concurring in part and dissenting in part.
I agree with much of the majority’s well-reasoned opinion, but I respectfully disagree with the conclusions in Parts IV(A)(l)(a) and IV(D), which address (1) the sufficiency of the evidence for an award of punitive damages against Sun-ridge and (2) the constitutionality of the punitive damages assessed against Apartment Management Consultants (AMC). In my view, the evidence was sufficient for the jury to assess punitive damages against Sunridge. As to AMC, I agree with the majority that the punitive-damages award was so large that it resulted in a denial of due process. But I believe the majority reduces the punitive-damages award too far below the constitutional limit; Rather than order a remittitur of $1.95 million, I would reduce the amount of punitive damages to be assessed against AMC to $7.8 million, four times 'the amount of compensatory damages.
I,The jury had sufficient evidence to assess punitive damages against Sunridge. i .
In.considering the sufficiency of the evidence, we engage in de novo review. Shaw v. AAA Eng’g & Drafting, Inc., 213 F.3d 519, 529 (10th Cir.2000). For this review, we consider the evidence and reasonable inferences in the light most favorable to the plaintiff, Ms. Lompe. See Sanjuan v. IBP, Inc., 160 F.3d 1291, 1298 (10th Cir.1998).
Viewing the evidence in this light, we must determine if a rational jury could regard Sunridge’s misconduct as willful and wanton. Weaver v. Mitchell, 715 P.2d 1361, 1370 (Wyo.1986). Sunridge’s misconduct would be “willful and wanton” if Sun-ridge intentionally failed to replace the furnaces “in reckless disregard of the consequences and under circumstances and conditions that a reasonable person would know, or have reason to know that such conduct would, in a high degree of probability, result in harm to another.” Id.
The evidence of'Sunridge’s willful and wanton misconduct included
1. Sunridge’s receipt of an inspection report upon purchasing the apartment complex, which recommended that Sunridge replace the furnaces,
2. Sunridge’s repeated failures to act on contractors’ bids to replace the furnaces, ■
3. Sunridge’s inaction in the face of reports of other incidents involving carbon monoxide leaks, and
4. extraordinary health risks from Sun-ridgé’s failure to replace the furnaces.
Together, these four forms of evidence would have allowed the trier of fact to find willful and wanton misconduct on the part of Sunridge.
*1077Before buying the apartment complex, Sunridge obtained a property assessment from LandAmerica Assessment Corporation. LandAmerica informed Sunridge that most of the furnaces appeared to be the complex’s original units and were in fair condition. Appellants’ App’x at 1980. According to LandAmerica, the furnaces had an expected life of 25 years and were already roughly 28 yéars old. Id. at 1989; see also id. at 709-10 (expert testimony that “[t]he generally accepted useful life for a furnace within the industry is 20 years”). As a result, LandAmerica recommended that Sunridge set aside $76,800 in capital reserves to replace the furnaces within five years. Id. at 1989; see also id. at 709, 714 (expert testimony about the LandAmerica report).
Yet three years after buying the complex, Sunridge had not authorized replacement of any furnaces. Thus, the jury could reasonably infer that when Ms. Lompe’s furnace leaked, it was already about six years past its expected life. Because all of the furnaces were overdue for replacement, the jury could also reasonably infer that Sunridge had known that the risk of a carbon monoxide leak increased with each passing year.1
Ms. Lompe also presented evidence that Sunridge had obtained contractors’ bids in 2007 or 2008 to replace all of the furnaces. Id. at 640. But Sunridge did not act on the bids, conduct preventative maintenance on the furnaces, or otherwise attempt to make the furnaces safé. Id. at 640, 1213.
Along with this evidence of inaction, Ms. Lompe presented testimony that Sunridge had known about prior furnace malfunctions. For example, a furnace leaked carbon monoxide in the apartment of the complex’s manager. Ms. Lompe argued that Sunridge had been aware of this leak, presenting testimony that “[djepending on the situation,”- an incident report would have been sent to Sunridge. Id. at 1546. In addition, one of Sunridge’s two owners acknowledged that he would have expected to be told if a tenant evacuated because of a: carbon monoxide leak. Id. at 1549, 1577-78.
The majority notes that there was no direct -evidence of Sunridge’s knowledge. Maj. Op. at 1048. But direct evidence was unnecessary because the -jury could reasonably-rely-on two types of indirect evidence demonstrating- that Sunridge had failed to act'after learning of furnace leaks.
First, the jury could rely on the property manager’s routine practice of forwarding inspection reports to Sunridge. See Fed.R.Evid. 406 (stating that evidence of an organization’s routine practice can be admitted to prove that1 the organization acted in accordance with that routine practice on a particular occasion). This type of evidence does not require corroboration or testimony that the property manager carried out the routine practice on a particular occasion. Id.; see also Hancock v. Am. Tel. & Tel. Co., 701 F.3d 1248, 1261-62 (10th Cir.2012) (stating that courts may accept “Rule- 406 evidence” for an “inference that a routine practice was actually carried out”).
Second, AMC’s regional property manager testified" that Sunridge had received information about the furnace leak in Ms. Lompe’s apartment. Appellants’ Ápp’x at 1546. Nonetheless, according to the regional property manager, Sunridge failed to - do anything to replace the furnaces even after learning of the leak in Ms. *1078Lompe’s furnace. Id. at 1932. Based on this testimony, ■ the jury could reasonably infer that Sunridge had failed to take action even after learning of a furnace leak that injured a tenant.
If the jury had drawn this inference, as permitted, the jury could have reasonably found Sunridge’s inaction to be willful and wanton in light of the gravity of the risk. Carbon monoxide can be deadly. See Id. at 1531 (AMC’s regional property manager acknowledging this risk); see also Hanlon v. Lane, 98 Ohio App.3d 148, 648 N.E.2d 26, 30 (1994) (holding that “the danger of carbon monoxide poisoning from the use of an improperly vented gas furnace is an open and obvious danger as a matter of law”). Thus, as the majority notes, the leakage of carbon monoxide risked death for every tenant in the complex. Maj. Op. at 1066.
