dissenting.
I respectfully cannot accept all of the majority’s analysis. In my view, the district court improperly rejected Pacheco’s testimony about the need for a door on the rider. Accordingly, I would reverse and remand for a new trial.
I agree with the majority’s analysis about the district court’s exclusion of certain accident reports. The district court has broad discretion to decide which accident reports are relevant. “Even when substantial identity of the circumstances is proven, the admissibility of such evidence lies within the discretion of the trial judge.” Nachtsheim v. Beech Aircraft Co., 847 F.2d 1261, 1269 (7th Cir.1988). As the majority correctly notes, the reports admitted were sufficient for Ms. Weir to make her point.
The district court’s decision to exclude Daniel Pacheco’s proposed testimony about the need for a door on the rider cannot, however, be sustained even under the deferential review that we accord to trial court determinations about the admissibility of evidence.1 The court excluded that part of Pacheco’s testimony because it found that it was inadmissible under Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 592-93, 113 S.Ct. 2786, 125 *462L.Ed.2d 469 (1993). In examining the correctness of that determination, we must first place the district court’s ruling in context and then examine the criteria of Daubert.
A.
As the majority observes, Indiana courts recognize the doctrine of crashworthiness. See Indiana Code Ann. § 33-1-1.5-1 et seq. (West 1996) (product liability); Miller v. Todd, 551 N.E.2d 1139, 1142 (Ind.1990); see also Whitted v. General Motors Corp., 58 F.3d 1200, 1205-06 (7th Cir.1995). The Indiana Supreme Court has summarized the doctrine as follows:
Any design defect not causing the accident would not subject the manufacturer to liability for the entire damage, but the manufacturer should be liable for that portion of the damage or injury caused by the defective design over and above the damage or injury that probably would have occurred as a result of the impact or collision absent the defective design.
Miller, 551 N.E.2d at 1142 (citation omitted). Thus, even if Ms. Weir’s cross-over theory was rejected by the jury, Ms. Weir could recover from Crown if she succeeded in showing that the rider was not crash-worthy.
In a crashworthiness case, the plaintiff must show that the injury caused by the product was “a natural and probable consequence which was, or should have been, reasonably foreseen or anticipated in light of the attendant circumstances.” Marshall v. Clark Equip. Co., 680 N.E.2d 1102, 1108 (Ind.Ct.App.1997). Ms. Weir could have recovered from Crown if Crown should have anticipated that the absence of a door on the rider would lead to the type of injury she suffered. Crown may defend, however, by showing that Ms. Weir misused the product in an unforeseeable manner. See id. Thus, even if Crown is correct that Ms. Weir intentionally had placed her foot outside the rider, the critical inquiry for purposes of determining whether the rider is crashworthy is whether Ms. Weir’s misuse was reasonably foreseeable.
The Indiana Court of Appeals’ decision in Marshall is instructive. In that case, the plaintiff was injured while backing up a forklift. See 680 N.E.2d at 1103-04. The plaintiff argued that the forklift’s design was defective because it did not include a rear door. See id. at 1108. The defendants presented a misuse defense, arguing that the plaintiff had improperly operated the forklift with his foot outside the driver’s compartment. See id. The court found that the jury was required to determine “whether Marshall’s misuse of the forklift was reasonably foreseeable.” Id. “The foreseeability of an intervening misuse is usually a question for the jury.” Underly v. Advance Mach. Co., 605 N.E.2d 1186, 1189 (Ind.Ct.App.1993); see also Montgomery Ward & Co. v. Gregg, 554 N.E.2d 1145, 1156 (Ind.Ct.App. 1990).
The district court acknowledged Ms. Weir’s crashworthiness theory in a pretrial order on Crown’s motion in limine to exclude portions of Pacheco’s testimony. The court stated, “‘Design defect’ cases frequently arise in the context of an enhanced injury to the plaintiff caused by the absence of an alleged safety feature or alternative design that would have prevented the injury.” R.208 at 2. The court later connected the doctrine to Pacheco’s proffered testimony, stating that the lack of a barrier over the doorway “falls in the category of ‘injury enhancing’ defects.” Id. at 4. It then noted that plaintiffs in an enhanced injury case must demonstrate, through a “risk-utility” test, that there was a more cost effective alternative design available. Id.
