Michael Bavlsik v. General Motors

RILEY, Chief Judge. These appeals are driven, in large part, by the standards of review. About five years ago Michael Bavlsik was- driving his 2003 GMC Savana van when he collided with a boat being towed by another vehicle. Bavlsik was wearing his seatbelt, but that did not prevent him from hitting his head on the roof when the van rolled over. As. a result, Bavlsik sustained a cervical-spinal cord injury and is *803 now a quadriplegic. Bavlsik and his wife, Kathleen Skelly, sued General Motors, the company that designed and manufactured the van, for: (1) strict liability, asserting the seatbelt system lacked three specific safety features; (2) negligent design, based on GM’s failure to implement these safety features or conduct adequate testing on the van; and (3) failure to warn. After an eleven-day trial, the jury found GM negligent for failing to test the van and such negligence caused Bavlsik’s injuries. The jury rejected all other claims and theories. Bavlsik was set to recover $1 million (all for past damages), until the trial court granted GM’s renewed motion for judgment as a matter of law (JML) and set aside the verdict. On Bavlsik’s and Skelly’s motion, the trial court also conditionally granted a new trial solely as to damages. Both decisions are before us now. Bavlsik and Skelly contend they presented sufficient evidence to support the verdict, therefore GM was not entitled to JML. GM disagrees, and argues that if a new trial is necessary, then the parties should also retry the liability issue. We reverse the grant of JML, and affirm the grant of a new trial on damages only. See 28 U.S.C. § 1291 (appellate jurisdiction). I. BACKGROUND A. The Crash On July 7, 2012, Bavlsik was driving two of his sons and eight others home to St. Louis after spending a week at Boy Scout camp in northern Minnesota when he hit a boat and trailer being towed by a pickup truck. The initial collision did not cause any significant harm, but then Bavlsik’s vehicle—a twelve-passenger 2003 GMC Savana van he had purchased nine years earlier—swerved and completed a three-quarters roll at a relatively low speed. Bavlsik was wearing his seatbelt, but still slid far enough out of his seat to hit the roof of the van with enough force to dislocate his neck and sever his spinal cord. No one else was seriously hurt. Today, Bavlsik is a quadriplegic. He has “no motor movement below [his] chest,” however he was able to regain partial function of his arms after a nerve transplant and considerable rehabilitation work. Bavlsik’s limitations have had predictable effects on his life. Professionally, Bavlsik was able to resume his work as a doctor just a few months after the accident. Needless to say Bavlsik’s medical practice has changed—he “see[s] less patients in the office” due to his problems getting around, he has “lost a lot of patients,” and he has to work harder to accomplish routine tasks. Personally, Bavlsik misses the way life was when he could hike, bike, swim, and maintain an active lifestyle with his family. Bavlsik also worries about what the future holds, both for himself and his family. According to Skelly, she shares many of these feelings and concerns. And financially, not only have Bavlsik’s professional prospects been curtailed, but he will also need to pay for some form of care for the rest of his life. B. The Case Bavlsik and Skelly filed a products-liability suit against GM in the Eastern District of Missouri less than one year after the accident. See 28 U.S.C. § 1332(a)(1) (diversity jurisdiction). The complaint included claims for strict liability, negligent design, and failure to warn. Bavlsik sought past and future damages for loss of income, pain and suffering, medical expenses, and punitive damages; Skelly sought additional damages for loss of consortium. Both sides consented to a magistrate judge presiding over the action. See id. § 636(c)(1) (magistrate jurisdiction). The case culminated in a multi-week jury trial in September 2015. The founda *804 tion of the plaintiffs’ case-in-chief was crafted around four key facts: first, there was no pretensioner, a device that activates in the event of a crash and removes slack from the seatbelt; second, the van did not employ an all-belts-to-seat design, which (as the name implies) consists of attaching the seatbelt to' the seat rather than the body of the vehicle; third, the seatbelt did not use a sliding-cinching latch plate, which limits how freely the latch moves on the webbing of the belt; fourth, the van’s seatbelt system had not been