PATRICK E. HIGGINBOTHAM, Circuit Judge:

General Motors appeals from a judgment following a jury trial in a products liability crashworthiness case. It challenges both the sufficiency of the evidence to support the jury's findings of defective roof design and producing cause and the district court's evidentiary rulings regarding certain demonstrative evidence depicting rollover sequences. We affirm.

I

Holly Shipp, a nineteen-year-old Texas college student, and four of her friends were returning from a party in Holly's 1976 Pontiac TransAm automobile. While navigating the local "Dead Man's Curve," Holly lost control and the car completed a single rollover. The roof over the driver's seat collapsed, and though her four passengers were not seriously injured, Holly suffered a broken back and permanent paralysis. Holly sued, asserting that the car's roof had collapsed because it was defectively designed and that the roof's impact upon her shoulders broke her back. GM denied that the TransAm's roof was defectively designed or was a producing cause of injury. Rather, GM contended that Shipp, who was not wearing a seat belt, broke her back when she "fell" into the roof during the rollover. The jury accepted Shipp's theory and awarded her $750,000. After conforming the judgment to a written stipulation concerning past custodial care, the district court denied GM's motions for judgment notwithstanding the verdict and for new trial. GM here argues that there was insufficient evidence to support the jury's finding that the roof design was both defective and a producing cause of injury, and that the district court erred in certain evidentiary rulings.

II 1

We begin by noting that the jury's verdict, rendered after eleven days of trial, will not be lightly disregarded. Its findings must be upheld if this court, considering all of the evidence and all its reasonable inferences in the light most favorable to the winning party, finds that there is substantial evidence "of such quality and weight that reasonable and fair-minded men in the exercise of impartial judgment might reach different conclusions. . . ." Boeing Co. v. Shipman, 411 F.2d 365, 374 (5th Cir. 1969) (en banc); Liberty Mutual Insurance Co. v. Falgoust, 386 F.2d 248, 253 (5th Cir. 1967).

General Motors contends that our standard in reviewing the jury's finding of defective design is stricter given the fact its product met the only government standard relating to roof strength, Federal Motor Vehicle Safety Standard 216, 49 C.F.R. § 71.216 (1973). But obedience to the federal standard does not intensify our standard of review. While compliance is evidence, albeit persuasive and contradictory to plaintiff's proof, it is only a piece of the evidentiary puzzle. Of course, compliance with such minimum safety standards does not exempt or immunize a manufacturer from common law strict liability. Sours v. General Motors Corp., 717 F.2d 1511, 1517 (6th Cir. 1983); Ellis v. K-Lan Co., Inc., 695 F.2d 157, 161 (5th Cir. 1983). Certainly Congress did not intend such a result, for 15 U.S.C. § 1397(c) provides: "Compliance with any Federal motor vehicle safety standard issued under this subchapter does not exempt any person from any liability under common law."

2

In deciding whether, under Texas law, a product is defectively designed, a jury must balance the product's utility against the likelihood and gravity of injury from its use. Boatland of Houston, Inc. v. Bailey, 609 S.W.2d 743, 746 (Tex. 1980); Turner v. General Motors Corp., 584 S.W.2d 844, 847 (Tex. 1979). Texas courts have advanced balancing criteria to which strict liability parties should direct their evidence, but as the district judge did here, have only required that the jury be instructed in general terms to consider "the utility of the product and the risk involved in its use." Turner, 584 S.W.2d at 847.

Nonetheless, General Motors contends that under Texas law a plaintiff must offer evidence as to each criterion and that insufficiency of proof as to any one of them will defeat a finding of design defect as a matter of law. We disagree. The Texas Supreme Court has never explicitly made proof of each balancing factor a distinct element of a strict liability claim. See Boatland, 609 S.W.2d at 746 n. 2 ("a number of evidentiary factors may be considered in determining whether a product's design is defective") (emphasis added); Turner, 584 S.W.2d at 849 ("It has been pointed out that the evidence necessary to address the appropriate elements of balancing criteria should be overtly advanced by both parties in a strict liability action . . . .") (emphasis added). And certainly, that the jury is instructed in ultimate terms without detailing the criteria is at odds with the notion that proof of each is required.