The jury could rationally have found that Sunridge’s inaction was not willful or wanton. But the jury could also rationally credit Ms. Lompe’s version of events and infer that Sunridge
*had known when buying the apartment complex: that the furnaces had been operating for so long that they were dangerous to apartment residents,
• had been reminded of the need to replace the furnaces when one of them leaked carbon monoxide in the manager’s apartment, and
• had failed to take corrective action even after learning of multiple furnace leaks.
With these inferences from the evidence, the jury could reasonably have found that Sunridge’s misconduct was willful and wanton. Accordingly, I respectfully dissent from the majority’s contrary conclusion in Part IV(A)(l)(a).2
II. The majority’s constitutional cap on punitive damages to be assessed against AMC is too low.
The majority concludes that the assessment of $22.5 million in punitive damages against AMC constitutes a deprivation of substantive .due process. I agree that this assessment was so large that it denied AMC due process and that the district court applied the incorrect standard when reviewing the jury’s assessment. But I believe that the majority overcorrected the assessment given the factual findings made by the district court and the jury.
The jury found that AMC’s conduct had constituted “willful and wanton misconduct,” ' a finding the majority upholds. And in upholding this finding, the majority acknowledges that AMC intentionally failed to inspect the furnaces, recklessly risking the life of every tenant in the' complex. But then the majority engages in its own fact finding, downplaying AMC’s culpability based on' the majority’s own weighing of the evidence;
Instead, I would usé the district court’s and jury’s fact findings to determine the constitutional limit for assessing punitive damages against AMC. In my view, the district court’s and jury’s findings would allow an assessment of punitive damages up to $7.8 million, four times the compensatory-damages award of $1.95 million.
A. I would apply the district court’s and the jury’s factual findings to evaluate the constitutionality of the punitive damages assessed against AMC.
As the majority explains, the Supreme Court has established three “guideposts” bearing on our constitutional inquiry:
*10791. the degree of reprehensibility of the defendant’s misconduct,
2. the disparity between the actual or potential harm suffered by the plaintiff and the punitive-damages award, and
3. the difference between the punitive damages assessed by the jury and civil penalties authorized or imposed in comparable cases.
Maj. Op. at 1063 (quoting State Farm Mut. Auto. Ins. Co. v. Campbell, 538 U.S. 408, 418, 123 S.Ct. 1513 (2003)); see also BMW of N. Am., Inc. v. Gore, 517 U.S. 559, 574-75, 116 S.Ct. 1589 (1996) (setting out these guideposts). Like the majority, I would apply these three guideposts to determine the constitutionality of the punitive damages assessed against AMC. Unlike the majority, however, I would apply these three guideposts based on the factual findings made by the district court and the jury.
The majority is correct in stating that we apply de novo review to the ultimate issue: whether the amount of the punitive damages is so large that it violates AMC’s right to substantive due process. Maj. Op. at 1061-62. But the three-part' test is laden with factual issues, such as what the defendant did to merit punitive damages and the extent of the plaintiffs injury. Under Supreme Court precedent, we can disturb the resolution of these factual issues only if the district court’s factual findings are clearly erroneous.
In applying more intensive scrutiny to these factual matters, the majority relies on Cooper Industries, Inc. v. Leatherman Tool Group, Inc., 532 U.S. 424, 121 S.Ct. 1678 (2001). Maj. Op. at 1061-62, 1067 n. 21. But there, the Supreme Court distinguished between review of the ultimate constitutional issue (where de novo review is required) and the underlying factual matters bearing on the constitutional issue (where clear error - review is required): “While we have determined that the Court of Appeals must review the District Court’s application of the Gore test de novo, it of course remains true that the Court of Appeals should defer to the District Court’s findings of fact unless they are clearly erroneous.” Cooper Indus., 532 U.S. at 440 n. 14, 121 S.Ct. 1678. The Supreme-Court cautioned that even though the ultimate constitutional issue requires de novo review, that does not mean that the jury’s findings are also subject to de noyo review. Id. To the contrary, the Court noted that its requirement of de novo review on the ultimate constitutional issue does not permit a reviewing court to disregard the jury’s factual findings. See id. at 439 n. 12, 121 S.Ct. 1678 (“[Nothing in our decision today suggests that the Seventh Amendment would permit a court, in reviewing a punitive damages award, to disregard such jury findings.”).
In light of the Supreme Court’s observations, I would give due regard to the factual findings expressly made by the district court and the factual findings implicitly made by the jury. See Rhone-Poulenc Agro, S.A. v. DeKalb Genetics Corp., 272 F.3d-1335, 1347 (Fed.Cir.2001) (interpreting the quoted language in Cooper Industries “to mean that if a-punitive damages determination rests on purely factual issues, we are to assume that those factual issues have been resolved adversely to the defendant, absent contrary indication”), vacated sub nom. DeKalb Genetics Corp. v. Bayer CropScience, S.A., 538 U.S. 974, 123 S.Ct. 1828, 155 L.Ed.2d 662, reinstated sub nom. Rhone-Poulenc Agro, S.A. v. DeKalb Genetics Corp., 345 F.3d 1366 (Fed.Cir.2003); Leatherman Tool Grp., Inc. v. Cooper Indus., Inc., 285 F.3d 1146, 1150 (9th Cir.2002) (stating, after the Supreme Court’s remand, that “despite [the Supreme Court’s] holding that we must review the district court’s application of the Gore test de novo, we still must ‘defer *1080to the district court’s findings of fact unless they are clearly erroneous’ ” (quoting Cooper Indus., Inc., 532 U.S. at 440 n. 14, 121 S.Ct. 1678)).
B. The majority’s, constitutional analysis does not reflect the factual findings made by the district court and jury on matters affecting the reprehensibility of AMC’s conduct.
