Hostetler v. Johnson Controls, Inc.

Full title: AMOS and DEBBIE HOSTETLER, et al., Plaintiffs, v. JOHNSON CONTROLS, INC.…

Court: UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF INDIANA SOUTH BEND DIVISION

Date published: Jul 11, 2016

Case No. 3:15-CV-226 JD (N.D. Ind. Jul. 11, 2016)

Opinion

Case No. 3:15-CV-226 JD

07-11-2016

AMOS and DEBBIE HOSTETLER, et al., Plaintiffs, v. JOHNSON CONTROLS, INC., et al., Defendants.

JON E. DEGUILIO Judge United States District Court

OPINION AND ORDER

This is a putative class action arising out of alleged environmental contamination that originated at a plant formerly operated by defendant Johnson Controls, Inc. The plaintiffs, a number of individuals who own property or reside in an adjacent neighborhood, allege that contamination primarily consisting of trichloroethylene (TCE) entered the groundwater and migrated onto their properties. In 2007, Johnson Controls sold the plant to defendant Tocon Holdings, LLC, which briefly operated a manufacturing facility at the property, and more recently began demolishing the plant, which the plaintiffs argue has caused additional environmental hazards. The plaintiffs filed this action in state court, asserting a number of claims arising under state law on behalf of a proposed class consisting of individuals who have owned, rented, or occupied properties affected by the contamination. After litigating the case in state court for nearly a year, Johnson Controls filed a notice of removal, invoking federal jurisdiction under the Class Action Fairness Act. In response, the plaintiffs moved to remand this action back to state court. They argue that the removal was untimely, that Johnson Controls waived its right to remove, and that this case falls within the local controversy exception to the Class Action Fairness Act. For the reasons that follow, the Court denies the motion to remand.

I. FACTUAL BACKGROUND

Johnson Controls operated a plant in Goshen, Indiana from approximately 1937 through 2006, where it manufactured parts for thermostats and building control systems. The plant is adjacent to a residential neighborhood, and is also located near several business properties and Goshen High School. Through most of its operations at the plant, Johnson Controls used TCE—which can cause cancer and other serious illnesses—and other hazardous chemicals as part of its manufacturing process. The plant discharged its hazardous waste into an adjacent creek until 1965, after which it began storing the waste pending off-site disposal. Even then, however, there were a number of leaks and discharges of TCE and other hazardous chemicals into the ground.

In 1991, Johnson Controls began investigating TCE contamination of the soil at and around the plant. In a report dated the following year, its environmental consultant described a plume of contamination consisting of TCE and other hazardous chemicals that extended over a mile to the west of the plant. It also sampled the private drinking water wells of two residential homes in the plume and discovered extremely high concentrations of TCE. Over the next four years, Johnson Controls conducted more sampling and installed extraction wells designed to reduce the amount of TCE and other hazardous chemicals migrating underneath the neighborhood. In 1997, Johnson Controls also signed a Voluntary Remediation Agreement with the Indiana Department of Environmental Management, which established timetables for Johnson Controls to investigate and remediate the contamination. The complaint alleges, though, that Johnson Controls continued to operate its extraction wells in an ineffective and ultimately futile attempt to remediate the contamination, and that it failed to completely delineate the boundaries of the plume.

In 2007, Johnson Controls sold the property to Tocon, which operated a fiberglass manufacturing facility for a short time, after which manufacturing operations at the property ceased. The complaint alleges that Tocon purchased the plant with full knowledge that it was contaminated, but did nothing to stop the contamination from flowing offsite onto the plaintiffs' properties, and also failed to warn the residents of the neighborhood about the dangers of the contamination. In 2011, Johnson Controls began testing homes in the area for TCE vapors, and discovered dangerous levels of TCE vapors in multiple properties. It then installed vapor mitigation systems underneath fifteen of the properties, but the complaint alleges that those systems were ineffective and were not properly monitored. In addition, neither Johnson Controls nor Tocon warned the residents of the dangers of the contamination, and Johnson Controls' representatives allegedly told certain class members that the TCE vapors did not pose health risks. Finally, in early 2013, Tocon began demolishing the plant, but left large piles of debris on the property containing hazardous materials such as asbestos, lead, mercury, and PCBs. Tocon has taken no protective measure to prevent further contamination of the soil or groundwater.

On May 30, 2014, five named plaintiffs filed this action in state court. They asserted claims arising under state law against Johnson Controls and Tocon, including common law claims for trespass, nuisance, negligence, negligent infliction of emotional distress, and punitive damages, and a statutory environmental legal action claim. Underlying each of those claims are allegations that the defendants each allowed hazardous chemicals to migrate from the plant onto the plaintiffs' properties; that they failed to investigate and remediate the contamination; and that they failed to notify the residents of the existence and hazards of the contamination in their properties. The complaint seeks compensatory and punitive damages from both defendants, and seeks an injunction requiring both defendants to remediate the contamination to non-detect levels. The plaintiffs also asserted these claims on behalf of a proposed class (for liability purposes only) consisting of all persons within the area impacted by the plume of contamination who have "owned, rented, or occupied property that has been impacted by the Contamination caused by, or were otherwise exposed to, releases from the Plant." [DE 4-1 ¶ 101].

Litigation then proceeded in state court. Johnson Controls filed a motion to dismiss on August 25, 2014, seeking dismissal of the complaint in its entirety. That motion was fully briefed, and the state court held oral argument on the motion on December 4, 2014. The court took the motion under advisement before denying it in whole the following month. Plaintiffs also filed a motion for class certification on January 15, 2015. The parties also commenced discovery, and Johnson Controls filed a motion to compel, which the state court referred to a special master for hearing. Before those pending matters were resolved, though, Johnson Controls filed a notice of removal on May 28, 2015, removing the case to federal court. In response, Plaintiffs filed two separate motions to remand—the first for procedural deficiencies, and the second for lack of subject matter jurisdiction under the local controversy exception to the Class Action Fairness Act—which the Court consolidated. Those motions have now been fully briefed along with several other related motions, and are ripe for ruling.

II. DISCUSSION

Plaintiffs move to remand this case to state court. In support, they present three separate grounds. First, they argue that the notice of removal was untimely under the removal statute. Second, they argue that Johnson Controls waived its right to remove by litigating the case in state court prior to filing its notice of removal. Third, they argue that this case falls under the local controversy exception to the Class Action Fairness Act, which requires a court to relinquish jurisdiction over cases that meet prescribed statutory elements showing that the case is truly local in nature. The Court considers each in turn.

A. Timeliness

Plaintiffs first argue that Johnson Controls' notice of removal was untimely because the complaint was filed on May 30, 2014, but the notice of removal was not filed until May 28, 2015, nearly a full year later. In response, Johnson Controls argues that the statutory removal clocks never began running because Plaintiffs never expressly set forth an amount in controversy sufficient to invoke federal jurisdiction. Section 1446(b) sets forth two 30-day deadlines for removal. First, section 1446(b)(1) states:

The notice of removal of a civil action or proceeding shall be filed within 30 days after the receipt by the defendant, through service or otherwise, of a copy of the initial pleading setting forth the claim for relief upon which such action or proceeding is based . . . .

28 U.S.C. § 1446(b)(1). Second, section 1446(b)(3) states:

[I]f the case stated by the initial pleading is not removable, a notice of removal may be filed within 30 days after receipt by the defendant, through service or otherwise, of a copy of an amended pleading, motion, order or other paper from which it may first be ascertained that the case is one which is or has become removable.

28 U.S.C. § 1446(b)(3). Thus, if the complaint itself sets forth all of the predicates for removal, a defendant must remove the case within 30 days of receiving the complaint. If that information is not contained in the complaint, though, the removal clock can be triggered by an amended complaint or other paper setting forth the requirements for removal.

These provisions make "clear that the 30-day removal clock is triggered only by the defendant's receipt of a pleading or other litigation paper facially revealing that the grounds for removal are present." Walker v. Trailer Transit, Inc., 727 F.3d 819, 823-24 (7th Cir. 2013). Consistent with every other circuit to have addressed this issue, the Seventh Circuit applies this rule literally and strictly. Id. at 824 (collecting cases from the Second, Fourth, Fifth, Eighth, Ninth, and Tenth Circuits); see also Romulus v. CVS Pharmacy, Inc., 770 F.3d 67, 74 (1st Cir. 2014). Accordingly, the Seventh Circuit has held that "[t]he 30-day removal clock does not begin to run until the defendant receives a pleading or other paper that affirmatively and unambiguously reveals that the predicates for removal are present." Walker, 727 F.3d at 824 (emphasis added). "As applied to the amount-in-controversy requirement, the clock commences only when the defendant receives a post-complaint pleading or other paper that affirmatively and unambiguously specifies a damages amount sufficient to satisfy the federal jurisdictional minimums." Id. at 825. The defendant's subjective knowledge of any of those facts is immaterial. Id.

