1 WO 2 3 4 5 6 IN THE UNITED STATES DISTRICT COURT 7 FOR THE DISTRICT OF ARIZONA 8 9 Jodi Moshier, et al., No. CV-23-00225-PHX-DLR 10 Plaintiffs, Consolidated with: Case No. CV-23-00822-PHX-DLR 11 v. ORDER 12 Safeco Insurance Company of America, 13 Defendant. 14 15 Before the Court is Plaintiffs’ Motion to Compel (Doc. 46), which is fully 16 briefed (Docs. 49, 51). For reasons below, the motion is granted. 17 I. Background 18 Plaintiffs allege that Defendant Safeco Insurance Company of America (“Safeco”) 19 failed in the past to stack Uninsured Motorist Coverage (“UM”) and Underinsured Motorist 20 Coverage (“UIM”) limits as required by Arizona’s Uninsured Motorist Act. Plaintiffs claim 21 that the insurance benefits paid by Safeco to every member of the putative class were 22 wrongfully capped at the maximum amount of the single-vehicle UM/UIM benefits. 23 Plaintiffs have employed experts to model damages. Plaintiffs’ experts need a 24 limited sample of data regarding Safeco’s past UM/UIM settlement payments along with 25 third-party bodily injury settlement payments to demonstrate how such damages can be 26 modeled on a class-wide basis. At issue are Plaintiffs’ Requests for Production (“RFP”) 27 Nos. 40, 41, and 43, in which Plaintiffs request that Safeco produce all data reflecting the 28 amounts of all private passenger automobile and motorcycle (1) UM/UIM claim 1 settlements paid by Safeco for the past six years (RFP 40), (2) liability insurance claim 2 settlements paid to third parties by Safeco for the past six years (RFP 41), and (3) liability 3 claim settlements for the past six years (RFP 43). In their motion to compel, Plaintiffs state 4 that “[t]o limit the burdens associated with gathering and producing such data, Plaintiffs 5 would accept a limited sample of the data at the class certification stage (approximately 6 200 rows of data per coverage type).” (Doc. 46 at 10.) Plaintiffs argue that the data is 7 relevant and necessary for the presentation of their damages model and that it can be 8 produced by Safeco with minimal effort. 9 Safeco objects to producing data about the settlement payments primarily on 10 relevance grounds. Safeco argues that the data Plaintiffs seek could not show that Safeco 11 breached its contract with any putative class member or improperly handled any putative 12 class member’s claim because the data sought is not data about any putative class members’ 13 claims. Safeco argues that each putative class member must prove the damages they would 14 have been legally entitled to recover under the policy and then prove that Safeco underpaid 15 their UM or UIM benefits. Relying on Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338 (2011), 16 Safeco argues that Plaintiffs cannot try the case “by formula” by using other people’s 17 injuries to prove liability or damages and that the use of models prepared by experts based 18 on past settlements would deny Safeco the opportunity to challenge the reasonableness of 19 the link of the claimed damages. Citing Tyson Foods, Inc. v. Bouaphakeo, 577 U.S. 442 20 (2016), Safeco further argues that this is not an appropriate case for the use of 21 representative evidence to prove damages. 22 Even if the Court determines that non-claimant specific data is relevant to a 23 claimant’s claim, Safeco argues the motion should be denied because the data sought is not 24 representative of any plaintiff’s claim or any putative member’s injury. Safeco points out 25 that personal settlements do not define the amount of a claimant’s legally recoverable 26 injury because evaluating personal injury claims is not an exact science. 27 Safeco also argues that the liability settlement data is not relevant because it 28 involves fundamentally different types of claims than UM and UIM claims. UM and UIM 1 claims are first-party claims while liability claims are third-party claims. 2 Finally, Safeco argues that Plaintiffs are not required to prove their case through 3 common proof at the class certification stage. 4 Aside from relevance, Safeco argues that it should not be required to produce the 5 data because such production would impose an undue burden and Safeco has an obligation 6 under Arizona law to keep the requested information confidential. 7 II. Analysis 8 A. Relevance 9 Information is relevant if it has “‘any tendency’ to make a fact in dispute ‘more or 10 less probable’.” Cont’l Cirs. LLC. v. Intel Corp., 435 F. Supp. 3d 1014, 1018 (D. Ariz. 11 2020) (quoting Fed. R. Civ. P. 401). At the pre-class-certification stage, requested 12 information that “is likely to produce substantiation of the class allegations” is relevant. 13 Mantolete v. Bolger, 767 F.2d 1416, 1424 (9th Cir. 1985). In determining the relevance of 14 pre-class-certification discovery, the Court does not decide whether the requirements for 15 class certification are met, the proposed damages model satisfies Daubert, Plaintiffs will 16 prevail on the common question presented, or the damages model will be accepted at trial. 17 These questions are reserved for later stages of the litigation. 18 Plaintiffs’ data discovery request is geared toward proof of damages. (Doc. 46 at 7.) 