tested to see how it would perform during a rollover accident. There was no dispute about whether these four facts were true. Rather the case hinged on the significance of these facts. Bavlsik’s and Skelly’s expert, Larry Sicher, testified that the lack of the three features he identified rendered the van’s seatbelt system defective, testing would have revealed as much, and implementing any of these design alternatives would have prevented Bavlsik’s injuries. Sicher’s testimony, was the primary way the plaintiffs tried to satisfy their burden for the factual questions facing the jury. On the strict liability claim, did the lack of the three proposed safety features mean the van was “in a defective condition unreasonably dangerous when put to a reasonably anticipated use?” On the negligence claim, did the absence of any of these features and the lack of testing mean GM breached a duty by designing the van as it did? 3 And for both claims, there was the issue of causation—would these features or some type of testing have prevented Bavlsik’s injuries? When the plaintiffs rested their case on day six of trial, GM moved for JML. See Fed. R. Civ. P. 50(a). According to GM, there was insufficient proof “that any al-tentative design ... would have made any difference,” and as for testing it was unclear “what the 'test should have been” or “in what way the information gathered from such a test should have been used.” Bavlsik and Skelly countered, citing their expert’s testimony about the effect the proposed features have on keeping passengers safely in their seats during a rollover. Bavlsik and Skelly also highlighted testimony about the “importance of testing” and posited that had there been adequate testing, “maybe [GM] could have considered some alternative—some of the many alternative designs that were offered into evidence in this case.” The trial court orally denied JML, so GM proceeded with its case-in-chief. At the close of all evidence, GM.renewed its motion for JML “for the same reasons previously stated,” plus its supposed “direct evidence that ... none of the alternatives ... are actually effective and.that there is nothing feasible that could have been done that would have prevented the injury.” Again, the trial court orally denied the motion. , The trial court submitted the plaintiffs’ claims .on a general verdict-form with special interrogatories that listed all of their theories within each claim (including lack of testing for negligent design). The jury returned a verdict after over four hours of deliberation, finding GM was negligent for not testing the van’s seatbelt system, and that negligence directly caused Bavlsik’s injuries. The jury found GM was not strictly liable or negligent for failing to implement any of the specific safety features Bavlsik and Skelly had proposed. With the verdict, Bavlsik was to recover $1 million—all for past damages, none for future damages—and Skelly was to recover nothing. GM did not object to the jury instructions, the verdict form, or the verdict itself. *805 Both sides filed post-trial motions. Bavlsik and Skelly moved for a new trial only on the damages issue. See Fed. R. Civ. P. 59(a). GM renewed its motion for JML, see Fed. R. Civ. P. 50(b), and alternatively moved for a new trial only on the failure-to-test portion of the negligent-design claim, see Fed. R. Civ. P. 59(a). This time the trial .court granted GM’s request for JML, reasoning “[t]he jury’s finding of no defect rendered the other finding of negligent failure to adequately test a legally insufficient basis for liability.” From this, Bavlsik and Skelly appeal. In addition, the trial court conditionally granted Bavlsik and Skelly a new trial on damages only, because the jury’s award was “shockingly inadequate.” See Fed. R. Civ. P. 50(c)(1). From this, GM conditionally cross-appeals. II. DISCUSSION A. Judgment as a Matter of Law We must first decide whether the district court was right to grant GM’s renewed motion for JML, which is a question we review de novo. See Stults v. Am. Pop Corn Co., 815 F.3d 409, 418 (8th Cir. 2016). Here, GM is entitled to JML only if “a reasonable jury .would not have a legally sufficient evidentiary basis” to return a verdict for Bavlsik and Skelly on their failure-to-test theory of negligent design. Fed. R. Civ. P. 50(a)(1). “[T]he law places a high standard on overturning a jury verdict because of the danger that the jury’s rightful province will be invaded when judgment as a matter of law is misused.” Hunt v. Neb. Pub. Power Dist., 282 F.3d 1021, 1029 (8th Cir. 2002) (citation