Importantly, "[t]he moment a case becomes removable and the moment the 30-day removal clock begins to run 'are not two sides of the same coin.'" Id. at 824 (quoting Kuxhausen v. BMW Fin. Servs. NA LLC, 707 F.3d 1136, 1141 n.3 (9th Cir. 2013)). Thus, a case might be removable on the date it is filed if the defendant has personal knowledge of facts outside the complaint sufficient to establish federal jurisdiction. The defendant could then immediately remove the case to federal court by affirmatively pleading those facts in its notice of removal. That does not mean, however, that the removal clock will have begun to run at that point, as the defendant's subjective knowledge cannot trigger the removal clock: "the timeliness inquiry is limited to . . . examining [the] contents of the clock-triggering pleading or other litigation paper; the question is whether that document, on its face or in combination with earlier-filed pleadings, provides specific and unambiguous notice that the case satisfies federal jurisdictional requirements and therefore is removable." Id. at 825 (also noting that, "as the text of the rule itself makes clear, the 30-day clock is triggered by pleadings, papers, and other litigation materials actually received by the defendant or filed with the state court during the course of litigation"); Cutrone v. Mortg. Elec. Registration Sys., Inc., 749 F.3d 137, 146 (2d Cir. 2014) ("[E]ven if a defendant could remove immediately upon the filing of a complaint because the case satisfies the requirements of federal subject matter jurisdiction, a complaint or other documents from the plaintiff that does not explicitly convey the removability of a case does not trigger the removal clocks of 28 U.S.C. §§ 1446(b)(1) and (b)(3).").

Here, Plaintiffs admit, as they must, that their complaint does not facially reveal any dollar figure as to the amount in controversy.¹ However, they note that the complaint seeks not only monetary damages, but far-reaching injunctive relief as well, and that the amount in controversy can be met by the cost to the defendant of complying with an injunction. Although, again, they concede that the complaint itself does not identify the cost for Johnson Controls to comply with the requested injunction, they argue that Walker's brightline rule should not apply in these circumstances, since a defendant's cost to comply with an injunction is more likely to be within the knowledge of a defendant than a plaintiff, and since Johnson Controls likely knew from the outset that its cost to comply with the injunction would far exceed $5 million (the amount-in-controversy threshold under the Class Action Fairness Act, 28 U.S.C. § 1332(d)(2)). The Court cannot agree that those circumstances justify any different rule in this case, though.

1 The complaint itself does not reveal the number of class members, either, and thus did not trigger the removal clock for that additional reason. However, Plaintiffs' motion for class certification, filed January 14, 2015, supplied that information. [DE 1-17].

First, both of the factors that Plaintiffs argue distinguish this case from Walker were equally present in Walker. There, the plaintiff, Walker, filed a class action suit seeking damages calculated as a percentage of the defendant's profits off of certain fees the defendant charged to its clients. Walker, 727 F.3d at 821. The complaint itself did not specify those amounts, though, as they were likely unknown to Walker. Thus, the removal clock did not begin to run upon the filing of the complaint. Later, in response to a request for admissions, Walker modified his damages theory to seek a percentage of the fees in their entirety, not just the defendant's profits off those fees. While the first theory would not have met the $5 million amount in controversy threshold, the second theory did. Based on its review of its own records, which established the amount of the fees in question, the defendant determined that the amount-in-controversy requirement was met and filed a notice of removal. At that point, according to its own affirmative representations in its notice of removal, the defendant subjectively knew that the amount in controversy was met. Yet, the Seventh Circuit held that even then, the removal clock still had not begun to run. The removal clock can only be triggered by the plaintiff's service of a complaint or other paper facially revealing the amount-in-controversy, but the defendant's removal "was not based on Walker's response to the requests for admission alone; it took Walker's admission and an estimate from a Trailer Transit executive to show that the jurisdictional limits were met." Id. at 825-26. Thus, the removal clock never actually began to run.² Id.

2 The removal clocks only set deadlines, they do not create exclusive windows within which a case must be removed, so a defendant can remove a case even if the removal clocks never begin. Walker, 727 F.3d at 824-25; Cutrone, 749 F.3d at 147.

This case is indistinguishable. Johnson Controls may be better positioned than Plaintiffs to determine its cost to comply with the requested injunction. But the defendant in Walker was in the same position, as it knew better than the plaintiff what fees it had charged to its own clients. Likewise, Johnson Controls may have subjectively known at the time the complaint was filed that its cost to comply with the injunction would exceed $5 million. But the defendant in Walker subjectively knew that it had charged fees in excess of $5 million by the time it filed its notice of removal, yet the Seventh Circuit held that the removal clock never began to run in that case. Plaintiffs' arguments thus do not distinguish this case from Walker in any meaningful respect. Moreover, although Plaintiffs argue that equitable relief is different in kind from monetary damages and should be subject to a different rule, the Eighth Circuit has held requests for injunctive relief to the same standard as requests for monetary damages. Reece v. Bank of N.Y. Mellon, 760 F.3d 771, 774 (8th Cir. 2014) ("Ordinarily, 28 U.S.C. § 1446(b) gives a defendant thirty days to remove a complaint to federal court, but Reece only sought equitable relief (an injunction prohibiting Mellon's non judicial foreclosure of his home), so Mellon was not obligated to remove to federal court within the thirty-day period.").

In addition, Plaintiffs' argument that the inquiry should depend on the respective availability of information to the parties has been repeatedly and expressly rejected in the courts of appeals. For example, in Cutrone, the plaintiff argued that the trigger for the removal clock should depend upon "the availability of information to the parties," and that Walker and comparable cases should only apply where "'plaintiffs are in possession of information to explicitly specify damages."" 749 F.3d at 145. The Second Circuit disagreed, finding that the brightline rule placing the burden on plaintiffs "is preferable to the uncertainties faced by defendants in determining removability" and also "avoids courts 'expending copious time determining what a defendant should have known or have been able to ascertain at the time of the initial pleading [or other relevant filing].'" Id. (alteration in original) (quoting Mumfrey v. CVS Pharm., Inc., 719 F.3d 392, 399 (5th Cir. 2013)). Other courts of appeals that have addressed similar arguments have reached the same conclusion. E.g., Romulus, 770 F.3d at 76 (finding that the clock had not begun even where the defendant could have ascertained the amount in controversy by applying information in its possession to the allegations in the complaint); Mumfrey, 719 F.3d at 399 (rejecting the plaintiff's argument that the removal clock started when "he pleaded for lost wages and CVS, as his employer, knew his salary"); Kuxhausen, 707 F.3d at 1141 (rejecting the plaintiff's argument that the defendant "should have consulted its business records" to calculate the amount in controversy based on the allegations in the complaint); Moltner v. Starbucks Coffee Co., 624 F.3d 34, 37-38 (2d Cir. 2010).

Plaintiffs further argue that this brightline rule creates a potential for abuse by defendants, who, unless limited by a removal clock, could test the waters in state court even if they know all along that they could remove the case whenever they want. Again, however, the courts of appeals have repeatedly addressed and rejected this argument, finding that the brightline rule is mandated by statute and also supported by countervailing policy interests of clarity and predictability. Romulus, 770 F.3d at 76; Cutrone, 749 F.3d at 147; Roth v. CHA Hollywood Med. Ctr., L.P., 720 F.3d 1121, 1126 (9th Cir. 2013); Moltner, 624 F.3d at 38. In addition, plaintiffs are not helpless against such tactics. As the Ninth Circuit explained in Roth:

[P]laintiffs are in a position to protect themselves. If plaintiffs think that their action may be removable and think, further, that the defendant might delay filing a notice of removal until a strategically advantageous moment, they need only provide to the defendant a document from which removability may be ascertained. 28 U.S.C. § 1446(b)(3). Such a document will trigger the thirty-day removal period, during which defendant must either file a notice of removal or lose the right to remove.

720 F.3d at 1126. Thus, the potential for manipulation by defendants does not justify an alternative rule.

Applying these principles here, the Court finds that Johnson Controls' notice of removal was timely. The complaint itself did not state the amount in controversy. Though it described the relief sought, and Johnson Controls could have determined based on its own inquiry that the amount-in-controversy requirement was met, the complaint did not facially reveal either the amount of monetary damages sought or the cost for Johnson Controls to comply with the injunction. Like in Walker, the notice of removal did not rely on the complaint alone; it took the complaint's description of the injunctive relief requested and an estimate from Johnson Controls as to the cost to comply with that injunction to plead the amount in controversy. Nor have Plaintiffs identified any other pleading or other litigation paper that unambiguously and unequivocally revealed that information. Therefore, the 30-day clock never began to run, so Johnson Controls' notice of removal could not have been untimely. Accordingly, the Court denies Plaintiffs' motion to remand on this basis.

