The question is not whether the expert has hands-on experience but whether his testimony meets scientific standards of hypothesis testing and support by data. Tyus, 102 F.3d at 263; Rosen v. Ciba-Geigy Corp., 78 F.3d 316 (7th Cir.1996); Pries v. Honda Motor Co., 31 F.3d 543 (7th Cir.1994). Personal experience may lead to knowledge, see Roback v. V.I.P. Transportation Inc., 90 F.3d 1207, 1214-16 (7th Cir.1996), or the lack of experience may lead an expert to use inappropriate tests, so a court is entitled to be skeptical about testimony by a witness outside his normal field of practice. See Navarro v. Fuji Heavy Industries, Ltd., 117 F.3d 1027, 1031 (7th Cir.1997); Braun v. Lorillard Inc., 84 F.3d 230, 234-35 (7th Cir.1996). Still, practical experience is not essential to expert testimony and sometimes gets in the way of scientific detachment. Minasian v. Standard Chartered Bank, PLC, 109 F.3d 1212, 1216 (7th Cir.1997). gm’s lawyers lampoon the methods Syson used to test the sun visor and to reach conclusions about the engineering compromises that would optimize a sun visor’s *720 performance in light of the risks involved. But their cri de coeur is not backed up by-references to any body of scientific knowledge. What tests do engineers use to resolve questions of the kind Syson addressed? What tests should he have performed? What data did he overlook? Counsel apparently want appellate judges to make a priori judgments about how scientific inquiry should be conducted. That way quackery lies. A profession resolves questions of method in the same way it reaches conclusions about other empirical issues; which method is best is a question itself subject to scientific inquiry. See Kenneth R. Foster & Peter W. Huber, Judging Science: Scientific Knowledge and the Federal Courts 137-62 (1997). For all this record reveals, Syson performed the tests that gm’s staff engineers or the faculty of mit use to reach conclusions. (Syson was a design engineer at GM earlier in his career.) A litigant that wants a court of appeals to set aside a district judge’s decision to admit expert testimony has to do more than appeal to a lawyer’s sense of how science should be done. That is all GM has done, and we therefore lack any basis on which to upset the district judge’s decision, whether or not that decision was correct as an independent matter. This is not to say that the subject is altogether closed. Some of Syson’s performance at trial was unscientific. For example, he attached a 10-pound weight to a sun visor and let it swing at the end of a pendulum, to see how much damage it would do to a concrete block 1% inch thick if dropped from a height of 3 feet. Visors more flexible than the one in the 1984 Buick Regal did less damage. But what did this prove about the proper design of automobile visors? Syson called his experiment “interesting, but not scientific”. Indeed so. gm is free to renew its objection to this aspect of Syson’s testimony, and the district judge should evaluate the scientific provenance and rigor of each of Syson’s tests independently using the principles we have mentioned. On one issue we are confident that the district judge erred by allowing Syson to testify, and repetition should not be permitted at a retrial. DePaepe’s lawyer asked Syson why gm had reduced the amount of padding in its sun visors. He replied that gm had done so to save money, (gm’s position is that it cut the padding because excessively thick sun visors pose hazards of their own.) The district court overruled gm’s objection to Syson’s testimony about motive or purpose, remarking that “as an expert, he can speculate.” With all respect to the district court, the whole point of Daubert is that experts can’t “speculate.” They need analytically sound bases for their opinions. District courts must be careful to keep experts within their proper scope, lest apparently scientific testimony carry more weight with the jury than it deserves. Syson lacked any scientific basis for an opinion about the motives of gm’s designers. He could give an opinion as an engineer that reducing the padding saved a particular amount of money; he might testify as an engineer that gm’s explanation for the decision was not sound (from which the jury might infer that money was the real reason); but he could not testify as an expert that gm had a particular motive. Because Syson did not participate in the deliberations leading to the design of the sun visor, he could not testify as a fact witness on the subject, either. The judgment of the district court is vacated, and the case is remanded for further proceedings consistent with this opinion and our 1994 opinion. This time, Circuit Rule 36 will apply on remand.