omitted). The proper analysis for considering renewed JML motions reflects our hesitancy to interfere with a jury verdict: “[T]he [trial] court must (1) consider the evidence in the light most favorable to the prevailing party, (2) assume that all conflicts in the evidence were resolved in favor of the prevailing party, (3) assume as proved all facts that the prevailing party’s evidence tended to prove, and (4) give the prevailing party the benefit of all favorable inferences that may reasonably be drawn from the facts proved. That done, the court must then deny the motion if reasonable persons could differ as to the conclusions to be drawn from the evidence.” Ryther v. RARE 11, 108 F.3d 832, 844 (8th Cir. 1997) (en banc) (first alteration in original) (quoting Haynes v. Bee-Line Trucking Co., 80 F.3d 1235, 1238 (8th Cir. 1996)); accord Browning v. President Riverboat Casino-Mo., Inc., 139 F.3d 631, 634 (8th Cir. 1998) (“Judgment as a matter of law is proper only when the evidence is such that, without weighing the credibility of the witnesses, there is a complete absence of probative facts to support the verdict.”). With this perspective in mind, we must determine whether there was legally sufficient evidence to support the jury’s liability finding. 4 We hold there was. The jury found GM liable for “not adequately test[ing]” the van. This theory of liability was presented to the jury as a subpart of the broader negligent-design claim, so Bavlsik and Skelly had the burden of establishing the traditional negligence elements: duty, breach, causation, *806 and damages. See Stanley v. Cottrell, Inc., 784 F.3d 454, 463 (8th Cir. 2015) (“To prove a negligent design claim under Missouri law, a plaintiff must show that the defendant breached its duty of care in the design of a product and that this breach caused the injury”). We assess these elements in turn. GM makes no meaningful argument it did not have a duty to exercise reasonable care in designing the van. Both Missouri appellate courts and our court have recognized companies have a duty to exercise due care when they design and manufacture a potentially dangerous product, which includes taking reasonable steps to reduce the likelihood of such injury. See, e.g., Johnson v. Auto Handling Corp., No. SC 95777, 523 S.W.3d 452, 457-58, 464-66, 2017 WL 2774620, at *1-2, *7-8 (Mo. June 27, 2017); Mathes v. Sher Express, L.L.C., 200 S.W.3d 97, 109 (Mo. Ct. App. 2006); McKnight ex rel. Ludwig v. Johnson Controls, Inc., 36 F.3d 1396, 1411 (8th Cir. 1994). 5 Thus GM had a duty to exercise reasonable care in designing the van. Whether GM breached that duty was “a question of fact for the jury.” Lopez v. Three Rivers Elec. Co-op., Inc., 26 S.W.3d 151, 157 (Mo. 2000). To support their failure-to-test theory, Bavlsik and Skelly relied on their design expert, Sicher, who testified about the important role testing plays (or should play) in the design process: “[T]he basic design principles are set your goals, determine how you’re going to test or evaluate them, and then start testing them.” If the test results are unsatisfactory, then the company should “go back and either do a redesign or evaluate whether that was a true indicative method of evaluating [the goal] properly.” That is, in Sicher’s opinion, rollover testing is vital to producing a careful design because it allows a company like GM to discover how a vehicle performs during a rollover and what alternatives, if any, can improve that performance—to him, failure to test means failure to exercise reasonable care. Cf. McKnight, 36 F.3d at 1411 (recognizing “[f]ailure to test is a viable theory of recovery under Missouri law” in a manufacturing defect case); Zesch v. Abrasive Co. of Phila., 353 Mo. 558, 183 S.W.2d 140, 145 (1944) (“[WJhere it is shown that the imperfection could be disclosed by a test, it would seem reasonable that the manufacturer in the exercise of ordinary care would be under a duty to make the test.”). Such testing did not happen here. In a deposition played to the jury, GM’s corporate representative admitted the company conducted no rollover testing to assess the seatbelt system’s performance before bringing the van to market in early 2003. Without testing, GM could not know whether the van provided adequate protection to occupants during a rollover, or whether any reasonable alternatives would have afforded additional protection. To be sure, there was evidence GM conducted *807 compliance testing and met certain required safety standards. Yet as the jury was instructed, proof of such compliance “is relevant to, but