B. Waiver

Plaintiffs next argue that, even if Johnson Controls complied with the removal clock, it nonetheless waived its right to remove by litigating the case on the merits in state court prior to removal. Plaintiffs note that Johnson Controls actively litigated the case in state court for nearly a year prior to removing. Most notably, Johnson Controls filed a motion to dismiss the complaint in its entirety, it briefed and argued that motion, it agreed to a deadline for the state court to rule on that motion, and it received the state court's order denying the motion in whole, all prior to filing its notice of removal. Plaintiffs argue that this conduct, by which Johnson Controls asked the state court to adjudicate the case on its merits, was inconsistent with a desire to have the claims decided in a federal forum, so Johnson Controls should be held to have waived that right.

In general, a party can waive its right to remove a case to federal court by taking actions in the state court that objectively and unequivocally manifest an intent to have the case adjudicated in the state court and to abandon the right to a federal forum. 16 Moore's Federal Practice § 107.132[1] (Matthew Bender 3d ed.). However, in Rothner v. City of Chicago, 879 F.2d 1402 (7th Cir. 1989), which appears to be the only case in which the Seventh Circuit has addressed this topic, the Seventh Circuit set a demanding standard to justify remanding a case under the waiver doctrine. Casting aside much of the previous precedent on waiver, the court noted that "the doctrine of waiver developed out of the necessity to interpret the pre-1948 removal statutes," which did not set definite time limits on removal, and that the 1948 revisions that implemented those time limits "for the most part mooted the question of waiver." Id. at 1415. While the court acknowledged that waiver still existed as a common law doctrine, it believed that waiver would be "rare" and would only occur in "extreme situations" such as "where the parties have fully litigated the merits" of the case prior to removal. Id. at 1416.

In responding to Plaintiffs' waiver argument, Johnson Controls relies entirely on the Seventh Circuit's decision in Walker, which was discussed above. The defendant in Walker filed the notice of removal sixteen months after the case was filed in state court, and only after the defendant opposed and lost a motion for class certification and filed a motion for summary judgment. 727 F.3d at 821-22. Johnson Controls argues that if those actions did not waive the defendant's right to remove, then its own actions in state court could not have, either. Johnson Controls' reliance on Walker is curious, though, as waiver was not at issue in Walker, and the court did not address or even mention that doctrine. Walker addressed only the timeliness of removal under the statutory removal clocks, and its holding was that the removal "was not untimely," not that the defendant had not waived its right to remove. 727 F.3d at 826.

Nonetheless, Walker is instructive for the reason that waiver was not and could not have been raised in that case: "because waiver refers to a litigant's intentional relinquishment of a known right, the defendant's actions in state court can waive the right of removal only where it is unequivocally apparent that the case is removable at the time those actions were taken." Becker v. Lindblom, No. 10-CV-792, 2010 WL 4674305, at *2 (W.D.N.Y. Nov. 18, 2010), report and recommendation adopted, 2011 WL 335864 (W.D.N.Y. Jan. 31, 2011) (internal quotations and citations omitted). Thus, "a defendant who actively invokes the jurisdiction of the state court and interposes a defense in that forum is not barred from the right to removal in the absence of adequate notice of the right to remove." Akin v. Ashland Chem. Co., 156 F.3d 1030, 1036 (10th Cir. 1998); see also Koch v. Medici Ermete & Figli S.R.L., No. 13-1411, 2013 WL 1898544, at *2 (C.D. Cal. 2013) ("[T]he first requirement [to finding waiver] is that 'it must be unequivocally apparent that the case is removable.'" (quoting Moore's Federal Practice, currently located at 16 Moore's § 107.132[1])). In Walker, all of the defendant's actions that could arguably give rise to waiver took place before the plaintiff asserted a new damages theory that could satisfy the amount-in-controversy requirement (at which point the defendant promptly removed). Because the case was not yet removable at the time of those actions, they could not have waived the defendant's right to remove.

This case is similar in that the removal clock never began to run, so all of Johnson Controls' actions in state court occurred prior to any removal clock beginning. The only difference is that in Walker, the defendant could not have removed the case earlier, since it previously appeared that the amount-in-controversy could not be met, whereas here, Johnson Controls could have removed the case at any time by affirmatively pleading its own subjective belief as to the amount in controversy. In order for that distinction to make a difference, though, the Court would have to find that notice of removability for waiver purposes has a lower bar than notice of removability for removal clock purposes, and permits an inquiry into the defendant's subjective knowledge. The Court cannot do so.

First, although that particular distinction is seldom addressed expressly, courts commonly apply the same standard when determining whether the defendant had adequate notice of removability for waiver purposes as they apply when determining when the removal clock began to run. For example, in Akin, the Tenth Circuit first found that the removal was timely because the initial pleading "did not provide unequivocal notice of the right to remove." 156 F.3d at 1035-36. Then, applying the same standard in the waiver context, the court held that the defendant did not waive its right to remove because the defendant's actions in state court "were taken before it was unequivocally apparent that the case was removable." Id. at 1036. Likewise, many district courts have relied on their analysis as to when the removal clock began in determining when a defendant had adequate notice of removability for waiver purposes. E.g., Misner v. State Farm Fire & Cas. Co., No. CIV-14-873, 2014 WL 6879094, at *8 (W.D. Okla. Dec. 4, 2014) (finding that a defendant did not waive the right to remove by taking action in state court prior to the time the removal clock began); Becker v. Lindblom, No. 10-CV-792, 2010 WL 4674305 (W.D.N.Y. Nov. 18, 2010) (same); Carvalho v. Equifax Info. Servs., LLC, No. C 08-1317, 2008 WL 2693625, at *4 (N.D. Cal. July 7, 2008) (same); Mattel, Inc. v. Bryant, 441 F. Supp. 2d 1081, 1090-91 (C.D. Cal. 2005) (same); Adams v. Lederle Labs., 569 F. Supp. 234 (W.D. Mo. 1983) ("Where the second paragraph of section 1446(b) provides the operative law for a case, it is obvious that a defendant's acts in the state court, taken before the case became removable or before the defendant was able to determine that the case was removable, cannot be taken as a waiver of the right to remove.").³

3 Even Fate v. Buckeye State Mut. Ins. Co., 174 F. Supp. 2d 876, 882 (N.D. Ind. 2001), upon which Plaintiffs heavily rely, applied the same notice standard in both the timeliness and waiver contexts. Although Walker subsequently overruled Fate as to the scope of that standard, at least in the timeliness context, Fate still suggests that the same standard applies to both.

Applying the same standard for notice of removability in both contexts makes good sense, too. As the Seventh Circuit discussed in Rothner, Congress enacted the removal clocks in order to "correct ambiguities that gave rise to forfeitures of the right to remove and . . . to make uniform and certain the time period for removal while insuring that the defendant have adequate time to remove." 879 F.2d at 1416. The present statute accomplishes those goals by setting definite timelines triggered by unequivocal notice. See Walker, 727 F.3d at 823-24. The First Circuit likewise recognized these principles in Romulus:

We follow, as we must, the Congressional policy choice inherent in the statutory text. As we have previously explained, "the obvious purpose of starting the 30-day clock only after the defendant's receipt of a 'paper' revealing the case's removability is to ensure that the party seeking removal has notice that the case is removable before the limitations period begins to run against it."

770 F.3d at 76 (quoting Woburn Five Cents Sav. Bank v. Robert M. Hicks, Inc., 930 F.2d 965, 970 (1st Cir. 1991)) (emphasis omitted). Adopting a different standard in the waiver context would undermine that statutory framework, and would insert through the waiver doctrine the very uncertainty that Congress meant to eliminate through the removal clocks. As a construct of the common law, the waiver doctrine should not be permitted to upend the carefully drawn legislative policy in that manner.

The primary counterargument to this approach would be that it could allow even defendants who subjectively know they have a right to remove to linger and test the water in state court, thus creating a potential for judge-shopping and for delays after substantial progress has been made in state court. However, those same policy arguments can be raised equally in the timeliness context, as discussed above, and have been uniformly rejected in that context. Moreover, the Seventh Circuit in Rothner expressly stated that a "defendant's motive for removing is not a proper consideration for remand." 879 F.2d at 1408. In addition, as also discussed above, plaintiffs are typically in a position to prevent that sort of manipulation. If plaintiffs want to prevent defendants from taking a second bite at the apple by removing to federal court, they need only provide the defendants with clear and unequivocal notice of their right to remove before the defendants take their first bite at the apple in state court. At that point, the removal clock (or possibly the waiver doctrine) will force the defendants' hand. Applying the waiver doctrine even without such notice, though, would upset the interests of clarity, efficiency, and predictability that the removal clock is meant to serve, and would disrupt rather than supplement the statutory scheme.

Therefore, the Court finds that Johnson Controls did not waive its right to remove. As discussed above, the removal clock never began to run since Plaintiffs never provided Johnson Controls with clear and unequivocal notice that the requirements for removal were met. All of Johnson Controls' actions in state court thus occurred before it received the requisite notice of removability, so those actions cannot give rise to waiver.⁴ Plaintiffs' motion to remand on the basis of waiver must therefore be denied.