In light of the factual findings by the district court and the jury, I respectfully disagreé with the way that the majority assesses reprehensibility. In my view, the majority should have deferred to all of the district court’s findings and assumed that the jury had resolved evidentiary conflicts against AMC; instead, the majority makes new factual findings, seeing mitigating evidence where the jury presumably did not. I do not believe that this independent fact finding was what the Supreme Court had in mind in Cooper Industries.
1. Reprehensibility is the most important of the three guideposts, triggering a need to defer to the district court’s and jury’s factual findings.
Of the three guideposts, reprehensibility is the most important. See State Farm Mut. Auto. Ins. Co. v. Campbell, 538 U.S. 408, 419, 123 S.Ct. 1513 (2003). Under this guidepost, the factual findings by the district court and the jury take on special importance. See Haynes v. Stephenson, 588 F.3d 1152, 1157 (8th Cir.2009) (distinguishing the legal conclusion of reprehensibility from the findings of fact required to identify conduct that demonstrates reprehensibility); Clark v. Chrysler Corp., 436 F.3d 594, 601 n. 7 (6th Cir.2006) (explaining an appellate court’s deference to the district court’s factual findings regarding the reprehensibility guidepost); Leatherman Tool Grp., Inc. v. Cooper Indus., Inc., 285 F.3d 1146, 1150 (9th Cir.2002) (observing that in “determining the ‘degree of reprehensibility,”’ the reviewing court “must accept the underlying facts as found by the jury and the district court”).
2: The reprehensibility of a party’s conduct involves a continuum of blameworthiness.
The term “reprehensibility” can refer to varying degrees of culpability. Properly understood, the level of reprehensibility involves a continuum rather than a precise point. See BMW of N. Am., Inc. v. Gore, 517 U.S. 559, 575, 116 S.Ct. 1589, 134 L.Ed.2d 809 (1996) (observing that “some wrongs ,are more blameworthy than others”); see also Mendez v. Cty. of San Bernardino, 540 F.3d 1109, 1120 (9th Cir.2008) (“Reprehensibility falls along a scale.”), overruled on other grounds by Arizona v. ASARCO LLC, 773 F.3d 1050 (9th Cir.2014); Asa-Brandt, Inc. v. ADM Inv’r Servs., Inc., 344 F.3d 738, 748 (8th Cir.2003) (explaining that dore established a “hierarchy of reprehensibility” for evaluating the constitutionality of a punitive-damages award); Laura J. Hines & N. William Hines, Constitutional Constraints on Punitive Damages: Clarity, Consistency, & the Outlier Dilemma, 66 Hastings L.J. 1257, 1303 (2015) (“Today, courts routinely rank the degree of,the defendant’s reprehensibility. on a wide spectrum that runs from very little reprehensibility, through modest or intermediate reprehensibility, and up to substantial or extreme reprehensibility.” (footnotes omitted)). For example, it may be reprehensible to steal from someone. But it is even more reprehensible to kill the victim afterward.
We must decide where AMC’s conduct fell along the-continuum of reprehensibility, ranging from a mere accident to intentional ill will. Gore, 517 U.S. at 575-76, 116 S.Ct. 1589. But we are confined to a *1081cold record because we were not' at the trial, as the district court and jury were. Thus, the district court had-“a somewhat superior vantage” point to assess reprehensibility. Cooper Indus., Inc. v. Leatherman Tool Grp., Inc., 532 U.S. 424, 440, 121 S.Ct. 1678 (2001).
The majority acknowledges that the district court’s and jury’s findings are subject to dear-error review. But without finding clear error, the majority downplays certain findings and supplements them- based on the majority’s own assessment of mitigating evidence.
3. The jury found that AMC had acted with willful and wanton misconduct, satisfying the high threshold of culpability required by Wyoming law for any award of punitive damages.
The jury was asked to decide whether AMC’s conduct satisfied Wyoming law’s threshold for punitive damages. As the majority notes, that threshold requires a rigorous showing that the defendant’s conduct was especially blameworthy: punitive damages can be assessed only if the defendant’s conduct is “outrageous” because it is intentional, malicious, or involves willful and wanton misconduct. Maj. Op. at 1047-48. The majority further explains that Wyoming considers conduct “willful and wanton” only if it “approaches intent to do harm.” Id. (quoting Cramer v. Powder River Coal, LLC, 204 P.3d 974, 979 (Wyo.2009)). Thus, the district court instructed the jury that it could assess punitive damages only if AMC had intentionally acted or declined to act “in reckless disregard of the consequences, and under such circumstances and conditions that a reasonable person would know, or have reason to know, that such conduct would, with a high degree of probability, result in harm to another.” Appellants’ App’x at 128. The jury specifically found .that AMC’s misconduct satisfied this rigorous threshold of blameworthiness, and the majority concludes that this finding was sufficiently supported by the evidence. Id. at 158; Maj. Op. at 1057.
Because I agree that this finding is supported by the evidence, I would consider AMC’s misconduct as “willful and wanton” when assessing the reprehensibility guidepost. See Cabral v. U.S. Dep’t of Justice, 587 F.3d 13, 26 (1st Cir.2009) (noting that the jury’s finding supported the reprehensibility guidepost).
The majority points out that the amount of punitive damages that the jury chose to award is not a factual finding. Maj. Op. at 1061-62, 1063-64; see also Cooper Indus., 532 U.S. at 437, 121 S.Ct. 1678 (“[T]he level of punitive damages is not really a ‘fact’ ‘tried’ by the jury.” (quoting Gasperini v. Ctr. for Humanities, Inc., 518 U.S. 415, 459, 116 S.Ct. 2211, 135 L.Ed.2d 659 (1996) (Scalia, J., dissenting))). But the jury did make a factual finding when it determined that AMC’s alleged misconduct was so blameworthy that it met Wyoming’s high threshold for punitive damages. The majority concludes that this finding was supported by the record. Nevertheless, the majority then disregards.this finding when evaluating the reprehensibility of AMC’s conduct. See Maj, Op. at 1064 (“[T]he jury’s verdict in the liability phase of trial does not inform our constitutional analysis — [T]he. jury here made no findings that are relevant to our application of the Gore factor^.”).
