ORDER

J.P. STADTMUELLER, District Judge.

This is a class action alleging that Defendants have marketed their dog foods as being natural and of high-quality, and sold them at a premium price, when their advertisements were misleading at best, meaning that the products' price was unfairly inflated. (Docket #41). Defendants have filed a motion for summary judgment, seeking dismissal of the case in its entirety. (Docket #46). Plaintiff has filed his own motion for class certification. (Docket #50). The parties have also moved to exclude the opinions of various experts pursuant to Federal Rule of Evidence 702 and Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993). (Docket #71, #81, #96, #107, #121, and #126). The Court will herein address only one such motion— Defendants' motion to exclude two of Plaintiff's damages experts, Jon Krosnick ("Krosnick") and Colin Weir ("Weir"). (Docket #71). It appears to the Court that the motion is pivotal to the case and may drastically change the parties' approaches to the litigation moving forward.

Rule 702 provides that parties may propound opinion testimony from a witness if that witness is qualified as an expert by way of knowledge, skill, or experience. Fed. Evid. 702. To warrant admission of such opinion testimony, the propounding party must establish that:

Id. Put another way, the Court "must determine whether the witness is qualified; whether the expert's methodology is scientifically reliable; and whether the testimony will assist the trier of fact to understand the evidence or to determine a fact in issue." Myers v. Ill. Cent. R.R. Co., 629 F.3d 639, 644 (7th Cir. 2010) (quotation omitted).

This analysis reflects the need for the Court to act as a gatekeeper for the admission of expert testimony, because such testimony "can be both powerful and quite misleading because of the difficulty in evaluating it." Daubert, 509 U.S. at 595 (quotation omitted). The Court's inquiry here is flexible and it need not find all factors present or absent to decide admissibility. Krik v. Exxon Mobil Corp., 870 F.3d 669, 674 (7th Cir. 2017). But the Court should remain focused on the expert's principles and methodology, not his conclusions or the facts underlying his opinions; those go to the weight of the testimony and are reserved for the fact-finder's evaluation. Gopalratnam v. Hewlett-Packard Co., 877 F.3d 771, 780-81 (7th Cir. 2017). The party propounding the expert testimony bears the burden to convince the Court, by a preponderance of the evidence, that the testimony satisfies Daubert and Rule 702. Lewis v. CITGO Petroleum Corp., 561 F.3d 698, 705 (7th Cir. 2009).

Krosnick is an expert in creating and conducting surveys. In this case, he was hired by Plaintiff to conduct two. The first, called the damages survey, was aimed to test how consumers' purchasing decisions would change when the allegedly misleading statements on Defendants' product packaging were corrected. The offending statements, which include the phrases "Biologically Appropriate," "Fresh Regional Ingredients," and "Never Outsourced," are referred to in shorthand as BAFRINO.

The damages survey began by showing respondents unaltered photos of two bags of Defendants' dog food. It then revealed zero to eight "corrective statements" which, as the name implies, were designed to correct the misleading nature of each BAFRINO statement. The corrective statements were as follows:

(Docket #73-1 at 42-43). Krosnick founded the statements' content in either governmental sources or information obtained from Defendants.

After reading the packages and seeing the corrective statements, respondents were asked a series of questions, including whether they had owned a dog at all. If they did, the respondents were then asked at what price they would purchase the products, whether at, above, or below market price. Krosnick determined that those who saw more corrective statements were less willing to pay the market price of the products; respondents who saw one statement thought they would pay ten percent less, and those who read all eight would pay almost sixty percent less.

Krosnick's data was provided to Weir, an economist, to calculate damages on a classwide basis. Weir took the sales figures for the subject products in Wisconsin for the class period and multiplied them by the diminution-in-value percentage obtained from respondents who had seen seven or eight of the corrective statements. Weir determined that the class damages, namely their overpayment to Defendants based on the unjustified premium price they were charged, stands at almost $9 million.

The second survey tested consumers' reaction to the potential presence of pentobarbital in the products. It was conducted similarly to the damages survey, except Krosnick concluded that there was no need to analyze diminution-in-value; it is entirely illegal to sell pentobarbital-laced dog food. Weir calculated that consumers would be entitled to approximately $2.25 million as a full refund for the purchase price of the products which had a risk of pentobarbital. Because they are not materially different for present purposes, the Court will refer to the surveys in the singular as the Survey.