not determinative of, whether the manufacturer exercised ordinary care in the design of its motor vehicles.” The jury was still free to accept Sicher’s testimony and find GM breached its duty by not conducting any sort of rollover testing before selling the van. 6 We proceed to the more hotly contested issue at trial and on appeal—causation: whether GM’s negligence “directly caused” the harm, meaning Bavlsik would not have been injured “ ‘but for’ ” the van’s negligent design and GM’s failure to test. Poage v. Crane Co., No. ED 103953, 523 S.W.3d 496, 508-09, 2017 WL 1632580, at *3-4 (Mo. Ct. App. May 2, 2017) (quoting Callahan v. Cardinal Glennon Hosp., 863 5.W.2d 852, 863 (Mo. 1993)). In this case, the causation requirement means Bavlsik and Skelly had to prove both that testing would have shown the van did not provide adequate protection during a rollover, and that testing would have prompted GM to explore and implement a safer design capable of preventing Bavlsik’s injuries. Like breach, causation “is a factual question left for the jury.” Id. at 508, 2017 WL 1632580, at *3. Here the jury was properly instructed on the causation question, and found Bavlsik and Skelly satisfied their burden. GM contends there is legally insufficient evidence to support this finding, suggesting at oral argument that Bavlsik’s and Skelly’s causation evidence is “vague, and speculative, and woefully inadequate to support th[e] verdict.” We disagree. The heart of the plaintiffs’ causation evidence came from Sicher, who unequivocally opined that testing would have shown the van was not safe during a rollover, and “that there were designs available ... that would have prevented Dr. Bavlsik’s injuries.” Sicher, a mechanical engineering expert with over twenty years of experience, explained the many tests he relied upon to reach this critical opinion. 7 First came the seminal “Malibu II” report in 1990, which revealed significant neck injuries were likely during a rollover, even if the occupant was wearing a seatbelt. Then there were a number of tests over the next fifteen-plus years that showed, according to Sicher, how various design alternatives can improve seatbelt effectiveness and reduce rollover injuries. Last came research GM conducted on the Savana model in 2007 and 2014, which Sicher said is proof the Savana’s seatbelt system “d[id] not provide a reasonable level of protection in a rollover.” Taken together, Sicher’s testimony and the peer-reviewed literature he relied upon support a reasonable inference that pre-2003 testing would have revealed the Savana seatbelt system was inadequate and could have been improved by adding feasible safety features. Sicher also demonstrated a sufficient understanding of the relevant case-specific circumstances. At 6’1” and about 260 pounds, Bavlsik was not a small man. Bavlsik had about four inches of clearance between his head and the roof when he *808 was seated in a “normal driving position” like he was when the collision occurred. Upon impact, Bavlsik’s van began a counterclockwise yaw and completed a three-quarters roll at around 11 to 15 miles per hour, beginning on the passenger’s side and stopping on the driver’s side. Evidence suggested the van was flipped exactly 180 degrees when Bavlsik sustained the injury. According to Sicher, he accounted for all these factors in reaching his ultimate conclusion that his proposed design changes would have prevented Bavlsik’s injuries. For its part, GM sought to downplay its failure to test by casting doubt on whether test results would have shown any of Sieher’s proposed designs to be effective. GM’s counsel cross-examined Sicher at length about supposed flaws in his methodology and whether the available testing actually supported his conclusions. GM then called two of its own engineering experts, who rebutted Sicher’s opinions and opined there were no feasible design changes that would have prevented Bavlsik’s injuries. GM’s argument seemed to be that testing would have done nothing more than show there was a problem with no solution. The jury heard considerable expert testimony about whether testing would have revealed design alternatives capable of protecting Bavlsik during the rollover. Sicher “testified extensively” about his opinion on the matter and was “subjected to lengthy and detailed cross-examination.” Adams v. Toyota Motor Corp., Nos. 15-2507, -2516, -2635 to 2638, 867 F.3d 903, 918, 2017 WL 3445112, at *7 (8th Cir. Aug. 11, 2017). GM’s experts did the same and were similarly .challenged. This is a classic example of conflicting