3

Shipp relied heavily upon circumstantial evidence and the opinion testimony of her expert, Dr. Michael Kaplan. GM, in its sufficiency challenge, in turn fired upon Kaplan's opinions. At the same time, GM does not quarrel with the admission of Kaplan's testimony, perhaps recognizing that a trial court's qualification of a witness as an expert, Fed.R.Evid. 702, will not be disturbed absent an abuse of discretion. Ellis, 695 F.2d at 162; Garwood v. International Paper Co., 666 F.2d 217, 223 (5th Cir. 1982).

Dr. Kaplan testified that GM could and should have designed a stronger roof for the vehicle, with more headroom, to withstand the relatively low forces involved in this single revolution accident and to provide a reasonable amount of occupant protection. He believed the national standard to be an inadequate minimum because it was based on static forces and failed to establish a "non-encroachment zone" for occupant protection. Kaplan suggested a better method for testing the amount of roof deformation under dynamic rollover forces, which he demonstrated in conducting a "drop test" of a 1976 Pontiac TransAm. He cited a number of scientific studies which stressed the importance of establishing a protected space around automobile occupants and which supported his finding of a "very strong correlation" between roof crush and the frequency and severity of rollover injuries. Kaplan, albeit in a sketchy way, then proposed an alternative roof design factoring in roof strength and headroom which, while "similar to the cars on the road today," would minimize roof intrusion into occupant space. When asked how GM could have made a "safer roof, a stronger roof" for the TransAm, Kaplan suggested that heavier metal be used more efficiently by "boxing" it with certain welding techniques. Though Kaplan never designed a TransAm automobile with such features, he testified that testing of redesigned roof structures on other passenger cars supported the feasibility of his alternative technique. GM's own expert admitted that a stronger roof could indeed have been made for the TransAm, although such increased "strength" would reduce the energy absorbing characteristics of the design and force the energy to be absorbed in the lateral plane of multiple rollovers.

General Motors vigorously argues that Shipp never established that a stronger roof is a safer roof, and that the defectiveness of a product can only be determined by reference to safer alternatives. See Kindred v. Con/Chem, Inc., 650 S.W.2d 61, 62 (Tex. 1983); Boatland, 609 S.W.2d at 746. Kaplan testified, however, that Shipp's accident "involved forces that were sufficiently low so that a manufacturer . . . could have designed a roof to withstand it." He further stated that there was a "causal connection between how much roof deformation you get in a rollover accident and how seriously somebody is injured." The TransAm sustained substantial roof deformation, in his opinion, because the "[b]asic design of the roof is simply just too weak." In other words, roof weakness led to roof deformation, which in turn, resulted in serious injury. Otherwise stated, a stronger roof is a safer roof.

Despite the force of GM's argument, we are persuaded that the risk to passengers of roof deformation and the trade-offs involved in a more rigid roof design were questions for the jury. Holly Shipp offered evidence of the dangerousness of the TransAm roof and the feasibility and availability of a safer alternative. That no other passenger car employed this design will not alone defeat recovery. See General Motors Corp. v. Turner, 567 S.W.2d 812, 819 (Tex.Civ.App.-Beaumont 1978), rev'd on other grounds, Turner v. General Motors Corp., 584 S.W.2d 844 (Tex. 1979). The jury had before it evidence enabling it to engage in the necessary balancing and supporting its finding of defective roof design.

4

GM next contends that there was insufficient evidence to support a finding that the defective roof was a producing cause of Holly Shipp's extensive injuries. The attack on this finding is two-tiered. GM first argues that the evidence conclusively established that Shipp could not have been injured in the manner which she claimed. At the next level, it argues that even if sufficient evidence supports such factual causation, Shipp failed to meet her legal burden of proving causation in this, a crashworthiness case.

a

Shipp draws a causal inference from the fact that the bulk of roof deformation occurred over her seat and that she was the only passenger seriously injured. GM demurs, pointing out that "substantial" roof deformation also existed over the uninjured passenger seated behind Shipp. The jury, with both this argument and photographs of the damaged TransAm before it, apparently either agreed with Kaplan that the deformation over the left rear portion of the car was not as severe as that over the driver's seat or similarly discounted the inconsistency with other explanations. Regardless, the jury was entitled to infer that roof crush caused Shipp's paralysis.