The majority seems to suggest that we cannot defer to a jury’s findings in the absence of special interrogatories. See id. at 1063 (noting that the jury was not asked to answer -special interrogatories). I respectfully disagree. In my view, we should assuhié that the jury resolved evi-dentiary conflicts adversely to ,AMC unless the record indicates otherwise. See United States v. Dowell, 430 F.3d 1100, 1110 *1082(10th Cir.2005) (explaining that' a jury’s verdict “encompassed ... factual findings” embedded in the statutory provision that the jury applied); Ag Servs. of Am., Inc. v. Nielsen, 231 F.3d 726, 733 (10th Cir.2000) (“[W]e ... find necessary inferences from the verdict indicating that certain views of the evidence were not taken by the jury as they could not have rationally supported the result.” (emphasis in original)).
Because the jury found willful and wanton misconduct, the jury must have regarded AMC’s.misconduct as “something in the nature of special ill-will” that constituted “oppression, fraud or malice.” Farmers Ins. Exch. v. Shirley, 958 P.2d 1040, 1051 (Wyo.1998) (quoting McCullough v. Golden Rule Ins. Co., 789 P.2d 855, 861 (Wyo.1990)).
4. The district court’s factual findings render AMC’s conduct particularly reprehensible.
Even if we were to disregard the jury’s implicit findings, we must still defer to the district court’s factual findings.' See CGB Occupational Therapy, Inc. v. RHA Health Servs., Inc., 499 F.3d 184, 189 (3d Cir.2007). After hearing the evidence, the district court denied- AMC’s and Sun-ridge’s renewed motion for judgment as a matter of law dr, in’the alternative, for new trial or remittitur. Appellants’ App’x at 287-314. In doing so, the district court made four factual findings bearing on the reprehensibility of AMC’s conduct:
1. The harm to Ms. Lompe was physical, not merely economic.3 .
2. “[I]f [Ms. Lompe] had been in her -apartment even minutes longer, she likely would have lost consciousness and the carbon monoxide exposure could have disabled her for the rest of her life or could, have been fatal.”
3. “The potential for similar injuries to ■ others was great where the defendants failed to correct known problems with furnaces in the apartment complex, exposing other tenants to carbon monoxide poisoning as well.”
4. “This was not an isolated incident.”
Id. at 309, 313. Neither AMC nor the majority suggests that the district court clearly erred in making these factual findings.
These four findings reflect particularly reprehensible conduct by AMC. In Bielicki v. Terminix International Co., we recognized that a defendant acts reprehensibly when operating in an “atmosphere of condoning the disregard of safety practices” that could result in .physical injuries to others. 225 F.3d 1159, 1163, 1165 (10th Cir.2000); see also Willow Inn, Inc. v. Pub. Serv. Mut. Ins. Co., 399 F.3d 224, 233 (3d Cir.2005) (concluding that “a mix of purposefully indifferent inaction and intentionally dilatory action” is reprehensible). The jury presumably found that AMC had disregarded safety practices, for even the majority acknowledges that AMC had recklessly failed to conduct regular furnace inspections, risking the life of every tenant in the apartment complex. Maj. Op. at 1066., Under Bielicki, AMC’s reckless disregard for safety practices is particularly reprehensible.
5. The majority improperly relies on mitigating evidence to find that AMC’s conduct was not particularly reprehensible.
The majority downplays AMC’s culpability based on evidence that the majority *1083sees as mitigating. In my view, this approach elevates our fact-finding role over the jury’s, effectively undermining the jury’s role in resolving evidentiary conflicts. ,
Though we exercise de novo review on the ultimate question of constitutionality, we are not fact finders. Nonetheless, the majority independently evaluates the evidence, finding in the first instance that AMC’s conduct was not “particularly reprehensible” based on what the majority regards as mitigating evidence:
• AMC was not cited by regulators for the prior incidents of poisoning,
• AMC voluntarily provided each tenant with a carbon monoxide detector, and
• AMC purportedly conformed to industry standards.
Maj. Op. at 1066-67. But-faced with this same evidence, the jury regarded AMC’s misconduct as willful and wanton. Appellants’ App’x at 128, 163, 168. The district court then concluded that the jury must have “viewed the defendants’ conduct as egregious.” Id. at 314. In light of the jury’s apparent conclusions from conflicting evidence, I believe that we should regard AMC’s conduct as particularly reprehensible' for purposes of the first guidepost.
In my view, even the majority’s fresh look at the evidence would support a large assessment of punitive damages. For example, while regarding some of the evidence as mitigating, the majority concludes that Ms. Lompe’s injury resulted from repeated actions or failures to act, rather than an isolated incident:
[Tjhree prior [carbon monoxide] leaks, including one in which [the apartment manager] was poisoned, put AMC on notice that the furnaces needed regular substantive safety inspections.to prevent future dangerous releases of [carbon monoxidé]. Failing to perform such inspections recklessly endangered not only Ms. Lompe, but also all the tenants at the apartment complex, and the risk included death.
Maj. Op. at 1066.
Though the majority finds that AMC’s “reckless[ ]” failure to conduct inspections endangered the lives of every tenant in the apartment complex, the majority concludes that AMC’s conduct was not particularly reprehensible in part because it did not involve “intentional malice, trickery, or deceit.” Id. at 1067 (quoting Jones v. UPS, Inc., 674 F.3d 1187, 1207 (10th Cir.2012)). I respectfully think that this characterization * understates- the reprehensibility of AMCs conduct and displaces the jury’s finding óf willful and wanton misconduct.