evidence that must be weighed and decided by a jury, not a court. See id. (“Though Toyota disagrees with Stilson’s opinions and conclusions-... ‘questions of conflicting evidence must be left for the jury’s determination,’ and we will not re-weigh the evidence.” (quoting Bonner v. ISP Techs., Inc., 259 F.3d 924, 930 (8th Cir. 2001))). It appears the jurors believed Sicher, at least in regards to testing, which they were entitled to do. 8 Thus viewing the evidence and accepting all inferences in the light most favorable, to Bavlsik and Skelly, we find legally sufficient evidence to support the jury’s causation finding. . We end' our JML analysis with damages, the final element and the one that requires little .analysis in this case. For Bavlsik, there was an abundance 'of evidence about the harm he suffered physically, emotionally, professionally, and financially. For Skelly, there was evidence she plainly sustained loss-of-coñsortiúm damages. (We discuss the extent of such damages in more detail below.) In sum, there was legally sufficient evidencé for a reasonable jury to' find GM liable for negligent design, specifically for failing to conduct adequate testing, JML was improper. B. New Trial Given our decision above, we must now decide whether the trial court was wrong conditionally to “grant a new trial only on plaintiff Bavlsik’s future damages and on plaintiff Skelly’s damages, past and future.” See Fed. R. Civ. P. 59 (new trial); see also Fed. R. Civ. P. 50(c)(1) (conditional new-trial rulings). GM contends the trial court abused its discretion by limiting the potential new trial .like this “because the entire record demonstrates the compromise character of the verdict.” GM says the parties should retry the failure-to-test *809 theory in its entirety, including the question of liability. Although GM makes a strong case, we are unable to say the trial court abused its ‘ considerable discretion and committed reversible error. A trial court “may, on motion, grant a new trial on all or some of the issues,” provided there is a good reason to do so. Fed. R. Civ. P. 59(a)(1). It is generally permissible for a trial court to grant a new trial on damages only. See Haug v. Grimm, 251 F.2d 523, 527-28 (8th Cir. 1958); see also Gasoline Prods. Co., Inc. v. Champlin Ref. Co., 283 U.S. 494, 499-500, 51 S.Ct. 513, 75 L.Ed. 1188 (1931). That is what happened here. The trial court first referenced Missouri law and recognized reversal on damages was warranted if the jury’s award was “shockingly inadequate.” See Riordan v. Corp. of the Presiding Bishop of the Church of Jesus Christ of Latter-Day Saints, 416 F.3d 825, 833 (8th Cir. 2005); see also Niemiec v. Union Pac. R.R. Co.. 449 F.3d 854, 858-59 (8th Cir. 2006) (noting “state law governs the question of adequacy of damages” even though “ ‘the appropriateness of a new trial is a federal procedural question decided by reference to federal law’ ” (quoting Sanford v. Crittenden Mem’l Hosp., 141 F.3d 882, 884 (8th Cir. 1998))). The trial court then used this standard to evaluate the evidence presented at trial and concluded the damages were indeed “shockingly inadequate.” We review this conclusion only for an abuse of discretion. See Dominium Mgmt. Servs., Inc. v. Nationwide Hous. Grp., 195 F.3d 358, 366 (8th Cir. 1999); see also Allied Chem. Corp. v. Daiflon, Inc., 449 U.S. 33, 36, 101 S.Ct. 188, 66 L.Ed.2d 193 (1980) (per curiam) (“The authority to grant a new trial ... is confided almost entirely to the exercise of discretion on the part of the trial court”). The trial court did not abuse its discretion. As the trial court put it, the plaintiffs proved Bavlsik sustained “a permanent injury that would require medical care of some sort for the rest of his life.” Such care does not come cheap—Bavlsik’s and Skelly’s expert on the issue predicted Bavlsik would -incur $7 million in life-care costs if he lived to age 79. This is in addition to whatever Bavlsik lost in earning potential, which another expert said will likely range between $296,000 (if Bavlsik worked until age 67) and $5.8 million (if Bavlsik quit immediately). As for Skellyj the evidence was clear regarding the impact her husband’s injury had and will continue to have on her. Though GM questioned the size of the plaintiffs’ claims, it did not suggest the jury should award $0 for future damages or loss of consortium. The parties fought about the extent of damages, not the existence of them. Rather than try to justify the award, GM posits the low. damages figure is one of several signs the verdict represents an improper