This is not a situation where all experts agreed that the accident could not have occurred as plaintiff testified. See, e.g., Scronce v. Howard Brothers Discount Stores, Inc., 679 F.2d 1204, 1206 (5th Cir. 1982). Shipp's biomedical engineering expert. Dr. Leon Kazarian, testified that Holly Shipp's injuries were caused when the TransAm's roof, a "driving load," intruded into her occupant space and onto her shoulders. He "heartily" disagreed with GM's experts that Shipp's injuries were of the type most commonly received in falls or diving accidents. Both the orthopedic surgeon and physical therapist who treated Shipp testified her injuries were compatible with the type resulting from a "crushing type of force." Even GM's radiologist could not rule out the "unlikely" possibility that the driving force from roof crush caused the type of vertebral injury exhibited in Holly Shipp.

GM argues that Shipp's causation theory hinged on the assumption that she remained in her seat at the time the roof force was applied to her spine. Kaplan, however, described the interior of a TransAm as having "virtually no headroom," with the driver surrounded by the console, gear shift and steering wheel. Shipp testified that she always drove with her seat pulled full forward, placing her in a "very tight" area. She explained her lack of head or face injuries by the fact she was slumped to the right over the vehicle's console when the roof crushed in the back of her seat onto her shoulders. Though Shipp's x-rays indicated that no rotational or lateral forces had been applied, GM's expert acknowledged the x-rays might not have exhibited actual rotational movement. Thus the jury had before it sufficient evidence to make the "critical assumption" that Holly remained seated as the roof "crushed" onto her unprotected shoulders and to infer that the crush was the cause-in-fact of her permanent paralysis.

b

GM contends that even if Holly Shipp established factual causation, she failed to prove that the defective roof design "legally" caused her injuries. GM argues that in crashworthiness cases, where the alleged defect "causes or enhances injuries but does not cause the accident," Duncan v. Cessna Aircraft Co., 665 S.W.2d 414, 428 (Tex. 1984), a manufacturer is liable only for "enhancement" damages comprised of "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." Larsen v. General Motors Corp., 391 F.2d 495, 503 (8th Cir. 1968). It invites us to predict the choice of Texas courts when they face the current debate over the degree and burden of proof necessary to establish causation in crashworthiness cases.

Jurisdictions have split on this issue. One line adheres to the holding of Huddell v. Levin, 537 F.2d 726 (3d Cir. 1976) (New Jersey), which requires that a plaintiff establish the nature and extent of enhanced injuries. Others do not require a crashworthiness plaintiff "to prove with specificity the injuries which flowed specifically from [a product's] deficiencies." Fox v. Ford Motor Co., 575 F.2d 774, 787 (10th Cir. 1978) (Wyoming). Rather, a plaintiff need only put forth "some evidence of enhancement to present a jury issue," Lahocki v. Contee Sand Gravel Co., 41 Md. App. 579, 398 A.2d 490, 501 (Md.Ct. Spec.App. 1979), rev'd on other grounds sub nom. General Motors Corp. v. Lahocki, 286 Md. 714, 410 A.2d 1039 (1980), or show merely "that the design defect was a substantial factor in producing damages over and above those which were probably caused as a result of the original impact or collision." Mitchell v. Volkswagenwerk AG, 669 F.2d 1199, 1206 (8th Cir. 1982) (Minnesota). These decisions follow the principles enunciated in the Restatement (Second) of Torts, sections 433A and 433B, in holding a manufacturer, whose defective product is a producing cause of an indivisible injury, jointly and severally liable for total damages as a concurrent tortfeasor.