In my view, AMC’s conduct was worse than trickery or deceit — AMC recklessly risked the life of every tenant in the entire apartment complex. See Mendez v. Cty. of San Bernardino, 540 F.3d 1109, 1120 (9th Cir.2008) (stating that on a scale of reprehensibility, reckless disregard for others’ health and safety is even more egregious than affirmative acts of trickery and deceit), overruled on other grounds, by Arizona v. ASARCO LLC, 773 F.3d 1050, 1058 n. 1 (9th Cir.2014); Action Marine, Inc. v. Cont’l Carbon Inc., 481 F.3d 1302, 1320 (11th Cir.2007) (noting that, though “punitive damages, may not be awarded to punish for harm inflicted on nonparties, [the court] may consider the risk of harm to others as part .of the reprehensibility analysis” and the fact that the defendant’s conduct endangered “a great number of people” who were not parties to the litigation rendered the defendant’s conduct “exceedingly reprehensible” (citing Philip Morris U.S.A. v. Williams, 549 U.S. 346, 127 S.Ct 1057, 166 L.Ed.2d 940 (2007))); see also Blaine LeCesne, Crude Decisions: Re-Examining Degrees of Negligence in the Context of the BP Oil Spill, 2012 Mich. *1084St. L.Rev. 103, 126 (“It is this focus on the design or purpose in the actor’s state of mind which places willful misconduct closer to the intentional conduct end of the spectrum, while gross negligence is nearer to the ordinary negligence end of that spectrum.”).
The majority correctly states that willful and wanton misconduct is not as reprehensible as “intentional malice.” Maj. Op. at 1067 (quoting Jones v. UPS, Inc., 674 F.3d 1187, 1207 (10th Cir,2012)). But this is not saying much. The majority elsewhere holds that under Wyoming law, “[a] .willful or wanton state of mind is one.ithat approaches intent to do harm.’ ” Id. at 1055 (quoting Cramer v. Powder River Coal, LLC, 204 P.3d 974, 979 (Wyo.2009)), Applying this test, the majority concluded that the jury had sufficient evidence to find that AMC had a state of mind approaching an intent to do harm. Id. at 1057. I would regard that state of mind as particularly reprehensible even if it did not involve intentional malice.
But I respectfully disagree with more than the majority’s placement of AMC’s misconduct on the repréhensibility continuum. In my view, the majority should not find mitigation in the first instance when the jury considered the same evidence and regarded the misconduct as willful and wanton. • ■ - '
For example, the majority states that AMC was not cited by regulators for the prior leaks of carbon monoxide. Id. at; 1066. But the apartment complex’s'sole maintenance employee from 2009 to 2012 testified that no one had ever inspected any of the furnaces in the complex. Appellants’ App’x at 1214. Without evidence that an inspection took place, the absence of a regulatory citation is neither surprising nor meaningful.
Regardless of whether a formal citation was issued, the district court stated that after the apartment manager was poisoned, the local gas utility had told AMC that the furnaces should be inspected. Id. at 291; see also id. at 655, 761. AMC apparently ignored that recommendation. Even- the majority acknowledges that AMC knew that its furnaces had outlived their expected duration and had leaked carbon monoxide on multiple occasions, endangering.’all the tenants. Maj. Op. at 1057. As a result, I question the majority’s reliance on the absence of a citation by the gas utility.4
The majority also points to evidence that carbon monoxide detectors were provided to the tenants. Id., at 1066, But the evidence also showed that (1) after Ms. Lompe’s furnace leaked carbon monoxide, approximately half of, the apartments lacked a working carbon monoxide detector, (2) the tenants with carbon monoxide detectors were never told anything about the detectors, and (3) the distribution of carbon monoxide detectors was not a substitute for proper maintenance of the furnaces. Appellants’ App’x at 657, 667—68, 674-76, 1127-28, 1204, 1220, 1231, 1256 (lack of working carbon monoxide detectors); id. at 1127-28, 1143, 1204 (tenants who had detectors were never told about *1085them); id. at 767-68 (detectors not substitutes for furnace maintenance).
AMC’s own maintenance worker testified that he had expressed concern to AMC about the lack of carbon monoxide detectors in all of the units and the need for hardwired carbon monoxide detectors. Id. at 1256-57. AMC replied to this concern by saying that hardwired detectors were “[n]ot in the budget.” Id. at 1257.
How are we to balance AMC’s provision of carbon monoxide detectors against evidence of the detectors’ ineffectiveness? And how does AMC’s provision of the detectors factor into AMC’s overarching misconduct? It is difficult, if not impossible, to resolve these questions on the basis of a cold record. Having heard all the evidence, mitigating and aggravating, the jury regarded AMC’s misconduct as willful and wanton. I would not lightly disregard this assessment based on our own characterization of AMC’s distribution of detectors as mitigating.
Finally, the majority says that there was some evidence that AMC “conformed to industry standards in dealing with the aging fleet of furnaces.” Maj. Op. at 1066. But there was also evidence to the contrary. For example, an. expert witness testified that AMC had violated the Wyoming fire safety code by failing to maintain a safé heating system. Appellants’ App’x at 708. The expert witness also testified that the industry standard required .inspections and maintenance of the furnaces. Id. at 703-04. And Ms. Lompe presented evidence that no one had ever inspected the furnaces. Id. at. 1214. The jury evaluated this conflicting evidence and found that AMC had engaged in willful and wanton misconduct. To-me, this finding requires us to consider AMC’s misconduct as particularly reprehensible, for resolution of the evidentiary conflicts was for the jurors, not for us on the basis of a cold record.