The Court concludes that Krosnick and his survey evidence must be excluded under the Daubert and FRE 702 standards. It begins by noting a preliminary issue with Plaintiff's claims at this juncture. Five of the eight corrective statements apply to allegations regarding heavy metal levels in Defendants' products. The Court has already dismissed the heavy metal allegations in deciding Defendant's motion to dismiss. (Docket #39). Plaintiff nevertheless re-pleaded comparable allegations in his operative amended complaint. (Docket #41 at 32-39). Plaintiff's only avenue to bring heavy metals back into this case is a successful appeal reversing the Court's decision on the motion to dismiss. His defiance of the Court's orders is neither advisable nor an effective strategy. Even if the Survey were not entirely inadmissible, the portions relating to heavy metals would certainly be excluded.

Beyond this concern, the Court detects four major flaws in the Survey itself. First and foremost, the Survey does not actually test Plaintiff's theory of the case. In the motion to dismiss, the Court observed that certain theories may be viable in this case, including that the food was rendered dangerous by the alleged contaminants,1 or that the food was over-priced in comparison to other pet foods which had greater or lesser concentrations of contaminants. (Docket #39 at 5-8). In his briefing on the pending motions, Plaintiff has rejected either of these avenues:

(Docket #82 at 6). Thus, Plaintiff has established his theory as follows. The foods are advertised as healthful and natural via the BAFRINO statements. However, they contain undisclosed levels of contaminants. Making the BAFRINO statements while failing to disclose the contaminants is an attempt to mislead consumers into buying Defendants' products at a premium price which, in their contaminated state, the products should not command. Put simply, Plaintiff feels that the BAFRINO statements improperly inflate the products' prices. He seeks to recover the difference between his actual purchase price and the price he would have paid had he known the BAFRINO statements were less-than-truthful. This is, in turn, Krosnick's goal in the Survey.

But the Survey does not test the effect of the BAFRINO statements on consumers' purchasing decisions. See Comcast Corp. v. Behrend, 569 U.S. 27, 35 (2013) (a model for class action damages must measure only those damages attributable to the theory of liability upon which the class is certified). To do that, the respondents should have first been shown unaltered bags of food and asked what price they would pay for that product. Next, they should have been shown bags with all of the BAFRINO statements omitted and have the same price question put to them. Plaintiff could further refine the results by removing some but not all of the statements in various combinations. These results would demonstrate the nature and magnitude of the misleading effect of the BAFRINO statements.

Instead, the Survey tests the irrelevant matter of how consumers would react knowing what particular contaminants may be present in the products. But Plaintiff's theory does not turn on what is actually in the food or what effect it has on the healthfulness of the products. The sole focus, from the consumer's perspective, is the misleading effect of the BAFRINO statements. Precisely how those statements are proved to be misleading is an entirely separate matter. In other words, the Survey conflates liability questions—proof that the BAFRINO statements were in fact false or misleading—with the matter of damages—what consumers would pay in the absence of the BAFRINO statements.2

Plaintiff's counsel should not be surprised by the Court's position. In another of the many identical cases filed by the same set of counsel across the country, one court recently found Krosnick's damages model wholly deficient. Jennifer Reitman et al. v. Champion Petfoods USA, Inc. et al., No. 2:18-CV-1736-DOC-JPR (C.D. Cal.), (Docket #115-1 at 21-25). Reitman's logic mirrors that of this Court, namely that the corrective statements do not align with Plaintiff's theory of the case. Id. at 24-25. Though Plaintiff's counsel has had the benefit of the Reitman decision for some time now, they have made no effort to reevaluate their use of Krosnick's testimony in this case or otherwise buttress their damages theory.3

The Survey's second error is that it relies on the flawed notion that any choice of wording for the corrective statements could actually "correct" the misleading effect of the BAFRINO statements. Language does not work that way; it is not a mathematical equation that can be balanced. This is true whether the statements are drafted by Plaintiff's counsel to be inflammatory, by Defendants' counsel to be sympathetic, or by some neutral third-party. Any attempt to "correct" the BAFRINO statements with additional statements emphasizes the basic defect in the Survey's premise, that is a lack of focus on the effect of the BAFRINO statements themselves.4

It is nevertheless abundantly clear why Plaintiff chose to structure the Survey in this fashion—to foment bias. The "corrective" statements are anything but. They are not a neutral explanation of the presence of contaminants in the products. Instead, the corrective statements are supremely provocative, intentionally preying upon consumers' fears of harm to their pets. This is true not only for the heavy metal, BPA, and pentobarbital statements, but also for the non-fresh and non-regional statements as well. The ready inference of the use of such ingredients is that they are less healthy and more dangerous than fresh and local ingredients. The corrective statements run directly counter to Plaintiff's theory of the case. He cannot in one breath disclaim any desire to attack the safety of the foods while also poisoning consumers against Defendants by advancing unmistakable implications to that effect. Plaintiff cannot have his cake and eat it too.5 Thus, even if the Court is wrong about the synergy between the Survey and Plaintiff's theory of the case, the Survey is biased beyond repair.