At trial the district court made plain that it did not think Ms. Weir could proceed with her crashworthiness claim:
Let me just mention this. I thought I made it fairly clear that the crash worthiness circumstance, or any problem with the crash worthiness was not really what we were going to do. Any part of *463the design that you would have thought aggravated the injury was out because you have got to build something, you have to test it, you have to go through all the activities to be sure that your theory is correct. I think that is clear from the case law. I’m not making this stuff up. So I don’t anticipate entertaining evidence about that.
R.213 at 344^45. It appears from this statement, read in isolation, that the district court’s basis for dismissing Ms. Weir’s crashworthiness claim was that she had not constructed an alternate model. Such a ruling surely would have been inappropriate on this record. When the alternate model discussed by an expert is in regular commercial production, the expert cannot be faulted for not building his own prototype, but instead is allowed to evaluate data relating to the existing models.2 After trial, however, the district court explicitly stated that it had not based its ruling at trial on the fact that Ms. Weir’s experts had not constructed a model of their own. Indeed, when Ms. Weir moved for a new trial, the district court explained that its basis for dismissing the crashwor-thiness claim was the fact that Ms. Weir had offered no cost/benefit analysis to demonstrate that a door should have been added to the rider.
This circuit has acknowledged that, in order to demonstrate a defect under Indiana law, a plaintiff must perform a cost/benefit analysis. See Pries v. Honda Motor Co., 31 F.3d 543, 545 (7th Cir.1994) (“To demonstrate a defect, the plaintiff must compare the costs and benefits of alternative designs.”). Pacheco presented such an analysis in his offer of proof. He testified that the cost of the doors would be only a small percentage of the total cost of the forklift. He stated that accidents involving riders caused lower limb injuries.3 He testified further that he had read numerous reports of crashes involving riders with doors, and in none of them had the operator suffered an injury to their lower extremities. Pacheco compared the costs of riders with and without doors, and also compared the accident benefits of each. In fact, counsel specifically asked Pacheco if the benefits of doors on riders outweighed their costs, and Pacheco answered yes.
Because of this proffered testimony, it was incorrect, unless the requirements of Daubert were not met, for the district court to conclude that the plaintiff had offered no cost/benefit analysis. We have stated that plaintiffs must offer a more cost-effective design, and Ms. Weir has fulfilled that obligation. See Anderson v. P.A. Radocy & Sons, Inc., 67 F.3d 619, 625 n. 5 (7th Cir.1995). Crown argues that Pacheco is wrong, and that riders with doors were not more cost-effective than those without; indeed, another of Ms. Weir’s own experts, Dr. William Ovens, testified that adding a door would create numerous problems with the rider. However, there was evidence in the record from which a reasonable juror could have concluded that a rider with a door was more cost-effective than one without. It was for the trier of fact to evaluate this testimony.
B.
Having determined that Ms. Wen-should have been allowed to present her *464crashworthiness argument to the jury, I now consider whether Pacheco’s testimony should have been admitted to support that argument. I focus here only on that portion of Pacheco’s testimony in which he testified that the addition of a door to the rider would have helped prevent Ms. Weir’s injury.4
To be admissible as expert testimony, Pacheco’s testimony must be based upon valid scientific knowledge that would assist the trier of fact with a matter at issue. See Daubert, 509 U.S. at 592-93, 118 S.Ct. 2786; Walker v. Soo Line R.R. Co., 208 F.3d 581, 586 (7th Cir.2000). Although Pacheco’s qualifications are not at issue in this case, a district court may reject a proffered expert as unqualified. See United States v. Vitek Supply Corp., 144 F.3d 476, 486 (7th Cir.1998). Even if an expert is qualified, he need not be allowed to testify on all subjects. This court has warned about the dangers of allowing qualified experts to offer opinions that do not rely on proper methodologies and are therefore speculative. See Cummins v. Lyle Indus., 93 F.3d 362, 368 (7th Cir.1996); Rosen v. Ciba-Geigy Corp., 78 F.3d 316, 318 (7th Cir.1996). Further, a district court may conclude that the testimony would not assist the trier of fact with a matter at issue, essentially a relevancy determination. See United States v. Shay, 57 F.3d 126, 132-33 (1st Cir.1996).
1.