C. The majority puts undue weight on the fact that the jury awarded a substantial amount of compen- ' satory damages.
We must also consider the ratio of punitive to compensatory damages. As the majority notes, the Supreme Court has provided only general guidelines for the constitutionally permissible ratio. Maj. Op. at 1067-68 & n. 24. For example, the Supreme Court stated that (1) a 4-to-1 damages ratio might be close to the constitutional boundary and (2) a ratio significantly exceeding single digits might cross the constitutional line. State Farm, Mut Auto. Ins. Co. v. Campbell, 538 U.S. 408, 425, 123 S.Ct. 1513 (2003); see also Clark v. Chrysler Corp., 436 F.3d 594, 620 (6th Cir.2006) (Moore, J., concurring in. part & dissenting in part) (“[T]he Court has never explicitly said that a 4:1 ratio actually is close to the. constitutional line, just that it might be.” (emphasis in original)); Mathias v. Accor Econ. Lodging, Inc., 347 F.3d 672, 676 (7th Cir.2003) (“The Supreme Court did not ... lay down a 4-to-1 or single-digit-ratio rule — it said merely that ‘there is a presumption against an award that has a 145-to-l ratio;’ ” (quoting State Farm, 538 U.S. at 425, 123 S.Ct. 1513)). The Supreme Court added that even a 1-to-1 ratio might represent the constitutional limit when the award of compensatory damages is substantial.-State Farm, 538 NS. at 425, 123 S.Ct. 1513.
Despite these conflicting and ambiguous signals from the Supreme Court, the majority finds a pattern among the circuit court cases. Some courts have held that when a defendant’s conduct is not particularly, reprehensible and the compensatory damages are substantial, a 1-to-l ratio represents the constitutional boundary. See Maj. Op. at 1073-76 (citing cases). Other courts have held that even when the compensatory damages are substantial, the constitutionally permissible ratio can far *1086exceed 1-to-1. See, e.g., Brand Mktg. Grp, LLC v. Intertek Testing Servs., N.A., Inc,, 801 F.3d 347, 366 (3d Cir.2015) (concluding that “a 5:1 ratio is not the type of gross disparity between compensatory and punitive damages that renders a punitive award suspect by itself’ for a compensatory-damages award of $1,045 million); Gibson v. Moskowitz, 523 F.3d 657, 665 (6th Cir.2008) (“[T]he 2 to 1 ratio ... falls well short of the high end of this range and indeed parallels the kind of relationship that the Court has said will often suffice [for a compensatory-damages award of $1.5 million].”); Planned Parenthood of the Columbia/Willamette Inc. v. Am. Coal. of Life Activists, 422 F.3d 949, 962 (9th Cir.2005) (stating that when the award of economic damages is significant, a 4-to-l ratio is a “good proxy for the limits of constitutionality” if the “behavior is not particularly egregious”); Diesel Mach., Inc. v. B.R. Lee Indus., Inc., 418 F.3d 820, 840 (8th Cir.2005) (rejecting an argument that “a four-to-one ratio is per se constitutional” when the plaintiff was awarded $665,000 in compensatory damages)5; Hangarter v. Provident Life & Accident Ins. Co., 373 F.3d 998, 1015 (9th Cir.2004) (concluding that a 2.6-to-l ratio is "well within the Supreme Court’s suggested range for constitutional punitive damages awards” for a compensatory-damages award of $1.92 million); see also Laura J. Hines & N. William Hines, Constitutional Constraints on Punitive Damages: Clarity, Consistency, & the Outlier Dilemma, 66 Hastings L.J. 1257, 1302, 1314-15 (2015) (studying 507 published opinions and concluding that when courts characterize the compensatory awards as substantial, “the majority do not appear to feel particularly bound by a 1:1 ratio”).6
How should we reconcile these conflicting signals from the Supreme Court and *1087the variety of approaches in other circuits? In my view, there is only one way: the ratios must be assessed on a case-by-case basis because the reprehensibility of the defendant’s conduct and the, size of a compensatory-damages award involve con-tinuums rather than precise points. For example, an award of $300 million in compensatory. damages would strike anyone as substantial. But so would an award of $3 million. -Nonetheless, application of the same ratio would result in far different assessments of punitive damages. For example, a 4-to-l ratio would result in
• a $1.2 billion assessment of punitive damages if the compensatory damages were $300 million, and
• “only” a $12 million assessment of punitive damages if the compensatory damages were $3 million.
Accordingly, the Supreme Court has declined to impose a bright-line ratio and has cautioned reviewing courts to assess the permissible ratio in light of the compensatory-damages award. See State Farm Mut. Auto. Ins. Co. v. Campbell, 538 U.S. 408, 425-26, 123 S.Ct. 1513 (2003).
This caution reflects the Supreme Court’s recognition that when the compensatory damages are extraordinarily large, any multiplier above 1 might render the punitive-damages award excessive. Id. at 425, 123 S.Ct. 1513. Consider the $300 million hypothetical above: even a 1-to-l ratio would yield punitive damages of $300 million. Thus, any multiplier greater than 1 might be overly severe and violate the defendant’s right to due. process even if the defendant’s conduct is particularly reprehensible.
> Here, the compensatory damages were $1,950,000. Like the majority, I regard this amount as substantial. But the Supreme Court has not suggested a hard- and-fast rule limiting punitive damage awards to a 1-to-l ratio whenever the compensatory damages are considered substantial.7
Against this backdrop, I would regard the- substantiality of the award as a continuum, much like the continuum on reprehensibility. Just as the term “reprehensibility” describes various degrees of culpability, the term “substantial” describes many different dollar amounts. Compare Clark v. Chrysler Corp., 436 F.3d 594, 607 (6th Cir.2006) (noting that a $236,000 compensatory-damages award was “not particularly large”), and Flax v. DaimlerChrysler Corp., 272 S.W.3d 521, 539 (Tenn.2008), (“Although the United States Supreme Court has made no effort to demonstrate when damages are ‘substantial,’ we do not believe that an award of $2,500,000 is so large as to require a ratio of 1 to 1.”), with Bains LLC v. Arco Prods. Co., 405 F.3d 764, 776 (9th Cir.2005) (stating that $50,000 in economic damages was “substantial”).