compromise. While it is true a retrial on only damages is sometimes proper, it is inappropriate “where there is good reason to believe that the inadequacy of the damages awarded was induced by unsatisfactory proof of liability and was a compromise.” Haug, 251 F.2d at 527-28; see also Phav v. Trueblood, Inc., 915 F.2d 764, 767 (1st Cir. 1990) (“Where a verdict is set aside because of an inadequate damages' award, retrial of all the issues is required ‘if the verdict could only have been a sympathy or compromise verdict.’” (quoting Spell v. McDaniel, 824 F.2d 1380, 1400 (4th Cir. 1987))). Generally such situations necessitate a new trial on all claims and issues. See Carter v. Moore, 165 F.3d 1071, 1083 (7th Cir. 1998) (“When a court recognizes that the jury’s verdict is a result of impermissible compromise, such a verdict taints the entire proceeding and the proper remedy is a new trial on all issues.” (emphasis added)). *810 “A compromise verdict results when the jury, unable to agree on the issue of liability, compromises that disagreement by awarding a party inadequate damages.” Boesing v. Spiess, 540 F.3d 886, 889 (8th Cir. 2008). “[S]everal factors” are often probative of whether the jurors improperly compromised, like “a close question of liability, a grossly inadequate award of damages, and other circumstances such as the length of jury deliberations.” M. Yet “the overarching consideration must be whether the record, viewed in its entirety, clearly demonstrates the compromise nature of the verdict.” Id. The trial court explicitly rejected any suggestion of a compromise verdict, perceiving “no question regarding the jury’s limited finding of liability.” We note two overlapping forms of deference are at play in our review of this conclusion—we defer first to the jury, as we start with the assumption jurors fulfilled their obligation to decide the case correctly; and we defer second to the trial court, which “has a far better sense of what the jury likely was thinking and also whether there is any injustice in allowing the verdict to stand.” Nichols v. Cadle Co., 139 F.3d 59, 63 (1st Cir. 1998); see also id. (“[T]here is ... a settled hostility toward [finding] ‘compromise’ verdicts.”); Fairmount Glass Works v. Cub Fork Coal Co., 287 U.S. 474, 485, 53 S.Ct. 252, 77 L.Ed. 439 (1933) (“Appellate courts should be slow to impute to juries a disregard of their duties, and to trial courts a want of diligence or perspicacity in appraising the jury’s conduct.”). We proceed with this doubly deferential standard in mind. First, as discussed above, both sides agree the damages award is seriously inadequate—although they disagree on the degree of the inadequacy. Given that reduced damages are part of the very definition of a compromise verdict, this factor exists in nearly every case where a court finds an improper compromise. See, e.g., Mekdeci ex rel. Mekdeci v. Merrell Nat’l Labs., 711 F.2d 1510, 1513-14 (11th Cir. 1983). “Insufficient damages alone, however, do not establish a compromise verdict.” Reider v. Philip Morris USA, Inc., 793 F.3d 1254, 1260 (11th Cir. 2015). If the rule were any different, a retrial on damages only would never be appropriate, and that is not the case. To be sure, the low verdict amount is consistent with a compromise verdict. However, it falls short of convincing us the trial court abused its discretion in deciding the better route was to order a new trial to remedy the inadequate damages problem. Next, GM draws our attention to an “odd pattern of jury deliberations.” See Phav, 915 F.2d at 768 (“In addition to inadequate damages, the telltale signs of a compromise verdict are a close question of liability and an odd chronology of jury deliberations.”). The basis for this argument is a note the jury submitted two hours into their deliberations asking whether Bavlsik would recover the stipulated amount of past medical expenses— about $577,000—“regardless of our decision.” The trial court informed the jury Bavlsik would recover that- sum only if the jury found GM liable. The jury reached a unanimous verdict for $1 million two hours later. GM tries to analogize this to the situation in Skinner v. Total Petroleum, Inc., 859 F.2d 1439 (10th Cir. 1988). In Skinner, the jury asked to see “testimony relating solely to the issue of liability” during the second day of deliberations. Id. at 1446. Then the jury explicitly “informed the court that it was unable to reach a unanimous decision.” M. The court told the jury to continue its effort, and a short time later there was a unanimous verdict. See id. The Tenth