4 If Johnson Controls could be charged with adequate notice of removability from the outset, this case could be a candidate for waiver. Litigating a case in state court for nearly a year and filing, briefing, arguing, and seeking a ruling on a motion to dismiss that asks the state court to adjudicate the case on its merits is inconsistent with a desire to have the claims adjudicated in a federal forum, which is the purpose of removal. See Hughes v. UPS Supply Chain Solutions, Inc., 815 F. Supp. 2d 993, 998 (W.D. Ky. 2011) ("If a potentially dispositive motion, such as a motion to dismiss, is made and argued by the defendant, the state court's adverse decision cannot be appealed to this Court by way of removal." (internal quotations omitted)). Since the requisite notice is not present here, though, the Court need not determine whether these circumstances meet the exacting standard for waiver the Seventh Circuit described in Rothner (albeit mostly in dicta), which is a question on which district courts in this circuit have split.

C. Local Controversy Exception

Finally, Plaintiffs move to remand under the local controversy exception to the Class Action Fairness Act. Under the Class Action Fairness Act, federal courts have subject matter jurisdiction over class actions when three elements are met: (1) the aggregate amount in controversy exceeds $5 million, (2) any member of the class of plaintiffs is a citizen of a state different from any defendant (minimal as opposed to complete diversity), and (3) the proposed class has at least 100 members. 28 U.S.C. § 1332(d)(2), (d)(5)(B). It is undisputed that each of those elements are present here. The aggregate amount in controversy greatly exceeds $5 million; Johnson Controls is a citizen of Wisconsin and many of the plaintiffs are citizens of Indiana; and there are well over 100 members of the proposed class.

However, the Class Action Fairness Act also contains limited exceptions under which an action that meets those threshold criteria can or must be remanded. One of those is the local controversy exception, which requires a court to decline to exercise jurisdiction over a class action in which:

(I) greater than two-thirds of the members of all proposed plaintiff classes in the aggregate are citizens of the State in which the action was originally filed;

(II) at least 1 defendant is a defendant—

(aa) from whom significant relief is sought by members of the plaintiff class;

(bb) whose alleged conduct forms a significant basis for the claims asserted by the proposed plaintiff class; and

(cc) who is a citizen of the State in which the action was originally filed; and

(III) principal injuries resulting from the alleged conduct or any related conduct of each defendant were incurred in the State in which the action was originally filed; and

(ii) during the 3-year period preceding the filing of that class action, no other class action has been filed asserting the same or similar factual allegations against any of the defendants on behalf of the same or other persons . . . .

28 U.S.C. § 1332(d)(4)(A). The party seeking to invoke this exception bears the burden of proof of establishing each of those elements. Myrick v. WellPoint, Inc., 764 F.3d 662, 664-65 (7th Cir. 2014); Hart v. FedEx Ground Package Sys., Inc., 457 F.3d 675, 679-80 (7th Cir. 2006).

The parties agree that the third and fourth elements are met here, as the injuries from the contamination would have been incurred in Indiana, and no other class action asserting a similar claim has been filed. The parties argue at length over the first two elements, though. Relying on a survey of potential class members and a report from a statistician, Plaintiffs argue that over two-thirds of the members of the potential class are citizens of Indiana. They also argue that Tocon, also a citizen of Indiana, is a defendant from whom they seek significant relief and whose conduct forms a significant basis for their claims. In response, Johnson Controls argues that Plaintiffs' survey is flawed and invalid, and it has also filed a motion to strike the expert report under Rule 702. It further argues that Tocon is not a substantial defendant because Tocon has owned the property in question for a comparably short amount of time and did not dump TCE or otherwise contaminate the groundwater. The Court considers the two elements in turn.

1. Citizenship of Two-Thirds of the Class

The first element of the local controversy exception is that "greater than two-thirds of the members of all proposed plaintiff classes in the aggregate are citizens of the State in which the action was originally filed." 28 U.S.C. § 1332(d)(4)(A)(i)(I). As the party seeking remand, Plaintiffs have the burden of proving the class members' citizenship by a preponderance of the evidence. Myrick, 764 F.3d at 64-65; In re Sprint Nextel Corp., 593 F.3d 669, 673 (7th Cir. 2010). Unless a plaintiff defines a proposed class as consisting only of citizens of the forum state (which Plaintiffs here did not), they "need[] to produce some evidence that would allow the court to determine the class members' citizenships on the date the case was removed." Myrick, 764 F.3d at 665. For individuals, which make up the vast majority of the class here, citizenship is determined by domicile—"the person's long-term plan for state of habitation." Id. 664. Because domicile is not equivalent with residence, "a court may not draw conclusions about the citizenship of class members based on things like their phone numbers and mailing addresses." Sprint, 593 F.3d at 674. Proving the citizenship of each class member individually is not necessary for these purposes, though; the Seventh Circuit has suggested that a plaintiff could prove the citizenship of the potential class members by determining the citizenship of a "representative sample" of the class, such as by conducting a survey of a random sample of class members, and then "us[ing] statistical principles to reach a conclusion as to the likelihood that two-thirds or more of the proposed class members are citizens" of the forum state. Id. a 675-76; Myrick, 764 F.3d at 665.

Plaintiffs have attempted to do that here. It is worth emphasizing at the outset, though, that the class Plaintiffs proposed is quite broad, so they have created quite a difficult task for themselves. Plaintiffs sought certification of a class consisting of "all persons within the Class Area [meaning the area impacted by the contamination] who have owned, rented, or occupied property that has been impacted by the Defendants' contamination and toxic vapors caused by releases from Defendants' activities at the Site between 1992 and May 30, 2014." [DE 1-17]. Thus, the proposed class spans more than two decades of owners, renters, and occupants of over 100 homes within the class area (11 of which were demolished at some point during this period). Many of those individuals could have moved away years ago, as even the complaint stresses. [DE 4-1 ¶¶ 107, 108 ("Many putative Class members were renters who have moved out of the Neighborhood after the exposure to the Contamination. Many persons have sold their homes in the Neighborhood following their exposure to the Contamination.")]. The class area also includes nine business properties and Goshen High School. Moreover, by including all persons who have ever "occupied" any of the properties during that time period, the class would include even short-term occupants or temporary guests. Those individuals could be very difficult to identify or locate, particularly when they may have occupied the properties decades ago.

To determine the citizenship of these class members, Plaintiffs began by using public records to identify as many people as possible who have owned or resided at the addresses within the class area. They ultimately compiled a list of 2,130 individuals that the public record searches affiliated with those addresses during the relevant time period. Because they decided to conduct their survey by telephone, Plaintiffs then eliminated from their list individuals for whom the public records did not contain a phone number. They also eliminated suspected duplicates and individuals believed to be deceased, to arrive at a list of 1,314 potential class members who could be contacted by phone. Plaintiffs' counsel then sent that list to a survey firm, which placed calls to each individual (making multiple calls to up to three different phone numbers per individual) and asked them to complete a survey inquiring about their citizenship. Ultimately, 153 potential class members completed surveys, of whom 117—or 76 percent—reported being Indiana citizens on the date this action was filed in state court and as of the date of the survey.⁵ Plaintiffs then had a statistician, Dr. Cohen,⁶ calculate confidence intervals from those results. Dr. Cohen concluded that, using a 95 percent confidence level, between 69 and 83 percent of the members of the class are citizens of Indiana. At a 50 percent confidence level, the citizenship rate would be 74 to 79 percent. Accordingly, Dr. Cohen opines that a preponderance of the evidence shows that at least two-thirds of the class is citizens of Indiana.

5 Plaintiffs' counsel also separately collected affidavits from their clients and the clients' family members (but only those who are current residents of Indiana) and had an investigator interview current residents of the homes in the class area to determine their citizenship. Those materials can essentially be disregarded from this analysis, though, because Plaintiffs have made no effort to weight those results proportionally to the size of these groups relative to the total population of potential class members. Accordingly, those materials have no value from a statistical perspective, as even Dr. Cohen effectively concedes, as he declined to calculate any confidence interval based on those materials.

6 Dr. Cohen co-authored his report with Laura Steiner, an expert on survey design and implementation. Since the issues in dispute mostly pertain to the report's statistical analysis, not the design or implementation of the survey, the Court refers primarily to Dr. Cohen, as he was responsible for that aspect of the report.

Johnson Controls disagrees. It has submitted its own expert report from a statistician, Dr. Wyner, who critiques Dr. Cohen's report at length, and who argues that Dr. Cohen's data provides no reliable basis upon which to express any opinion as to the citizenship of the class as a whole. Johnson Controls has also moved to exclude Dr. Cohen's report under Rule 702 of the Federal Rules of Evidence. Under Rule 702, a qualified expert may offer a relevant opinion if "the testimony is based on sufficient facts or data;" "the testimony is the product of reliable principles and methods;" and "the expert has reliably applied the principles and methods to the facts of the case." The Seventh Circuit has observed that while a courts gatekeeping role is different when, as here, the court is the factfinder, "[t]hat is not to say that the scientific reliability requirement is lessened in such situations . . . ." In re Salem, 465 F.3d 767, 777 (7th Cir. 2006). There is no dispute here that Dr. Cohen is qualified as an expert in statistics, so the question here is whether his testimony meets the reliability prongs. The Court finds that it does not, as Dr. Cohen failed to apply any valid statistical analysis to consider whether the survey responses from the 153 potential class members could reliably be extrapolated to the class as a whole.