As an initial matter, it is not clear that the district court ever applied the Daubert standards to this portion of Pacheco’s proffered testimony. This court has reversed district court decisions that do not show a proper consideration of the Daubert factors. See United States v. Hall, 165 F.3d 1095, 1102 (7th Cir.1999) (describing earlier proceedings in that case). Here, the court properly evaluated another portion of Pacheco’s proffered testimony, his proposed self-designed alternate design. The court’s order does not, however, discuss the subject of Pacheco’s study of accident reports involving forklifts with compartment doors. There is no point in the record at which the district court specifically considers the scientific value of this proffered testimony.
Of course, a district court’s ruling is not defective simply because it failed to recite the Daubert standards. See Walker, 208 F.3d at 590. Our focus must be on whether the district court has applied the principles of Daubert to the proffered testimony at issue. See Walker, 208 F.3d at 590; Hall, 165 F.3d at 1102. Here, the district court, although properly rejecting other parts of Pacheco’s testimony, simply did not discuss specifically the proffered testimony at issue.
2.
In the absence of any explanation from the district court, its ruling is unknown. However, assuming that it considered under Daubert this portion of Pacheco’s proposed testimony, it would have determined whether Pacheco’s methodology was scientifically valid and whether that testimony would have assisted the trier of fact with a matter at issue. In the following discussion, therefore, I shall assume, arguendo, that the district court would have concluded that Pacheco’s testimony failed both prongs of the Daubert analysis. Neither conclusion can be sustained even under our deferential review of such questions.
In the district court’s evaluation of the admissibility of expert testimony, “the focus ... must be solely on the principles and methodology, not on the conclusions they generate.” Cummins, 93 F.3d at 370 (quoting Daubert, 509 U.S. at 595, 113 S.Ct. 2786). In his offer of proof, Pacheco testified that he has studied the protection afforded by doors that previously have been available on forklifts. He claimed to have reviewed accident reports of forklifts involving doors and found no such accidents that caused an injury to the lower *465extremities. Further, from those records he determined that injuries were frequent on riders without doors.
Pacheco relied on data from accident reports to determine that more injuries occurred in crashes involving riders without doors than in crashes involving riders with doors. Review of reports and records is an appropriate method for experts to learn the data about which they plan to testify. See Walker, 208 F.3d at 591. For testimony based on reports to be admissible, the reports themselves must be reliable sources of information. Federal Rule of Evidence 703 demands that experts obtaining data from reports use only reports reasonably relied on by experts in the field. See Fed.R.Evid. 703; United States v. Gardner, 211 F.3d 1049, 1053-54 (7th Cir.2000). There does not appear to be any dispute that these accident reports are reasonably relied upon for the purpose of obtaining data about rider accidents. First, most of the reports were prepared by Crown itself for the purpose of compiling safety histories of its products.5 Second, when considering Ms. Weir’s crossover theory, the district court admitted numerous accident reports into evidence, as discussed in the majority opinion. Although Crown objected that some of these accident reports were irrelevant to the cross-over theory, it has not argued that these reports were unreliable. These reports fulfill the mandate of Rule 703 as materials reasonably relied upon by experts in the field.
Crown focuses on the fact that Pacheco did not cite any articles supporting his position or address contrary authority stating that doors are unnecessary on forklifts. In a Daubert analysis, it helps to have cited articles; it is not, however, required. The fact that professional organizations disagree with Pacheco about the need for doors — another point raised by Crown — is a proper subject for cross-examination, and Crown could have countered Pacheco’s testimony with experts of its own. That concern goes to the weight of Pacheco’s testimony, not its admissibility-
3.
Even though Pacheco’s testimony was based on appropriate scientific methodology, the district court properly could have rejected it if it would not assist the trier of fact with a matter at issue in the case. As discussed above, the district court erred in concluding that the crashworthiness of the rider was not a matter at issue in the case. However, had the district court allowed the crashworthiness theory to proceed, the existence of a more cost-effective and crashworthy alternate design would have been a matter at issue. Establishing the viability of an alternate design is necessary to show proximate cause. See Marshall, 680 N.E.2d at 1108. In this regard, Pacheco’s testimony would have helped Ms. Weir to 'show that a more cost-effective alternative design would have been more crashworthy: he testified that a forklift with a door could have prevented Ms. Weir’s injury and that it would have only taken a minimal expense to improve dramatically the rider’s safety.6 Had the jury been allowed to consider the crashworthiness issue, Pacheco’s testimony would have assisted the jury to decide whether a rider with a door would have been more cost-effective and more crashworthy.