The third and fourth flaws in the Survey relate to its methodology. The Survey did not even attempt to sample the population which would form a class in this case should one be certified. Plaintiff's third amended complaint asks the Court to certify a class of Wisconsin residents who purchased Defendants' products in Wisconsin from July 1, 2013 to the present. (Docket #41 at 72). But the Survey was targeted to a broad population consisting of all American adults. Out of the almost seven thousand people surveyed, less than one percent actually fall within the proposed class. Many of the respondents did not even own a dog or had ever purchased dog food. Among the dog-owners, there was no focus on those who would actually be in the market for Defendants' premium-priced products. And as to those who did purchase those products, only a few hundred did so in the relevant time frame. The Survey comes nowhere close to obtaining a significant sample of the population harmed by Defendants' BAFRINO alleged lies, and who would stand to recover if the class was certified.

Plaintiff responds in two ways. First, Krosnick opined that it would have been inappropriate to limit the survey population to just those who purchased Defendants' products, as they were the ones who fell for the deception. This belief is based on forty-year-old FTC guidelines which generally address consumer deception. (Docket #80-10). This leads to Plaintiff's second response, which is to claim that Defendants' rebuttal expert, Dominique Hanssens ("Hanssens") engaged in speculation to critique Krosnick's choice of survey population. Hanssens says that Krosnick should have accounted for potential material differences between the populations of American adults versus those who would actually buy Defendants' products. Plaintiff says that this is speculative, and that Defendants lack data to show that "the responses of people that had purchased [Defendants'] food in the past differed from the responses of those that had not, or that the responses of people from Wisconsin differed from the responses of people from other states." (Docket #78 at 24). But this attack improperly reverses the Daubert burden. It is Plaintiff's obligation to establish that the survey population was appropriate, and so he should have produced this data. His failure to do so means that the Court is not equipped to compare the Survey respondents to the proposed class. The final issue with the Survey is the lack of a meaningful control group. Krosnick claims the Survey contains multiple control groups, one for each corrective statement. He states that among all of the random showings of the corrective statements, about half of respondents saw any particular statement, and half did not. Thus, these groups can be used as test and control groups for each statement. Two flaws are apparent in this design. First, given the many problems with the corrective statement approach discussed above, the only true control group would be respondents who saw no corrective statements. Only twenty-three people, less than half of one-percent of respondents, comprise this group. Moreover, even if Krosnick had obtained a more sizable control group, he never attempted to analyze that group, or his own "control" groups, to provide context for the Survey results. Without a valid control group, the Court cannot conclude that the Survey is reliable or helpful to the factfinder. See DeKoven v. Plaza Assoc., 599 F.3d 578, 581-82 (7th Cir. 2010); Muha v. Encore Receivables Mgmt., Inc., 558 F.3d 623, 626 (7th Cir. 2009).

In light of the foregoing, the Court must exclude Krosnick's opinion and his Survey evidence. Weir's calculations are entirely founded on the Survey results. With the exclusion of those results from this case, Weir's opinion must go as well. Without these two opinions, the parties' approach to class certification, summary judgment, and the litigation generally will likely be significantly altered. The Court will, therefore, deny the pending motions for summary judgment and class certification without prejudice. The Court will also grant various pending motions to restrict documents. (Docket #40, #49, #59, #77, and #86).

Accordingly,

IT IS ORDERED that Defendants' motion to exclude the opinions of Jon Krosnick and Colin Weir (Docket #71) be and the same is hereby GRANTED, and the opinions of Jon Krosnick and Colin Weir be and the same are hereby EXCLUDED from the evidence in this case;

IT IS FURTHER ORDERED that Defendants' motion for summary judgment (Docket #46) and Plaintiff's motion for class certification (Docket #50) be and the same are hereby DENIED without prejudice; and

IT IS FURTHER ORDERED that the parties' motions to restrict documents (Docket #40, #49, #59, #77, and #86) be and the same are hereby GRANTED.