The first step in conducting a reliable survey is defining the universe of people to be surveyed. "Courts consider the selection of the proper universe as one of the most important factors in assessing the validity of a survey as well as the weight that it should receive. Selection of a proper universe is so critical that even if the proper questions are asked, the results are likely to be irrelevant." Leelanau Wine Cellars, Ltd. v. Black & Red, Inc., 452 F. Supp. 2d 772, 781 (W.D. Mich. 2006) (internal quotations and citations omitted); see also Shari Seidman Diamond, Reference Guide on Survey Research, in Fed. Judicial Ctr., Reference Manual on Scientific Evidence 359, 376 n.76 (3d ed. 2011) (hereafter "Survey Reference Guide") ("Identification of the proper target population or universe is recognized uniformly as a key element in the development of a survey." (collecting sources)). Here, Plaintiffs attempted to identify the potential class members by consulting three databases of public records in order to find people who were associated with any of the mailing addresses within the class area during the pertinent period. They ran the addresses in question through the Polk City Directory, a database that contains demographic data compiled from a variety of sources; through a title search, to determine the ownership of each home; and through the Transunion/TLO database, a database available to licensed private investigators. As explained by Plaintiffs' investigators, these sources would have identified any individual that was linked to any of the addresses in question by "some kind of official document," such as a deed, a voter's registration, utility records, and the like. [DE 1-17; DE 50-4 p. 80-81; DE 50-6 p. 37-39]. Plaintiffs consulted these sources extensively, and ultimately compiled a list of 2,130 individuals that the public records linked to any of the addresses in question.

However, while Plaintiffs appear to have squeezed as much useful information out of these sources as they have to offer, at no point did they consider whether the individuals reflected in these records would represent a fair cross-section of the class as a whole. As came to light during the deposition of Plaintiffs' investigator, these records systematically failed to capture at least two groups of people: individuals who were minors at the time they occupied the class area, and individuals who were only temporary occupants (including short-term guests) in the class area. Minors typically do not own property, have utilities in their name, or otherwise appear in the sorts of public records that these databases reflect. Thus, for the most part, individuals who were minors during their occupancy in the class area did not appear in the public records list. Similarly, temporary occupants are unlikely to appear in these databases unless some record happened to associated them with the address during their stay at the property. As Plaintiffs' investigator repeatedly noted, "[t]here's got to be a record" of some sort associating an individual with the addresses in question for the individual to be reflected in these sources. [DE 504- p. 82, 107]. These are potentially substantial omissions. Minors make up about a quarter of the population of Goshen according to census data, yet are largely unrepresented in these public records.⁷ Moreover, as Dr. Cohen observed during his deposition, temporary occupants could constitute "tens of thousands of individuals" given the length of the class period, [DE 50-7 p. 9-10], which would dwarf the 2,130 individuals (apparently consisting mostly of adult owners and residents of the properties in question) reflected in Plaintiffs' list.

7 Minors share the citizenship of their parents, but the class period spans over two decades, over which time many of the minors would have become adults, and citizenship is measured as of the date of removal, not as of the time of the individuals' occupancy in the class area.

Even more notably, Dr. Cohen acknowledged during his deposition that he never even considered whether the public records list was representative of the class as a whole. [See DE 50-7 p. 59 ("I didn't determine whether the exclusion of minors had any impact. I didn't do any kind of bias test.")]. He simply accepted the list given to him by Plaintiffs' counsel as comprising the entirety of the class. In fact, he clarified that his opinion only pertained to the citizenship of the list that was provided to him, and did not extend to the citizenship of the class as a whole.⁸ In other words, Dr. Cohen's report answered the wrong question: he opined only on the citizenship of individuals in the public records list, not the class as a whole. Accordingly, having failed to even consider the composition of the class as a whole and having only considered a subset of that class that is most likely to have long-term ties to the area, Dr. Cohen had no reliable basis on which to extrapolate the citizenship of those individuals to the class as a whole. As it turns out, he did not even attempt to do so, leaving Plaintiffs without an expert opinion as to the citizenship of the class as a whole.

8 [DE 50-7 (Cohen dep.) p. 25-28 ("Q. Did you draw conclusions about the incidence of citizenship in the class as a whole or only in the list you were provided? A. I could only draw it from the list I was provided."), p. 107 ("Q. . . . When you say more than two-thirds of the potential class members are Indiana citizens, the potential class members you're talking about are the names that populated the list that you were provided by plaintiffs' attorneys? A. Yes."), p. 116 (clarifying that his conclusion applied to the citizenship "of the list we were provided with," and stating "I do not" know if that list is representative of the universe); see also DE 50-16 (Steiner dep.) (stating that "[m]y understanding was that the list was the universe of the potential class members," and that if potential class members "were not on the list, I would probably have to evaluate what the impact would be and to know what the scope of that problem was to know whether it would cause a problem or not," but that she had not done that analysis)].

The second step in conducting a reliable survey is to select an appropriate sample of the total population from which to take the survey. Survey Reference Guide, at 380 ("Identification of a survey population must be followed by selection of a sample that accurately represents that population."). "For the survey's results to be accurate, it must use a sampling method that ensures the sample is representative of the entire population . . . ." Hurt v. Commerce Energy, Inc., No. 1:12-cv-758, 2015 WL 410703, at *5 (N.D. Ohio Jan. 29, 2015); see also Survey Reference Guide, at 379 ("If the sampling frame does not include important groups in the target population, there is generally no way to know how the unrepresented members of the target population would have responded."). "Usually this is done using some version of random sampling." Hurt, 2015 WL 410703, at *5; see generally Myrick, 764 F.3d at 665 (suggesting that plaintiffs could prove the citizenship of a class by "tak[ing] a random sample" of class members (emphasis added)); Sprint, 593 F.3d at 675 (suggesting that plaintiffs could prove the citizenship of a class by submitting "evidence going to the citizenship of a representative sample" (emphasis added)).

Here, however, Plaintiffs did not select a random sample of class members to administer the survey to. They instead took their list of 2,130 individuals from the public records, and eliminated individuals who they could not contact by phone—primarily individuals for whom no phone number was listed in the public records—to arrive at a sample of 1,314 potential class members. This was not a random sample, but a "sample of convenience," meaning the sample was chosen based on which individuals were easiest to survey, not whether they would fairly represent the composition of the class. This type of sample creates a potential for systemic bias that would prevent the results of the survey from being reliably extrapolated from the sample to the class as a whole.⁹ Hurt, 2015 WL 410703, at *5 ("Plaintiffs have not presented any credible analysis suggesting that their sample was representative of the entire population. . . . This was merely a convenience sample where Plaintiffs' counsel selected the people they had at hand, rather than people who would be representative of the entire population."); In re Countrywide Fin. Corp. Mort.-Backed Sec. Lit., 984 F. Supp. 2d 1021, 1039 (C.D. Cal. 2013) ("The Court cannot countenance the use of this type of convenience sample that is 'easy to take but may suffer from serious bias.'"); David H. Kaye & David Freeman, Reference Guide on Statistics, in Fed. Judicial Ctr., Reference Manual on Scientific Evidence 211, 285 (3d ed. 2011) (noting that a convenience sample is "easy to take but may suffer from serious bias").

9 For example, the records from the beginning of the class period were unlikely to contain phone numbers. [DE 50-4 p. 72-73]. Because those individual had the most amount of time to move out of state, they were the most likely to have done so as a matter of probability, yet they were mostly excluded from the sampling frame.

While such samples do not necessarily render a survey inadmissible,¹⁰ they at least require an analysis as to what effect they may have on the validity of the results. See Survey Reference Guide, at 382 (noting that "when respondents are not selected randomly from the relevant population, the expert should be prepared to justify the method used to select respondents," and that "[s]pecial precautions are required to reduce the likelihood of biased samples"). Again, however, Dr. Cohen admitted during his deposition that he did not even consider whether the sample used here was appropriate:

10 They do, however, mean that confidence intervals should not be assigned, which would undermine Plaintiffs' reliance on Sprint. Survey Reference Guide, at 383 (noting that "[c]onfidence intervals technically should not be computed" when convenience samples are used); [DE 50-3 ¶ 20 ("[U]nder generally accepted statistical principles, no confidence interval can be applied at all [when a convenience sample is used] because the sample was not random.")].

Q. Did you do anything to determine if the exclusion of the people that didn't have contact information introduced any type of bias into the data?

A. No.

Q. So as you sit here today, you can't tell us if that exclusion worked bias in the results or not?

A. I can't.

[DE 50-7 p. 59].

