dissenting.
This case is before us on remand from the United States Supreme Court for reconsideration in light of Williamson v. Mazda Motor of America, Inc., — U.S.-, 131 S.Ct. 1131, *60179 L.Ed.2d 75 (2011). Although I previously joined in the result that the majority reaches now for a second time, my reconsideration in light of Williamson leads me to the opposite conclusion.
In Williamson, the Court discussed Geier v. American Honda Motor Co., 529 U.S. 861, 120 S.Ct. 1913, 146 L.Ed.2d 914 (2000). In doing so, it made clear that state tort suits are preempted only when the evidence shows that retaining manufacturer choice is “a significant objective of the federal regulation.” Williamson, 131 S.Ct. at 1136 (emphasis in original). The Court’s language emphasizes the clear evidence it relied on in Geier to support a finding of preemption. Id. at 1137. (“DOT’S contemporaneous explanation of its 1984 regulation [found to preempt state tort suits in Geier ] made clear that manufacturer choice was an important means for achieving its basic objectives.”) (emphasis added). The Court then lists four specific reasons that had been articulated by the agency itself in that contemporaneous explanation for determining that manufacturer choice was needed. Id. (citing agency explanation that phase-in period for requiring airbags was needed because doing so would “give manufacturers time to improve airbag technology and develop other, better passive restraint systems”; avoid a public backlash; avoid potential injuries to unbelted occupants, particularly children; and avoid possibility that airbags would not be replaced when needed because of the high cost of doing so). The Court further explained that the history of the regulation at issue in Geier and the Government’s understanding of it also supported the conclusion that tort liability was preempted. Id. at 1136-37.
Turning to the regulation at issue in Williamson, the Court found that it, unlike the regulation at issue in Geier, did not reflect a significant objective of preserving manufacturer choice. The Williamson Court took the time to note each of the specific reasons given by the agency for the need to preserve manufacturer choice in Geier and their absence from the agency explanation in Williamson. Id. at 1138. The Court then turned to the reasons cited by the agency in Williamson for declining to require a single standard, and, despite some references to minor safety concerns, found that the agency had declined to require a particular safety measure *61because of cost-effectiveness concerns. Id. at 1138-39. Thus, although the agency clearly chose to maintain manufacturer choice, the Court found that it did not do so as an affirmative, significant objective.
Turning then to this case and the evidence regarding the NHTSA’s decision not to impose in FMVSS 205 a requirement that advanced glazing be used in side windows, as I read the relevant agency documents, they fail to demonstrate that the NHTSA had any objective of maintaining manufacturer choice, much less a significant objective in doing so. The NHTSA discontinued its study of the benefits of advanced glazing and withdrew its proposal to require it in side windows because it believed its resources would be better devoted to developing regulations related to other ejection mitigation devices and because of cost and minor safety concerns. See Notice of Withdrawal, 67 Fed.Reg. 41,365 (June 18, 2002). The NHTSA had explained in its Ejection Mitigation Using Advanced Glazing Final Report (Aug. 2001) (Final Report) that it encountered difficulty in quantifying the neck injuries that should be attributed to the use of advanced glazing in place of tempered glass. See Final Report at 36 (“No assessment of actual neck injury levels due to shear loads or moments was made since no accepted lateral neck injury criteria exist.”). The agency further noted extreme variability in test results aimed at collecting data on neck loads. See id. (stating that both the lowest and second highest measurements of axial neck load were obtained in replicate tests on tempered glass impacts). None of these reasons express the agency’s belief that manufacturer choice is needed in order to achieve an agency objective and therefore is a significant objective in and of itself.
The NHTSA’s explanation of its decision not to pursue study of advanced glazing parallels that of the Coast Guard in Sprietsma v. Mercury Marine, 537 U.S. 51, 123 S.Ct. 518, 154 L.Ed.2d 466 (2002). In Sprietsma, the United States Supreme Court found that a state tort action was not preempted when the Coast Guard declined to require installation of propeller guards on all boats. The Sprietsma Court characterized the Coast Guard’s explanation for its decision as “reveal[ing] only a judgment that the available data did not meet the ... ‘stringent’ criteria for federal regulation.” Id. at *6266-67, 123 S.Ct. 518. Moreover, the Coast Guard’s decision not to mandate propeller guards was due in part to concerns that “feasible propeller guards might prevent penetrating injuries but increase the potential for blunt trauma caused by collision with the guard.” Id. at 61, 123 S.Ct. 518. Nonetheless, the Sprietsma Court found that the Coast Guard “most definitively did not reject propeller guards as unsafe[,]” citing the Coast Guard’s indication that it might promote propeller guard use as a means of reducing propeller strike accidents. Id. at 67 & n. 11, 123 S.Ct. 518. Advanced glazing is much the same. It prevents ejection but might increase the risk of comparatively minor injury. In addition, the NHTSA has most definitively not rejected advanced glazing as unsafe, continuing to require its use in windshields. See FMVSS 205; O’Hara v. General Motors Corp., 508 F.3d 753, 761-63 (5th Cir.2007).
The Sprietsma Court also noted that the Coast Guard focused on the lack of a universally appropriate propeller guard for all types of boat operation; it reasoned that “nothing in [the Coast Guard’s] official explanation would be inconsistent with a tort verdict premised on a jury’s finding that some type of propeller guard should have been installed on this particular kind of boat equipped with respondent’s particular type of motor.” Id. at 67, 123 S.Ct. 518. Although the NHTSA has made no comparable statement to the effect that no single standard might be universally appropriate for side window glazing, it is also clear that the NHTSA did not make vehicle-specific determinations or formulate an objective that manufacturers’ fleets contain a mixture of devices, as the DOT did in Geier. See Geier, 529 U.S. at 878-81, 120 S.Ct. 1913. Indeed, if, as the majority finds, the NHTSA’s regulation was designed to maintain manufacturer choice so that they would install the safer choice for each vehicle, the regulation is entirely consistent with a tort verdict premised on a jury’s finding that advanced glazing should have been used in a particular window of a particular vehicle model. See Sprietsma at 67,123 S.Ct. 518.
Finally, I do not read the NHTSA explanation as finding that use of advanced glazing would “decreas[e] safety for *63lawfully belted passengers” as the majority concludes.21 Therefore, I find no basis for concluding that requiring advanced glazing would frustrate what the majority acknowledges is merely a collateral federal purpose of increasing seatbelt use. Williamson, supra.
Thus, I would reverse the grant of summary judgment and remand for further proceedings.
. Indeed, the Final Report implies that advanced glazing would provide some safety benefits to belted as well as unbelted occupants. The report indicates that belted occupants may occasionally be ejected from vehicles. See Final Report at 15. It also indicates that a substantial portion of ejection injuries result from partial ejections, though without noting whether belted occupants may be partially ejected. Id. at 9. The Final Report does not indicate whether estimates of the overall costs and benefits of advanced glazing for belted occupants are available.