Circuit found the “sudden *811 arrival at unanimity, when just a few hours before [the jury] was still struggling with an apparently close issue of liability,” to be “suspect.” Id. Those are not the facts here. The jury deliberated for only four hours, a reasonable (if not short) length of time for a complex eleven-day trial. See Boesing, 540 F.3d at 889-90 (rejecting a compromise claim despite multiple days of deliberations, an Allen charge, and lower-than-expected damages); see also Burger King Corp. v. Mason, 710 F.2d 1480, 1488 (11th Cir. 1983) (“After a long, protracted trial, the jury required only two to three hours to reach its verdict. It obviously was not deadlocked.”). While the jury’s question could indicate a struggle with liability, it could also be read as a request for clarification on what the jury needed to account for in deciding damages or whether the jury could add to or reduce the stipulated number if they saw fit to do so. After all, this was the only pre-filled answer on the verdict form and the instructions made clear the plaintiffs’ right to recover was dependent on the jury first finding GM liable. Much like the inadequate damages, this factor raises the possibility the jurors compromised but it does not compel such a conclusion. See Boesing, 540 F.3d at 889-90; see also Phav, 915 F.2d at 768 (stating “[t]here [wa]s nothing in the deliberation process indicating a compromise verdict” even though the jury deliberated for four hours, asked if the verdict had to be unanimous, and had the court repeat the liability instruction). That leaves GM’s argument about the close liability question and the jury’s seemingly inconsistent verdict—how could the jury find rollover testing would have led to a better design capable of preventing Bavlsik’s injuries if the jury seemingly rejected the only design alternatives the plaintiffs offered? 9 We admit we share GM’s confusion about how the jury reached the conclusion it did. But the proper way to resolve this problem would have been for GM to object to the format of the verdict form or the final verdict itself—it did neither. See Chem-Trend, Inc. v. Newport Indus., Inc., 279 F.3d 625, 629 (8th Cir. 2002). The First Circuit faced a similar situation in Phav v. Trueblood, Inc., where the jury’s causation findings could not be reconciled with one another. See Phav, 915 F.2d at 765. The court affirmed a second trial solely on damages, notwithstanding this apparent inconsistency, because the defendant failed to object to the special interrogatories or the jury’s verdict. See id. at 769. Our sister circuit reasoned: [T]he use of special interrogatories puts the parties on notice that there might be an inconsistent verdict. If a slip has been made, the parties detrimentally affected must act expeditiously to cure it, not lie in wait and ask for another trial when matters turn out not to their liking. ... Because of defendant’s waiver, we do not consider the jury’s answers to the special questions as evidence of its confusion on liability. To decide otherwise would countenance agreeable acquies *812 cence to perceivable error as a weapon of appellate advocacy. Id. (citation and internal quotation marks omitted); see Fox v. City Univ. of N.Y., 187 F.R.D. 83, 92 (S.D.N.Y. 1999) (“In [Phav], the First Circuit held that a party’s failure to' object to the form of the special interrogatories submitted to the jury precluded a subsequent argument that the answers indicated a compromise verdict.”); see also Buchwald v. Renco Grp., Inc. (In re Magnesium Corp. of Am.), 682 Fed.Appx. 24, 29-30 (2d Cir. 2017) (summary order) (rejecting a compromise-verdict claim “because that would- sneak [a waived inconsistency claim] in through the back door, while undermining the principle that the jury must be given the opportunity to reconcile any apparent or alleged inconsistency in the first instance” (alteration in original) (citation and internal quotation marks omitted)) pet. for cert. filed (U.S. Aug. 09, 2017) (No. 17-228); cf. Reider, 793 F.3d at 1261 (“[N]ot all inconsistent verdicts are compromise verdicts.”). We agree with this reasoning and refuse to consider the jury’s unclear answers in deciding whether there was an improper compromise. Our analysis may have been different had GM preserved the issue for our review. But GM did not do so, perhaps because making a timely objection to the verdict might have reduced its odds of prevailing. Now the confusion lingers on appeal in a repackaged argument about a compromise verdict. We decline to. make Bavlsik and Skelly pay the price for