The admissibility of Pacheco’s testimony is not affected by the fact that he did not testify about the foreseeability of any misuse by Ms. Weir. In a crashworthiness case, unforeseeable misuse is an affirmative defense. See Marshall, 680 N.E.2d at 1108; Montgomery Ward, 554 N.E.2d at *4661151-52. The existence of a superior alternate design was an essential element of Ms. Weir’s claim; when the district court prevented her from pursuing that argument, it eliminated the need for Crown to raise any affirmative defenses. Further, had Crown raised the defense of unforeseeable misuse, it would have had the burden of proving that Ms. Weir’s misuse was not foreseeable. When an affirmative defense is raised the burden is on the defendant to establish its elements. See Schleibawm v. Kmart Corp., 153 F.3d 496, 501 (7th Cir.1998); GeP-N-Go, Inc. v. Markins, 544 N.E.2d 484, 486 (Ind.1989). Because the district court never allowed the unforeseeable misuse defense to be raised, and because, even if the defense had been raised, Ms. Weir would have been under no obligation to enter testimony on the subject, the fact that Pacheco did not discuss the foreseeability of any misuse does not affect our consideration of whether Pacheco’s testimony would have assisted the trier of fact 'with a matter at issue.
Crown argues that Pacheco’s testimony was properly rejected because he argued that, even with a door, the rider would have been defective because it did not include a ridge between foot pedals. The testimony about a ridge, however, was directed toward showing how Crown could have prevented a brake failure; his testimony about compartment doors was directed toward showing how Crown could minimize the potential for injury in the event of a brake failure. Whether Pacheco thought that the rider would still be defective if a door was added was immaterial; what was important was his scientifically-grounded testimony that the addition of a door would reduce the chance of injury to lower extremities in the event of a crash.
The majority’s conclusion that Pacheco’s testimony was properly excluded rests in part on the fact that the district court was attempting to minimize confusion over Ms. Weir’s contradictory statements about the position of her feet. It is true that Ms. Weir’s own testimony on the subject of the position of her feet at the time of the crash was inconsistent. However, even if we indulge in the assumption most favorable to Crown — that Ms. Weir intentionally placed her foot outside the running lines of the rider — Pacheco’s testimony about the need for a door on the rider was still relevant because a jury could have concluded that Crown’s design caused an enhancement to Ms. Weir’s injury and awarded her partial recovery under Indiana’s crashworthiness doctrine. This would not be “an unjustified sympathy verdict,” as the majority characterizes it, but would instead be exactly the sort of recovery the crashworthiness doctrine was intended to facilitate. Therefore, this design defect was not irrelevant to Ms. Weir’s injury.
Pacheco’s opinion on the need for doors on the forklift was grounded in proper research methodologies and should have been admitted. From this testimony, a reasonable jury could have inferred that, even if the forklift’s brakes failed, Ms. Weir’s injury could have been prevented by Crown. I would remand the case for a new trial at which this portion of Pacheco’s testimony would be admitted.
*467APPENDIX
[[Image here]]
. Pacheco also at one point offered testimony about an elaborate design change, and on appeal Ms. Weir concedes that the district court correctly excluded that testimony.
. In design defect cases, courts have frequently noted the testimony of experts comparing allegedly defective products to safer designs already in existence. See Chaullc v. Volkswagen of America, 808 F.2d 639, 642-43 (7th Cir.1986); accord Violette v. Smith & Nephew Dyonics, Inc., 62 F.3d 8, 13 (1st Cir.1995); Miles v. Olin Corp., 922 F.2d 1221, 1227 (5th Cir.1991); Johnson v. Colt Indus. Operating Corp., 797 F.2d 1530, 1535 (10th Cir.1986); Martin v. Michelin N. Am., 92 F.Supp.2d 745, 752 (E.D.Tenn.2000) (memorandum); Bush v. Michelin Tire Corp., 963 F.Supp. 1436, 1446 (W.D.Ky.1996) (memorandum).
. Pacheco was allowed to testify at trial about the dangers of riding forklifts. Pacheco testified that forklift accidents occur at "a fairly high frequency,” and that when one occurs there is "a very great chance that the injury will be very serious.” R.218 at 613.
. See note 1, supra.
. Some of the accident reports relied upon by Pacheco in evaluating injuries on forklifts with doors came from K-Mart, a Crown customer.
. Pacheco’s testimony about the need for a door was relevant only to a crashworthiness claim; there is no argument that the absence of a door caused Ms. Weir’s rider to collide with the parked rider.