The courts of Texas, whose pronouncements we are here bound to follow, have not explicitly addressed the burden of proof in crashworthiness cases. GM contends that Texas would adopt the burden outlined in Huddell. It insists that Shipp should have proved not only that there was a safer design, but also the nature and extent of the injuries she would have suffered had it been utilized. Shipp, GM continues, was then required to apportion her injuries between those precipitated by defective design and those attributable to rollover forces, Huddell, 537 F.2d at 737-38; she fell short of this burden in failing to establish she would not have sustained her injuries had she been driving a TransAm designed with her hypothetically safer roof.

GM would deduce that the adoption of comparative causation in Duncan v. Cessna Aircraft Co., 665 S.W.2d 414 (Tex. 1984), signals the Texas view of the yet unfaced causation question — a view expressing a shift in the burden of proof and a retreat from the concurrent causation doctrine. We read a different signal.

The Texas version of comparative fault does not increase a plaintiff's burden, but continues to require only that a plaintiff prove that a defect was a producing cause of injury. While Duncan found comparative causation particularly appropriate in crashworthiness cases, where responsibility could be apportioned "between all whose action or products combined to cause the entirety of plaintiff's injuries," id. at 428, it cast the allocative burden on the defendant. A nonnegligent plaintiff may recover her total damages regardless of allocated percentages, from any or all of the jointly and severally liable defendants. The injured plaintiff is thereby compensated and the defendants are left to themselves to quarrel as to contribution. Id. at 429.

As explained in Lewis v. Timco, Inc., 716 F.2d 1425 (5th Cir. 1983) (en banc) (federal maritime law), while comparative fault inevitably measures relative causation, and is indeed sometimes referred to as comparative causation, causation that does not have its origin in fault is irrelevant. The ultimate consequence of Duncan is that "comparative" principles are inapplicable when there is only one liability-producing cause. The typical second collision case involves several causal "faults." A plaintiff's total damages ordinarily culminate after an act of independent negligence causes the initial accident, precipitating a product's defect. The decisions on which GM relies thus apportion total injury among several causes, each of which is a potential source of liability. We have here neither a joint tortfeasor nor a negligent plaintiff whose actions combine "to cause the entirety of a plaintiff's injuries." Though several causes may have concurred in causing Shipp's injuries, there was but one liability-producing cause on which to cast responsibility.

Forcing Shipp to identify in her proof the precise damages she would have sustained had the product been properly designed, would shift from proof of a producing cause of injury to sole producing cause. See Richardson v. Volkswagenwerk A.G., 552 F. Supp. 73, 80 (W.D.Mo. 1982). We see no retreat in Texas from its rule that a defect need only be a "producing cause" of injury and that there may be more than one such cause. Davidson v. Stanadyne, Inc., 718 F.2d 1334, 1341 n. 11 (5th Cir. 1983) (Texas); General Motors Corp. v. Hopkins, 548 S.W.2d 344, 351 (Tex. 1977) overruled in part on other grounds, Turner, 584 S.W.2d at 847 Duncan, 665 S.W.2d at 438; Caterpillar Tractor Co. v. Gonzales, 599 S.W.2d 633, 637 (Tex.Civ.App.-El Paso 1980, writ ref'd n.r.e.). The district court correctly instructed the jury that "[t]here may be more than one producing cause of an injury," and accurately defined "producing cause" as "an efficient, exciting, or contributing cause, which in a natural and continuous sequence in connection with any other cause or causes, produced the injuries." (emphasis added). See Rourke v. Garza, 530 S.W.2d 794, 801 (Tex. 1975); C.A. Hoover Son v. O.M. Franklin Serum Co., 444 S.W.2d 596, 597 (Tex. 1969).