GM not acting on this perceived error in a timely manner. ... Having closely reviewed the record, we are not convinced the record so clearly demonstrates a compromise verdict that the trial court abused, its discretion in not recognizing as much. 10 See Boesing, 540 F.3d at 889. The facts are such there were a number of options the trial court could choose from in deciding whether a new trial was warranted, and if so, how much of the case should be retried. Because we are satisfied the issues regarding damages and liability are “distinct and separable” from one another, a new trial for Bavlsik’s future' damages and Skelly’s past and future damages was one of those permissible options. Champlin, 283 U.S. at 500, 51 S.Ct. 513. III. CONCLUSION We reverse the trial' court’s decision to grant GM’s renewed motion for JML, and affirm the trial court’s conditional grant of a partial new trial on damages. Footnotes [3] Neither side makes any argument about the failure-to-warn claim, and because we agree it has no relevance to this appeal, we will not elaborate on it here. [4] The trial court did not adhere to this evidence-centric approach in granting GM's renewed motion for JML, and instead relied largely (if not exclusively) on the jury's findings. Much of the parties’ briefing focused on whether this approach was compatible with our precedent regarding Rule 50 and what significance, if any, we should place on the jury’s other findings. However, at oral argument GM conceded the point and invited us to focus only on the evidence. We accept this concession, and assume for the sake of this appeal that everything else (including the jury’s other findings) is irrelevant. [5] We described a car manufacturer’s duty to design its vehicles with care in Larsen v. General Motors Corp., 391 F.2d 495, 502-03 (8th Cir. 1968). There, we held "[a] manufacturer is under a duty to use reasonable care in the design of its vehicle to avoid subjecting the user to an unreasonable risk of injury in the event of a collision.” Id. at 502. We went on to explain ‘‘[t]he manufacturers are not insurers but should be held to a standard of reasonable care in design to provide a reasonably safe vehicle in which to travel. ... At least, the unreasonable risk should be eliminated and reasonable steps in design taken to minimize the injury-producing effect of impacts.” Id at 503. Larsen attempted to divine and apply Michigan law, see id at 497, but it is a “landmark decision” of sorts on this issue, and we have looked to the Larsen court’s reasoned analysis in a products-liability case governed by Missouri law. See Polk v. Ford Motor Co., 529 F.2d 259, 264, 266 (8th Cir. 1976) (en banc). [6] Though it does not affect the analysis given our de novo review, we note the trial court’s only reference to the evidence itself was a comment that the jury’s conclusion GM "negligently failed to adequately test the seat belt restraint system” was “supported by legally sufficient evidence.” [7] GM moved in limine to exclude Sicher’s testimony, arguing he was not qualified to testify as an expert, the testing he relied on was unreliable and inapplicable to this case, and his conclusions were not adequately supported. The trial court rejected these arguments and certified Sicher as an expert. See generally Fed. R, Evid. 702. GM does not appeal this decision. [8] We reiterate, although it may seem counter-intuitive and odd from a practical perspective, our focus is confined to what the jury found in regards to this issue (GM is liable for failure to test), without any regard to what the jury found on other issues. [9] An inconsistent verdict is one in which the jury ”reach[es] contradictory factual findings.” Lowry ex rel. Crow v. Watson Chapel Sch. Dist., 540 F.3d 752, 762 (8th Cir. 2008). GM resists classifying this argument as one about inconsistent verdicts, but we see no other way to characterize it. If GM thought the failure-to-test theory was only viable if the jury found for Bavlsik and Slcelly on one of their other alleged theories, then GM should not have allowed the jury to be instructed that failure to test was an independent basis for liability (or at least objected when the jury reached that conclusion). The jury merely followed the instructions given to it. [10] To the, extent GM renews its arguments about whether substantial evidence supports the jury’s liability finding, we. refer to our lengthy discussion earlier in regards to JML and the trial court s conclusion on the issue. See Children's Broad. Corp. v. Walt Disney Co., 357 F.3d 860, 867 (8th Cir. 2004).