Ms. Lompe’s compensatory-damages award of $1,950,000 was large, and I would consider that fact along with the other two State Farm guideposts articulated in Gore. Gore, 517 U.S. at 575-76, 116 S.Ct. 1589 (identifying two other guideposts: reprehensibility and ■ the “difference between *1088this remedy and the civil penalties authorized or imposed in comparable cases”). The inquiry under State Farm does not end with characterization of the compensatory damages as “substantial.” We must also consider the other two guideposts to determine whether the amount assessed in punitive damages is constitutionally permissible. ■
D. The third factor, involving the disparity with penalties in comparable cases, is inapplicable here.
The third State Farm guidepost directs us to consider the disparity between 'the punitive-damages award in this cáse and civil penalties authorized or imposed in comparable cases. State Farm Mut. Auto. Ins. Co. v. Campbell, 538 U.S. 408, 418, 123 S.Ct. 1513 (2003). In my view, this guidepost does not help us evaluate the constitutionality of the punitive damages to be assessed against AMC. See CGB Occupational Therapy, Inc. v. RHA Health Servs., Inc., 499 F.3d 184, 189-90 (3d Cir.2007) (limiting the constitutional analysis to the first two guideposts in the absencó of relevant, comparable civil penalties).
The majority acknowledges that the penalties associated with “violations of common law tort duties often do not ‘lend themselves to a comparison with statutory penalties.’” Maj. Op. at 1070 (quoting Cont’l Trend Res., Inc. v. OXY USA Inc., 101 F.3d 634, 641 (10th Cir.1996)). Nonetheless, the majority concludes that this guidepost “reinforces” a substantial reduction in punitive damages because
• many, other states lack civil penalties , for failure to maintain heating equipment, and-
• other state courts have imposed puni- ■ tive damages much less than $22.5 million.
Id. at 1073.
For the sake of argument, we may assume that this guidepost supports a sub-stantia! reduction in the assessment of punitive damages. But I do not believe the majority’s analysis requires us to remit the punitive damages to be assessed against AMC to $1.95 million.
The majority identifies several state statutes that require maintenance of heating systems, pointing out that they do not authorize civil penalties.- Id. at 1071-72 & nn. 33-32. The majority views this factor as reinforcing its decision to cap the punitive damages to be assessed against AMC at $1.95 million. Id. at'1076. But, as.the majority notes, no state authorizes civil penalties for failure to properly maintain furnaces. As a result, this factor is neutral; it does not reinforce anything.
In addition, the majority cites three cases in two states (Alabama and Missouri) between 1991 to 1993 for the proposition that courts “have generally awarded much smaller punitive damages than those awarded here.” Id. at 1073-74 & n. 36. Again, I do not believe the cited cases help us to decide the amount of the remittitur. Each of these cases was decided before the Supreme Court decided BMW of North America, Inc. v. Gore, 517 U.S. 559, 116 S.Ct. 1589 (1996), which fundamentally changed the nature of our inquiry into the constitutionality of punitive damage assessments. The majority- seems to acknowledge this fact, but still holds that these cases should have put AMC on notice of the potential degree of punitive damages. See Maj. Op. at 1072-73. I respectfully disagree.
First, in Sears, Roebuck and Co. v. Harris, the Alabama Supreme Court upheld a $2.5 million punitive-damages award after reducing the compensatory damages to $350,001, yielding a constitutionally permissible ratio of oyer 7.1 to 1 — several degrees higher than my proposed 4-to-l ratio and magnitudes above the majority’s *10891 — to—1 ratio. 630 So.2d 1018, 1034-35 (Ala.1993).8
Second, in Killough v. Jahandarfard, the Alabama Supreme Court upheld an assessment of $2.5 million in punitive damages in a wrongful-death action because (1) the defendant had failed to equip the tenant’s home with a smoke detector and (2) a fire resulted in a child’s death from carbon monoxide poisoning. 578 So.2d 1041, 1042 (Ala.1991). But we do not know anything about the defendant’s conduct in maintaining the house or when the fire took place. We know only when the suit was filed: over 27 years ago. See id. at 1042. Even then, the state supreme court upheld an assessment of $2.5 million in punitive damages.
For our inquiry on punitive damages, these cases tell us -only that (1) more than 25 years ago, two Alabama juries awarded $2.5 million in punitive damages in cases involving carbon monoxide, and (2) in one of those cases, the ratio of punitive to compensatory damages was over 7.1-to-l. I do not think these Alabama cases could have put AMC on notice that any punitive-damages award would be limited to the extent of actual harm caused to its tenants (either to $1.95 million or any other amount).
In the third case that the majority cites, Kilmer v. Browning, a Missouri appellate court affirmed a $300,000 award based on “aggravating circumstances” that could be considered under a Missouri statute. 806 S.W.2d 75, 77, 80 (Mo.Ct.App.1991). But we do not know how much of this award was for compensatory damages versus punitive damages — we know only that “it was for the jury to determine if aggravating circumstances were present.” Id.
In my view, we cannot determine the constitutional limit for punitive damages to be assessed against AMC based on
• three cases from Alabama and Missouri that were decided about 25 years ago and
• a number of state laws that lack any civil penalties.
* ❖
I respectfully disagree with the majority’s adjusted assessment of punitive damages against AMC. The majority makes its own factual findings on AMC’s reprehensibility without applying the ' elear-error standard to the district court’s factual findings or considering the jury’s finding of willful and wanton misconduct. These findings require us to consider AMC’s conduct as particularly reprehensible — the most important of the three State Farm guideposts bearing on the constitutionality of the punitive-damages award.
I agree that the compensatory-damages award is substantial. But other courts have regularly permitted an assessment of punitive damages at a multiplier approximating four times the compensatory-damages award, even when the compensatory damages are substantial, as shown in the appendix.9
*1090. As a result, I would order a remittitur, but not one as severe as the majority’s. I would order a remittitur of $7.8 million (a 4-to-l ratio) rather than $1.95 million (a 1-to-l ratio).