Plaintiffs argue in response that they went "above and beyond a random sample" by instead trying to contact all of the individuals for whom they had contact information. While this may have been more laborious than a random sample, Plaintiffs offer no reason to believe it is reliable or eliminates the need to account for systemic biases in their survey. To the contrary, as even Dr. Cohen observes, "If there's a bias, a larger sample wouldn't alleviate the bias." [DE 71-2 p. 89]; see also Countrywide, 984 F. Supp. 2d at 1039 ("Systematic error, or bias, . . . cannot be reduced by increasing sample size."). Regardless of how many people Plaintiffs attempted to contact, their sample was still only a convenience sample of individuals reflected in the public records for whom they had phone numbers they could dial, not a sample that was drawn randomly from the class as a whole. Dr. Cohen's failure to consider this source of bias further diminishes the reliability of his report.

Finally, a reliable survey requires an analysis of whether the results were impacted by non-response bias, which occurs when the nonresponses to a survey are not random, but correlate to a trait the survey is intended to measure. As the court in Wallace v. Countrywide Homes Loans Inc., explained:

[T]he point of the survey evidence is to allow the trier of fact to draw classwide inferences from the information obtained from the respondents. In order to ensure that such inferences are valid, not only must the survey begin with a random sample, but nonresponses must also be random. A survey that "begins with a random sample," but does not "take measures to assure that nonresponses are random and provide analysis of the reasons of nonresponse," is not "the product of reliable principles and methods."

No. SACV 08-1463, 2012 WL 11896333, at *4 (C.D. Cal. Aug. 31, 2012) (quoting Marlo v. United Parcel Serv., Inc., 251 F.R.D. 476, 485 (C.D. Cal. 2008)); accord Medlock v. Taco Bell Corp., No. 1:07-cv-1314, 2015 WL 8479320, at *6 (E.D. Cal. Dec. 9, 2015). Non-response bias typically becomes a concern when the response rate falls below eighty percent. Survey Reference Guide, at 384. Response rates below that point—even far below that point—are commonplace and do not necessarily invalidate a survey, but they do require an analysis as to the reasons for the nonresponses and the effect they may have on the results.¹¹

11 Survey Reference Guide, at 383 ("It is incumbent on the expert presenting the survey results to analyze the level and sources of nonresponse, and to assess how that nonresponse is likely to have affected the results."); Reference Guide on Statistics, at 226 ("A large nonresponse rate warns of bias, although supplemental studies may establish that nonrespondents are similar to respondents with respect to characteristics of interest."); Hurt, 2015 WL 410703, at *5 (excluding a survey where "there was no serious testing to determine whether the respondents were representative of the entire class, and [the expert] did not account for potential bias in the results"); Medlock, 2015 WL 8479320, at *6 ("[A] lower response rate does not necessarily mean that the survey is invalid, but a lower response rate 'generally requires an analysis of the determinants of nonresponse.' It is the survey proponent's burden to assure that the nonresponses are random.").

Here, the response rate was only 11.6%, or 153 respondents out of the 1,314 individuals in the sample frame. Once again, however, Dr. Cohen failed to even consider this source of bias as part of his analysis. When asked if there was any non-response bias in his results, Dr. Cohen responded, "I don't know if there is necessarily. You can't tell." [DE 50-7 p. 106]. He then explained that non-response bias could be evaluated by comparing the characteristics of non-respondents against the characteristics of respondents, but noted that he had not been asked to perform that analysis. Id. All Dr. Cohen said was done to account for non-response bias was "to work as hard as you can to try and reach everybody on the list" by making multiple calls to each individual. [DE 50-7 p. 105]. While increasing the response rate might have been helpful, though, those efforts still only produced a response rate of 11.6%. At that point, a reliable statistical analysis would have required an analysis of the reasons for and effect of the non-response rate, but Dr. Cohen did not even consider those issues.

In sum, it is altogether unclear what statistical expertise Dr. Cohen brought to bear on this analysis other than plugging the numbers 153 (the number of survey respondents) and 117 (the number of Indiana citizens among those respondents) into a formula to calculate confidence intervals. He did not consider whether the public records list that counsel provided was representative of the class as a whole; whether the subset of people they attempted to survey from that list was an appropriate sample; or whether the high non-response rate from that sample affected the validity of the survey results. Nonetheless, having assumed away or neglected each of those critical steps, his report proceeded to extrapolate the survey results to the citizenship of the class as a whole. That is not the product of a reliable method, and Plaintiffs have not shown that it can be counted on to reliably estimate the citizenship of the class, as is required to be admissible under Rule 702. The Court therefore grants Johnson Controls' motion to strike. Likewise, even if the report met the standards for admissibility, Plaintiffs have not shown that these results are deserving of any weight such that the report could satisfy their burden of proof.

Plaintiffs offer a number of arguments and supplemental materials in reply in attempt to salvage their efforts, but they are both too little and too late.¹² As to considering the proper universe, Dr. Cohen states in his rebuttal report that he believes the public records list "was complete enough to provide probative evidence regarding the citizenship of the individuals who occupied housing units in the relevant area during the relevant time period." [DE 63-3 ¶ 8]. That is a meaningless qualification, though. As a subset of the larger class, the citizenship of any one individual class member may be "probative" of the citizenship of the class as a whole, in the sense that when combined with other evidence, it can help establish the citizenship of the class. That does not mean, though, that any one individual's citizenship can be extrapolated to prove the citizenship of the class. Dr. Cohen's role as a statistician is not to determine whether the public records list is probative of the overall class, but to determine whether it is representative of that class, such that its citizenship can be reliably extrapolated to the citizenship of the class. Even in his rebuttal report, Dr. Cohen does not purport to do so.

12 A rebuttal report is an opportunity to respond to criticisms of a methodology, not to attempt to apply a methodology for the first time after failing to reliably do so in the first instance. Bowman v. Int'l Bus. Mach. Corp., No. No. 1:11-cv-593, 2013 WL 1857192, at *7 (S.D. Ind. May 2, 2013) ("A rebuttal report is not the time to change methodologies to account for noted deficiencies; instead, it is to respond to criticisms of such methodologies."). Moreover, none of the materials Plaintiffs rely on in rebuttal could not have been presented at the time of the initial reports.

Plaintiffs also submitted a rebuttal affidavit from a former FBI agent they used to conduct the public records searches and interview current residents. He states that based on his experience, "It is not expected or more likely that children from the neighborhood will move out of state . . . than any other individual" and that "[s]hort-term occupants (including temporary guests) are not more likely to move out of Indiana than any other individual." [DE 63-6 ¶¶ 11, 15]. These opinions also suffer from a number of flaws, though.¹³ First, as discussed above, Plaintiffs did not reliably establish the likelihood that "any other individual" will move out of state. Second, particularly as to the short-term occupants, this opinion assumes that those individuals were citizens of Indiana at the time they occupied the class area, as it only addresses the likelihood that they will "move out of Indiana." However, these individuals, who could include short-term and transient guests, might not have been citizens of Indiana at the time. Additionally, these opinions are far too generalized to be of use in this context, since, as Dr. Wyner indicated, even a small discrepancy could lead to less than two-thirds of the class members being citizens of Indiana.¹⁴ In fact, though Dr. Cohen states in his rebuttal report that he has reviewed this affidavit, he never says that this is the type of information that a statistician would rely on, or even that he himself has relied on this affidavit or concluded from it that the survey respondents are representative of the class. He merely notes that he has "reviewed" the affidavit, which offers no insight into its value.

13 The agent offered other similarly generic and conclusory assertions, but they are unhelpful for similar reasons.

14 Moreover, the assertion that a short-term occupant of a home is no less likely to be a citizen of another state or to move to another state than an owner or renter of that home is difficult to credit. Likewise, it is difficult to believe that minors are no more likely to move out of a state than their parents. Plaintiffs also do not attempt to corroborate these broad assertions with evidence from their survey or any other statistical source.

Plaintiffs also submitted affidavits stating that, of the individuals in the public records list who did not respond to the survey, 81 to 85% of their last known addresses (of those that were available) were in Indiana. These materials likewise miss the mark for multiple reasons, though. First, non-response bias is evaluated by comparing known characteristics of respondents against known characteristics of non-respondents to determine whether they differ in a relevant way. [DE 50-7 p. 106; DE 71-1 ¶ 21]. Plaintiffs only presented the last known addresses of the non-respondents, and did not compare them against those of the respondents, so these numbers are not useful as a statistical matter to ascertain the presence or effect of non-response bias. Nor are they meaningful in their own right. As the Seventh Circuit has emphasized, citizenship cannot be proven with information like mailing addresses, even in the context of the local controversy exception. Sprint, 593 F.3d at 674; see also Hood v. Gilster-Mary Lee Corp., 785 F.3d 263 (8th Cir. 2015) (finding that last known addresses were insufficient to demonstrate the citizenship of nonrespondents). In addition, Plaintiffs offer no information as to how recent these last known addresses are, or how accurate they are in the first place. Thus, these materials add no material support to Plaintiffs' argument.