19 Safeco’s relevance objections are premised on the assumption that Plaintiffs’ method of 20 proving class-wide damages must calculate damages for each individual class member— 21 the “common proof of individual damages” approach. Safeco argues that there is no precise 22 method of calculating personal injury damages, that each class member’s injuries are 23 separate and individualized, and the amounts it paid to settle other unrelated claims cannot 24 be used to prove what Safeco owes these Plaintiffs. Plaintiffs, however, intend to show that 25 they have a model that can determine aggregate damages through the representative 26 approach. Rather than totaling individualized damages of each class member, Plaintiffs 27 claim that their experts have created a model to calculate class-wide damages in the 28 aggregate. 1 There is no general prohibition against proving class-wide damages through the 2 representative approach, where statistical evidence is used to prove class claims. “The use 3 of aggregate damages calculations is well established in federal court and implied by the 4 very existence of the class action mechanism itself.” In re Pharm. Indus. Average 5 Wholesale Price Litig., 582 F.3d 156, 197 (1st Cir. 2009). Plaintiffs concede that there are 6 individual variations in damages among class members and that there are some uninjured 7 class members. But individualized damages do not necessarily destroy predominance, see 8 Olean Wholesale Grocery Corp., Inc. v. Bumble Bee Foods LLC, 31 F. 4th 651, 669 (9th 9 Cir. 2022) (en banc), and at the class certification stage, Plaintiffs’ calculations need not 10 be exact, Comcast Corp. v. Behrend, 569 U.S. 27, 35 (2013). Plaintiffs claim that their 11 model can reasonably account for Safeco’s monetary liability. In its response to Plaintiffs’ 12 eventual motion for class certification, Safeco may of course produce its own model that 13 individually tallies damages. And after Plaintiffs present their model, the Court will need 14 to decide if Plaintiffs’ aggregate damages model establishes that damages are capable of 15 measurement on a class-wide basis. Comcast, 569 U.S. at 34. At that time, the Court will 16 take a close look at whether the presence of individualized damages questions undermine 17 predominance. Id. But the question the Court must answer today is merely whether the data 18 Plaintiffs seek is likely to produce substantiation of the class allegations. 19 Plaintiffs represent that their experts need Safeco’s data concerning its settlement 20 of past bodily injury and UM/UIM claims to calculate class-wide damages using the 21 representative approach. Plaintiffs represent that their expert’s model will use the 22 distribution of past settlement payments that were not reduced by any cap to estimate the 23 distribution curve for all the class members’ claims and calculate the probable aggregate 24 of settlements that would have been made to the class under what Plaintiffs claim are the 25 correct stacked policy limits. With that data, Plaintiffs’ experts can utilize established 26 statistical tools such as Kaplan-Meier curves and Weibull mixture models and attempt to 27 show that the aggregate damages can be calculated for all insureds whose benefits were 28 limited by caps. 1 Plaintiffs further have distinguished the two primary cases relied on by Safeco in its || objection to the use of statistical evidence to prove damages. Tyson Foods did not establish || categorical rule against statistical or other representative evidence to prove class claims. 4|| Plaintiffs’ use of statistical evidence is being employed here to establish a measure of 5 || damages, whereas in Tyson, the statistical evidence was employed to prove a breach of 6|| liability and causation. Tyson made clear that “[w]hether and when statistical evidence .. . can be used to establish classwide liability depends on the purpose for which the evidence 8 || is being introduced[.]” 577 U.S. at 1040. Representative proof of aggregate damages is a || well-accepted method of showing predominance. 7yson did not change that. And in Dukes, || the plaintiffs sought to certify an injunction class pursuant to Rule 23(b)(2). Damages were 11 || not an issue. Plaintiffs, here, have made clear that their use of statistical models goes only || to damages. The statistical models involved in Dukes addressed liability at the class 13 || certification stage, after discovery of data had been obtained. Although Dukes did not approve of the statistical models that had been presented regarding liability, it did not 15 || prohibit the discovery of the data upon which the models relied. 16 For these reasons, the Court finds Plaintiffs’ requested data relevant. 17 B. Burden 18 Safeco argues that there “is a burden to producing the requested data.” (Doc. 49 at 13.) However, Safeco does not describe the burden. Safeco supplies no affidavit from a 20 || custodian of records describing the anticipated labor or costs to retrieve the data. As for □□ confidentiality concerns, those can be resolved through a protective order, if appropriate. || Safeco has not established an unreasonable or disproportionate burden in producing the 23 || data. 24 IT IS ORDERED that Plaintiff's Motion to Compel (Doc. 46) is GRANTED. 25 Dated this 15th day of October, 2024. 26 - b 4 bla 28 Do . Rayes Semior United States District Judge _5-