III. Conclusion
I respectfully dissent from two' of the majority’s conclusions:
1. that the evidence was insufficient for an award of punitive damages against Sunridge and
2. that for AMC, the Constitution requires .a remittitur of $1.95 million in punitive damages.
Instead, I would uphold the punitive-damages award against Sunridge and order a remittitur on the punitive-damages award against AMC for $7.8 million.
Appendix: Survey of Ratios of Punitive Damages to Compensatory Damages When the Compensatory Damages Are $400,000 or More
This appendix displays the results of a survey of punitive damage awards when the underlying award of compensatory damages was at least $400,000. In each of the surveyed cases, a federal court of appeals upheld the award of punitive damages, either as a constitutionally permissible award by the jury or as a constitutionally permissible amount through re-mittitur; Each point represents a single damages award. The bold line represents a 1-to-l ratio of punitive damages to compensatory damages. Each case is labeled with a data point. The red triangle shows how my proposed ratio. of 4 to 1 would fit within the survey of awards upheld against due process challenges when the compensatoiy damages awarded are $400,000 or more.
[[Image here]]
*1091[[Image here]]
. Sunridge also argued that the assessment of punitive damages was excessive. The majority does not reach this argument with respect to Sunridge; as a result, I also decline to address this argument.
. The majority seems to collude that this finding is mitigated by the fact that the compensatory-damages award included " “both physical harm and economic loss.” Maj. Op. at 1066. But AMC concedes that the harm was physical and does not argue that the inclusion of economic damages is a mitigating factor. See Appellants’ Opening Br. at 46 (“The- harm here was physical, as opposed to economic.’’).
. The majority notes that the absence of- a .citation from the gas utility is "not incpnsis-tent” with its determination that AMC’s failure to act constituted recldess disregard of the , consequences, which would result in harm to another. .Maj. Op. at,1064-65 n. 18.1 am not suggesting otherwise, for my point is different, The majority concludes that the jury could have found that AMC acted in reckless disregard for its tenants, and the district court found that the local gas utility had. alerted AMC to the need to inspect its furnaces. Even if the gas utility did not cite AMC, AMC knew that- the furnaces were dangerous. Thus, AMC acted with reckless disregard for its tenants’ safety, which bears on the element of repréhensibility.
. The majority points to a 2012 opinion from the Eighth Circuit: Ondrisek v. Hoffman, 698 F.3d 1020 (8th Cir.2012). Maj. Op. at 1069-70. Citing Ondrisek, the majority states that the Eighth Circuit has reduced the award in three post-Gore cases when the compensatory damages were substantial. Id. (citing Ondrisek, 698 F.3d at 1029-30). But in Ondrisek, the Eighth Circuit held that the constitution would permit a 4-to-l ratio, even though compensatory damages were $3 million (over 50 percent higher than the $1.95 million assessment of compensatory damages against AMC). Ondrisek, 698 F.3d at 1030-31. In doing so, the Eighth Circuit stated that its "constitutional boundary” was a 4,8-to-l ratio "when multi-million dollar compensatory damages award[s] are involved.” Id.
In my view, the Eighth Circuit's opinion in Ondrisek supports some reduction in Ms. Lompe’s award of punitive damages, but not to a 1-to-l ratio. My proposed ratio, 4-to-l in this case, is below the Eighth Circuit’s constitutional boundary of 4.8-to-l in cases involving compensatory damages in the millions of dollars.
. In an appendix, I have surveyed cases involving compensatory-damages awards of $400,000 or more. The appendix shows that there is no discernible pattern in ratios approved in circuit courts, much less a trend toward awarding punitive damages in a 1-to-1 ratio when the compensatory damages are substantial.
The majority attempts to distinguish some of these cases in which a court approved of a ratio of damages exceeding 1-to-l, stating that the cases involved "intentionally culpable behavior ... or significantly greater actual damages to the plaintiff.” Maj. Op. at 1075 n, 38. In my view, this distinction is immaterial for two reasons. First, as the majority states, each case is fact specific. See id. at 1069 n. 28 ("[T]he disparity in holdings indicates that the maximum of punitive damages that comports with due process is highly dependent upon the relevant facts,”). The majority points to a number of cases that limit the ratio to 1-to-l. I am simply pointing to ' these cases to show that numerous courts have approved higher ratios even when the damages are substantial. Second, the majority’s distinction assumes that AMC was not particularly culpable in this case. But as discussed above, the jury’s verdict was tantamount to a finding that AMC’s misconduct *1087had "approached an intent to do harm.” See p. 1084, above.
. After the Supreme Court decided Gore, we held in Continental Trend Resources, Inc. v. OXY USA Inc. that even when the compensatory damages are more than $1 million, a roughly 6-to-l ratio can be constitutionally permissible. 101 F.3d 634, 643 (10th Cir.1996). Continental Trend Resources remains good law. The Supreme Court later,decided State Farm and noted that a 1-to-l ratio may push the constitutional bounds when the compensatory damages are substantial. See State Farm, 538 U.S. at 425, 123 S.Ct. 1513. But the Court did not set this as a hard-and-fast rule or suggest that any substantial award of compensatory damages would obviate the need to consider the other guideposts, including the most important one, the reprehensibility of the defendant's conduct. See generally State Farm, 538 U.S. 408, 123 S.Ct. 1513.
. The majority points out that if we combine all of the punitive-damages awards in Sears, amounting to $6,5 million, it would be less than one-third of the punitive-damages award in this case. Maj. Op. at 1072-73 n. 35. But my proposed remittitur of $7.8 million is only 20% higher than the punitive damages assessed roughly 25 years ago in Sears.
. In a far more exhaustive survey, covering 507 published cases, two scholars found that when compensatory damages ranged from $5 million to $9.9 million, 39% of the cases permitted punitive damages in a ratio exceeding 4 to 1. Laura J. Hines & N. William Hines, Constitutional Constraints on Punitive Damages: Clarity, Consistency, & the Outlier Dilemma, 66 Hastings L.J. 1257, 1314-15 (2015).