Plaintiffs also argue at length about how hard they tried to identify and contact all of the potential class members. If the admissibility and reliability of evidence were determined by how much effort a party put into creating it, Plaintiffs would certainly prevail. But the question here is whether the individuals that Plaintiffs succeeded in identifying were representative of the class as a whole,¹⁵ and whether the survey responses from a small subset of those individuals can reliably be extrapolated to the citizenship of the class. As to those questions, Dr. Cohen failed to even give them consideration, much less apply a reliable methodology in a reliable manner. At best, the analysis is incomplete, and at worst, it may be terribly misleading. In either case, the Court cannot conclude from this evidence that over two-thirds of the members of the expansive class were citizens of Indiana. Moreover, it is possible that Plaintiffs have given themselves an impossible task to reliably estimate the citizenship of this class given its scope and duration. Even if so, that would not relieve them of their evidentiary burden, regardless of the sincerity or exhaustiveness of their efforts.¹⁶

15 The Court does not suggest that a plaintiff must be able to identify every potential class member in order to appropriately define the universe of individuals to be surveyed, which Plaintiffs argue is the upshot of Johnson Controls' argument. Instead, they might have focused their efforts on a randomly selected sample of homes within the class area and endeavored to identify all potential class members from those homes and their citizenship, as suggested by Dr. Wyner, and extrapolated the citizenship of the class from that sample. There may be other options as well, or it may turn out that there are no reliable options. In either event, it is not acceptable to systematically exclude certain groups from the survey and fail to even consider the impact that may have on the results.

16 Evans v. Walter Indus., Inc., 449 F.3d 1159, 1166 (11th Cir. 2006) ("We understand that evidence of class citizenship might be difficult to produce in this case. That difficulty, however, is to a considerable degree a function of the composition of the class designed by plaintiffs. . . . We conclude that the evidence adduced by the plaintiffs wholly fails to present a credible estimate of the percentage of the plaintiff class who are citizens of Alabama."); see also Myrick, 764 F.3d at 665 ("[P]laintiffs contended that proving the citizenship of all class members is simply too expensive, so proof should be excused. Lawyers who launch class actions are not in a good position to complain about the expenses they entail; plaintiffs and their counsel must be prepared to meet them or be deemed inadequate representatives.").

Plaintiffs also argue at various points that Johnson Controls has not proven that the sample was not representative of the class. However, that gets the inquiry backwards. As the party that bears the burden of proof and as the proponent of the expert testimony, it is Plaintiffs' obligation to show that the sample is representative of the class and can be reliably extrapolated to the class as a whole. On that point, Johnson Controls is not merely raising a factual dispute, it is pointing out that Plaintiffs failed to consider this question altogether and thus failed to apply a reliable methodology. Thus, this is not a question of weight that rests on the factual underpinnings of a reliably applied methodology, but a threshold question of the adequacy of the methodology. Having admitted that these questions could invalidate his results through systemic bias, and having admitted that he has not even considered them, Dr. Cohen has essentially admitted that he did not apply a reliable statistical methodology to reach his results, so the Court cannot find that those results are admissible.

Plaintiffs also argue that, notwithstanding any shortcomings, their evidence at least satisfies the preponderance-of-the-evidence standard. In support, they rely on Keltner v. SunCoke Energy, Inc., No. 3:14-cv-1374, 2015 WL 3400234 (S.D. Ill. May 26, 2015), a suit that also involved environmental contamination of a neighborhood. There, the court held that, while the survey submitted by the plaintiffs to establish the citizenship of the class was "not without flaw, and additional information may have been helpful, the material presented by plaintiffs is sufficient to meet the preponderance of the evidence standard." Id. at *8. In Keltner, though, the proposed class consisted of residents of the affected area over the previous five years, meaning that "by definition, the proposed class members must reside or have recently resided in Illinois," the forum state. Id. at *7. Here, however, the time period spans over twenty years—more than four times longer—and Plaintiffs' own evidence shows that the ownership and occupancy of most of the homes changed many times over that period; some of the homes turned over ten or more times. [DE 1-17 p. 67-90]. In other words, the class here consists mostly of people who no longer live in the neighborhood, and who may not have lived there for a long time. In addition, the statistician there surveyed "a random sample" of the class and the court did not address the sorts of methodological omissions at issue here, and the results showed a higher rate of citizenship among the class than Plaintiffs' survey here. Accordingly, the Court does not find Keltner to be comparable or to justify a different result.

To the contrary, the Court finds that this case is more comparable to Hood v. Gilster-Mary Lee Corp., 785 F.3d 263 (8th Cir. 2015), where the plaintiffs did not use a representative sample or apply a valid sampling methodology, but merely attempted to contact each of the potential class members at their last known addresses. The survey respondents were overwhelmingly citizens of the forum state, and the last known addresses of the non-respondents were also mostly in the forum state. Nonetheless, the court held that the survey responses could not be extrapolated to the class as a whole given the significant potential for non-response bias, and that the last known addresses for the nonrespondents were insufficient to establish those individuals' citizenship. Accordingly, the court found that the plaintiffs had failed to meet their burden under the local controversy exception. Here, Plaintiffs attempted to contact a subset of the potential class at their last known phone numbers, but did not utilize statistical principles that would permit extrapolating the results of that survey to the class as a whole. Nor do the last-known addresses of the non-respondents bridge that gap. Thus, the result is the same.

Ultimately, it may be that over two-thirds of the members of the proposed class are citizens of Indiana. However, Plaintiffs have the burden of proving that fact, and the evidence they have provided gives the Court no reliable basis on which to conclude that they have done so. To the contrary, the Court finds that Plaintiffs' expert report is inadmissible under Rule 702 because it failed to reliably apply a reliable methodology. Moreover, even if the Court admitted the report, those same concerns as to the reliability of the methodology would so severely undermine the report's conclusions that the Court would be unable to find by a preponderance of the evidence that over two-thirds of the members of the proposed class are citizens of Indiana as of the pertinent date. That is particularly so in light of the opinions expressed in Dr. Wyner's reports that no statistically valid opinion as to the citizenship of the class as a whole can be drawn from Plaintiffs' submissions. Accordingly, the Court finds that Plaintiffs have failed to meet their burden of proof on this element, and that they have therefore failed to show that remand is warranted under the local controversy exception.

2. Substantial Defendant

Having failed to satisfy the previous element, Plaintiffs cannot qualify for remand under the local controversy exception. For completeness, though, the Court will also address the remaining disputed element, which is whether at least one defendant is a defendant "from whom significant relief is sought," "whose alleged conduct forms a significant basis for the claims asserted by the proposed plaintiff class," and "who is a citizen of the State in which the action was originally filed." 28 U.S.C. § 1332(d)(4)(A)(i)(II). Tocon is a citizen of Indiana,¹⁷ satisfying the third element, but the parties dispute whether it is a defendant from whom significant relief is sought and whose conduct forms a significant basis for the claims.

17 Unlike under the general diversity jurisdiction statute, a limited liability company is a citizen of the State where it has its principal place of business and the State under whose laws it is organized for purposes of the Class Action Fairness Act. 28 U.S.C. § 1332(d)(10). Tocon has its principal place of business in Indiana and is organized under the laws of Indiana, so it is a citizen of Indiana. --------

The first issue to address in resolving those questions is what materials the Court may rely on. Johnson Controls' notice of removal argues that these elements are met based on evidence extrinsic to the complaint, but its later filings insist that the Court can only consider the allegations in the complaint. Plaintiffs argued from the outset that only the allegations in the complaint matter for these purposes, but they have argued in the alternative that the extrinsic evidence supports their arguments, too, and both parties moved for leave to file additional extrinsic materials after the briefing concluded. Based on the plain language of the statute, the Court finds that these questions must be resolved based on the allegations in the complaint itself and without resort to extrinsic evidence.

Although the Seventh Circuit has yet to address this question, the Ninth Circuit has considered it at length, and the Court agrees with its analysis:

We hold that CAFA's language unambiguously directs the district court to look only to the complaint in deciding whether the criteria set forth in § 1332(d)(4)(A)(i)(II)(aa) and (bb) are satisfied.

The first criterion is whether "significant relief is sought" from a defendant who is a citizen of the state in which the suit is filed. 28 U.S.C. § 1332(d)(4)(A)(i)(II)(aa) (emphasis added). The word "sought" focuses attention on the plaintiff's claim for relief—that is, on what is "sought" in the complaint—rather than on what may or may not be proved by evidence. The second criterion is whether the defendant's "alleged conduct forms a significant basis for the claims asserted by the proposed plaintiff class." Id. § 1332(d)(4)(A)(i)(II)(bb) (emphasis added). Like the word "sought," the word "alleged" makes clear that the second criterion is based on what

is alleged in the complaint rather than on what may or may not be proved by evidence.

There is a revealing contrast between the language in subsections (aa) and (bb) and the language in subsection (cc). All three subsections specify criteria that must be satisfied before the local controversy exception to CAFA jurisdiction applies. Subsection (cc) requires that the defendant from whom relief is sought and whose alleged conduct is at issue be a defendant "who is a citizen of the State in which the action was originally filed." Id. § 1332(d)(4)(A)(i)(II)(cc) (emphasis added). Unlike the words "sought" and "alleged," used in subsections (aa) and (bb), the word "is," used in subsection (cc), indicates that an actual fact must be established.