Relatedly, Texas also adheres to the tenets of concurrent causation. If two or more causes produce a single injury and each is an efficient cause without which the damage would not have occurred, the injury may be attributed to any or all of the causes. McMillin v. L.D.L.R., 645 S.W.2d 836, 841 (Tex.App. — Corpus Christi 1982, writ ref'd n.r.e.); Dover Corp. v. Perez, 587 S.W.2d 761, 765 (Tex.Civ.App.-Corpus Christi 1979), supplemented, 591 S.W.2d 547 (Tex.Civ.App.-Corpus Christi 1979, no writ); South Austin Drive-In Theatre v. Thomison, 421 S.W.2d 933, 952 (Tex.Civ.App.-Austin 1967, writ ref'd n.r.e.); Chancey v. Van Luit, 306 S.W.2d 377, 380 (Tex.Civ.App.-Amarillo 1957, writ ref'd n.r.e.). Because the defect in a crashworthiness case is viewed as combining with other forces to produce a plaintiff's injuries, the entire injury alternatively could be attributed solely to the defect as a concurrent cause. GM's contention is therefore contrary to principles of legal causation long-established in Texas; principles left unchanged in any way relevant here by Texas' subscription to comparative fault.

Nor do we have any signal that Texas will otherwise distinguish crashworthiness cases as a group from ordinary products actions for purposes of either evidentiary sufficiency or burden of proof. The Texas courts perceive "no valid distinction in strict liability between a conscious design defect causing an accident and a conscious design defect causing an injury." Turner, 584 S.W.2d at 848. In holding strict liability principles equally applicable to second collision cases, Turner also found "no rational basis for a difference in the manner of submission of the issues to be determined by the factfinder." Id. at 848. While the Turner court was specifically dealing with the proper jury instruction on product defect, see supra notes 4-5 and accompanying text, it is logical to assume that the Texas Supreme Court would similarly find no rational basis for modifying the producing cause interrogatory. We are persuaded that Texas courts would conclude that it would be illogically harsh to force a plaintiff to segregate causation in crashworthiness cases, where "the collision, the defect, and the injury are interdependent and . . . viewed as a combined event." Turner, 584 S.W.2d at 848, quoting Huff v. White Motor Corp., 565 F.2d 104, 109 (7th Cir. 1977).

Shipp suggests that whatever change accompanied the Texas Supreme Court's adoption of comparative fault in strict liability cases, GM's concept of "enhancement" was encompassed within the producing cause and damages interrogatories submitted to the jury; that even with the additional burden such a construction would place upon her, its findings are supported by the evidence. The district court asked the jury whether "the defective design. . . was a producing cause of Holly Shipp's related injuries" (emphasis added). It requested the jury "to award those damages which would fairly and reasonably compensate Holly Shipp because of her fractured back and related injuries resulting from the defective design. . . ." (emphasis added). Shipp's argument is that GM was asked to pay only for those damages resulting from its defective product and thus was not charged with any injury attributable to other causal faults. She points out that GM did not object to either special issue, proposed interrogatories almost identical to those submitted, and did not request an interrogatory on apportionment or an instruction on enhancement. See Fietzer v. Ford Motor Co., 590 F.2d 215, 218 (7th Cir. 1978) (cited with approval in Duncan, 584 S.W.2d at 428).

Shipp's argument is not without force. We choose not to rest on this point, however, because it places too great a weight on the assumption the jury read "causation" to mean producing cause in interrogatory number three, but something more refined, when in answer to interrogatory number four, it awarded a sum to compensate "for injuries resulting from the defective design." (emphasis added.)

Holly Shipp's case as to design defect and producing cause was not compelling and GM has dealt it hard blows, yet a reasonable and fair-minded jury could infer from the circumstantial evidence and expert testimony that the roof of the TransAm was defectively designed and was a producing cause of Shipp's injuries. As it did, we must reject GM's challenge to the sufficiency of the evidence.

III

GM also attacks the admission of Shipp's exhibits demonstrating a drop test of a 1976 TransAm and the exclusion of GM's film illustrating unrestrained occupant movement in a rollover accident. The admission of such demonstrative evidence is within the trial court's sound discretion and will not be disturbed on appeal absent "abuse." Big John, B.V. v. Indian Head Grain Co., 718 F.2d 143, 146 (5th Cir. 1983); Jon-T Chemicals, Inc. v. Freeport Chemical Co., 704 F.2d 1412, 1417 (5th Cir. 1983). We find no abuse here.