Coleman v. Estes Exp. Lines, Inc., 631 F.3d 1010, 1015 (9th Cir. 2011). The court found that practical considerations support this interpretation as well:

If a determination whether "significant relief is sought" against the local defendant under subsection (aa) requires a factual determination about the respective ability of the various defendants to satisfy a judgment, that determination has the potential to be expensive and time-consuming. . . .

A factual determination whether the "alleged conduct" of the local defendant "forms a significant basis for the claims asserted" by plaintiffs under subsection (bb) is particularly likely to be expensive and time-consuming. Such a determination necessarily implicates the merits of the case. We see nothing in CAFA that indicates a congressional intention to turn a jurisdictional determination concerning the local defendant's "alleged conduct" into a mini-trial on the merits of the plaintiff's claims.

Coleman, 631 F.3d at 1016-17.

A related consequence of this rule is that a court may not look outside the complaint to determine whether the local defendant may be able to satisfy a judgment against it:

The statutory language is unambiguous, and a "defendant from whom significant relief is sought" does not mean a "defendant from whom significant relief may be obtained." There is nothing in the language of the statute that indicates Congress intended district courts to wade into the factual swamp of assessing the financial viability of a defendant as part of this preliminary consideration, which is one of six issues for a court to consider when deciding whether the "local controversy exception" is met.

Coffey v. Freeport McMoran Copper & Gold, 581 F.3d 1240, 1245 (10th Cir. 2009). Accordingly, Johnson Controls' arguments about Tocon's likely inability to satisfy any judgment are immaterial, so the Court denies its motion for leave to submit the order dismissing Tocon's bankruptcy petition. [DE 115]. Likewise, Plaintiffs' motion for leave to submit newly discovered evidence on these questions is denied, as that evidence cannot be considered at this stage. [DE 90].

The Court therefore turns to the allegations in the complaint to determine whether Tocon is a defendant from whom significant relief is sought and whose conduct forms a significant basis for the claims. Both of these elements turn on whether the local defendant is "significant," which is an ambiguous term that is not otherwise defined in the statute. Other provisions of the Class Action Fairness Act provide a useful comparison, though, as they refer to "primary defendants," whereas the local controversy exception requires only that the relief sought and the defendant's conduct be "significant." Compare 28 U.S.C. § 1332(d)(3), (d)(4)(B), (d)(5)(A), with § 1332(d)(4)(A)(i)(II). These provisions thus suggest that a local defendant can be secondary to a foreign defendant yet still be significant. The Third Circuit has held, however, that it is not enough for a local defendant to be of "more than trivial or of no importance." Kaufman v. Allstate N.J. Ins. Co., 561 F.3d 144, 157 (3d Cir. 2009). Rather, the "local defendant's alleged conduct must be an important ground for the asserted claims in view of the alleged conduct of all the Defendants." Id. This entails a comparative approach, considering the significance of the local defendant in the context of the action as a whole, since the overarching question is whether the action is local such that it should be litigated in state rather than federal courts. Benko v. Quality Loan Serv. Corp., 789 F.3d 1111, 1118 (9th Cir. 2015).

Here, Johnson Controls argues that Tocon is not a significant defendant because Tocon did not purchase the site until 2007 and did not use TCE during its operation of the plant at the site. Johnson Controls argues that because Tocon had little if any role in causing the contamination, it must not be a significant defendant in these claims that arise out of the contamination. This argument misconstrues the nature of Plaintiffs' claims, though. The conduct that Plaintiffs argue gives rise to liability is that the defendants both knew about the contamination, failed to adequately remediate it, allowed it to migrate into the Plaintiffs' properties, and failed to warn Plaintiffs of the presence and dangers of that contamination. For example, the complaint alleges that the defendants are each liable for "allowing or causing Contamination to migrate onto the Plaintiffs' properties," and that Tocon in particular "purchased the plant from [Johnson Controls] with full knowledge that it was contaminated, but did nothing to stop the contamination from flowing offsite onto the Plaintiffs['] properties, and took no steps to warn the Plaintiffs about the dangers of the contamination." [DE 4-1 p. 1-2, ¶ 119]. The complaint further alleges that Tocon "contributed to the release of TCE and other hazardous chemicals from the Plant into the Neighborhood by failing to prevent further off-site migration of Contamination from the Plant, by failing to timely remediate the off-site Contamination emanating from the Plant, and by contributing additional waste to the Neighborhood through its improper storage of solid and potentially hazardous waste." [Id. ¶ 139]. Moreover, in support of punitive damages, the complaint alleges that Tocon "took no steps to stop the contaminant migration or to warn the residents who were living in the homes," and that the defendants both "exhibited reckless disregard for the Plaintiffs' health by concealing the extent of Contamination from IDEM and Plaintiffs, and by failing to take sufficient or effective action to remove the Contamination from the Neighborhood or to adequately notify or warn current and former residents of the Contamination vapors." [Id. ¶¶ 151, 153].

In short, the crux of Plaintiffs' claims is not that the plant itself was contaminated, but that the TCE was allowed to migrate from the plant onto Plaintiffs' properties, that it was not investigated and remediated, and that the defendants failed to warn the Plaintiffs of the existence of the contamination and its dangers. Even if Tocon did not place the TCE in the ground in the first place, its conduct still forms a significant basis for the claims as thus framed. Johnson Controls is more significant as a defendant, as it owned the property for longer and TCE had begun migrating into the neighborhood before Tocon purchased the property. But Tocon is alleged to have allowed the migration to continue and also to have failed to warn Plaintiffs of the hazards in their neighborhood. Moreover, Tocon is the current owner of the property, and Plaintiffs allege that Tocon has continued to use the property in ways that will cause additional contamination and environmental dangers. Under those circumstances, Tocon's conduct forms a significant basis for these claims even in comparison to Johnson Controls'. Additionally, the complaint seeks compensatory and punitive damages equally against each defendant, and asks that the injunction requiring complete remediation be imposed against both defendants. Though Johnson Controls' exposure under these claims is likely greater than Tocon's since it owned the property for longer, that does not render Tocon insignificant, either in terms of the claims asserted or the relief sought. See Benko, 789 F.3d at 1118-19 (finding that the local defendant was significant even where it was one of six defendants and was responsible for 15 to 20% of the activities in question).

The Ninth Circuit's recent opinion in Allen v. Boeing Co., 821 F.3d 1111 (9th Cir. 2016) addressed a similar situation. There, the plaintiffs asserted environmental claims arising out of groundwater contamination from a Boeing plant. The only local defendant, Landau Associates, was not alleged to have ever caused any contamination at the plant. Rather, it was retained by Boeing as a contractor to investigate and remediate contamination that had accumulated over the previous decades of Boeing's operations at the plant. The plaintiffs asserted claims against both Boeing and Landau for their failure to adequately investigate and remediate the contamination and their failure to warn the property owners and residents of the hazards, which mirrors the claims here. Boeing argued that Landau could not have been a significant defendant because it never created any contamination, but the court rejected that argument:

It is true that Boeing's activities over several decades created the hazardous plumes. However, it does not follow that Boeing's liability (if any) for creating the pollutants necessarily dwarfs Plaintiffs' claims against Landau. The gist of Plaintiffs' claims is that the movement of the volatile organic chemicals off Boeing's property caused them harm, not that the existence of the chemicals at the Plant harmed them. Plaintiffs allege that Landau undertook in 2002 to investigate, remediate, and clean up the hazardous materials moving off Boeing's property and failed to take reasonable steps to do so. If Landau is shown to have failed, for more than a decade, to remediate the spreading toxic chemical plumes, its liability could be as great as Boeing's. Thus, the fact that Boeing created the pollution does not in itself render insignificant the damages caused by Landau's alleged failure to investigate and remediate the spreading pollution.

Allen, 821 F.3d at 1118-19. For similar reasons, Tocon's failures to stop the spread of contamination, to remediate the contamination, and to notify Plaintiffs of the dangers of the contamination make it a significant defendant.

Accordingly, based on the allegations in the complaint, the Court finds that Tocon is a local defendant whose conduct forms a significant basis for the claims and from whom significant relief is sought. Nonetheless, because Plaintiffs have failed to show that greater than two-thirds of the members of the proposed class are citizens of Indiana, they have failed to satisfy each of the elements of the local controversy exception. The motion to remand on that basis is therefore denied.

III. CONCLUSION

Plaintiffs' consolidated motion to remand [DE 21] is DENIED. The Court GRANTS Johnson Controls' motion to strike Plaintiffs' expert report [DE 50], and DENIES Plaintiffs' motion for leave to file newly discovered evidence [DE 90] and Johnson Controls' motion for leave to submit an order of the bankruptcy court [DE 115].

SO ORDERED.

ENTERED: July 11, 2016

/s/ JON E. DEGUILIO

Judge

United States District Court