The district court admitted film and photographs depicting a 1976 TransAm automobile, similar to the one involved in Holly Shipp's accident, being dropped onto the driver's side of its roof from an elevation of one foot. GM contends that the vertical drop demonstration was irrelevant to rollover accidents, occupant protection and roof strength. It complains it was prejudiced by the film's admission because, without a limiting instruction, it misled the jury into assuming the demonstration simulated Shipp's accident. See, e.g., Barnes v. General Motors Corp., 547 F.2d 275 (5th Cir. 1977). GM's relevancy challenges, however, go more to the weight the jury should have given the exhibits. It was relevant to roof strength and there was no abuse of discretion.

Finally, GM argues that the prejudice flowing from the admission of plaintiff's demonstrative evidence was compounded by the exclusion of its own film demonstrating occupant movement in a rollover accident. It contends the film was offered, not to simulate the accident, but to demonstrate the "principles of unrestrained occupant movement which are based upon the laws of physics, and which are therefore common to all rollover accidents."

Undoubtedly the demonstration was relevant to GM's theory that Shipp fell out of her seat and into the roof of her car during the course of the rollover. The trial judge recognized the film would "more than likely" show general principles of occupant movement. A central tenet of GM's argument, however, was that all rollover accidents are different. Yet, its filmed "accident" was a multiple, rather than a single revolution one, involving a different vehicle with a substantially different roof and passenger compartment than the TransAm involved in the Shipp rollover. The district court expressed its distrust of demonstrations involving vehicles other than a 1976 TransAm. After balancing the film's acknowledged relevance against the danger its admission would mislead, confuse, or prejudice the jury, Fed.R.Evid. 403, and determining that a limiting instruction would not defuse such improper influence, the district court excluded GM's film on the ground the jury would likely consider it as more than a simple demonstration of general principles.

Rule 403 determinations are often inextricably bound with the facts of a particular case and thus will not be disturbed absent a showing of "clear abuse." Wright v. Hartford Accident Indemnity Co., 580 F.2d 809, 810 (5th Cir. 1978). GM has failed to make such a showing. The district court was within its discretion in determining that it would be prejudicial and misleading to allow GM to demonstrate its causation theory using a different vehicle and a different accident.

GM projected its theory of causation through many other "vehicles," including still photographs, expert testimony, opening statement and final argument. The fact the "visual" aspect of the excluded exhibit may have greatly "impressed" the jury does not lessen its cumulative nature, see Golden Bear Distributing Systems, 708 F.2d at 951, and indeed increased the court's perception of its potential prejudice. That is, GM was not deprived of an opportunity to present evidence; rather it was not allowed to present it in the way it preferred. So viewed and so weighed on the Rule 403 scale, there was no error.

IV

Shipp asks this court to reform its judgment to allow her to recover an additional $48,136.08 in past medical expenses. She contends that defense counsel's stipulation of these expenses as "reasonable and necessary" obviated the necessity of introducing either the stipulation or evidence of such costs at trial. Shipp, however, has never filed a notice of cross-appeal. The district court's denial of her untimely motion for an extension to file such notice, Fed.R.App.P. 4(a), fell within its "broad discretion" to determine whether her explanation for such tardiness rose to the requisite level of "excusable neglect." Gann v. Smith, 443 F.2d 352, 353 (5th Cir. 1971). See Alvestad v. Monsanto Co., 671 F.2d 908, 911 (5th Cir.), cert. denied, 459 U.S. 1070, 103 S.Ct. 489, 74 L.Ed.2d 632 (1982); Pellegrino v. Marathon Bank, 640 F.2d 696, 698 (5th Cir. 1981). As the filing of a notice of appeal is a "mandatory precondition" to our exercise of jurisdiction, Curacao Drydock Co. v. M/V Akritas, 710 F.2d 204, 206 (5th Cir. 1983); Nelson v. Foti, 707 F.2d 170, 171 (5th Cir. 1983); Robicheaux v. Radcliff Material, Inc., 697 F.2d 662, 668 (5th Cir. 1983), we must decline to entertain Shipp's belated cross-points. The judgment of the district court is thus in all respects AFFIRMED.