Navarro v. Procter & Gamble Co.

Full title: ANNETTE NAVARRO, et al., Plaintiffs, v. PROCTER & GAMBLE COMPANY, et al.…

Court: UNITED STATES DISTRICT COURT SOUTHERN DISTRICT OF OHIO WESTERN DIVISION

Date published: Mar 8, 2021

Case No. 1:17-cv-406 (S.D. Ohio Mar. 8, 2021)

Opinion

Case No. 1:17-cv-406

03-08-2021

ANNETTE NAVARRO, et al., Plaintiffs, v. PROCTER & GAMBLE COMPANY, et al., Defendants.

JUDGE DOUGLAS R. COLE

OPINION AND ORDER

This matter comes before the Court on three Motions in Limine. The first is Defendants' motion to exclude Jeffrey Sedlik, Navarro's expert on actual damages and fraud damages. (Doc. 214). The remaining two motions are Navarro's. One motion seeks to exclude Robert Zeithammer, Defendants' expert on profit attribution. (Doc. 218). In the other motion, Navarro seeks to fully exclude the expert opinion of Steven Hazel, Defendants' omnibus damages expert, and to partially exclude the expert opinion of Doug Bania, Defendants' rebuttal expert to Sedlik. (Doc. 217). For the reasons below, the Court GRANTS IN PART AND DENIES IN PART Defendants' Motion in Limine to Exclude Jeffrey Sedlik (Doc. 214); DENIES Navarro's Motion to Exclude Robert Zeithammer (Doc. 218); and GRANTS IN PART AND DENIES IN PART Navarro's Motion to Exclude Steven Hazel and Partially Exclude Doug Bania (Doc. 217).

BACKGROUND

This is an action for copyright infringement. (See Fifth Am. Compl., Doc. 133, #4054-4109¹). As more fully explained in the Court's recent decision on the parties' cross-motions for summary judgment (Op. on Cross Mot. for Summ. J., Doc. 247), Navarro claims that Defendants willfully infringed her copyrights in certain photos that she provided to P&G for use pursuant to licensing arrangements between the parties. (See Fifth Am. Compl., Doc. 133 at ¶¶ 186-87, #4090-91). According to Navarro, P&G continued to use the photographs after the licenses expired, and also used them on products and in geographic areas outside those covered by the parties' agreements. (Id. at ¶ 246, #4104). Such use, she says, constituted willful infringement of her copyrights. (Id. at ¶ 1, #4055). She sued P&G and Walmart (who sells P&G products that bear the allegedly infringing photographs) in this Court. (Id. at ¶¶ 4-5, #4055-56). The matter was originally pending before another Judge on this Court, but in December 2019, it was transferred to the undersigned Judge.

1 Refers to Page ID Number.

One element of a copyright infringement case that a plaintiff must prove at trial is damages (at least if the plaintiff is seeking damages other than statutory damages). Apart from statutory damages, a plaintiff in a copyright infringement action may seek two different types of damages: actual damages and profit-based damages. 17 U.S.C. § 504(b). The first of those, actual damages, reflects the straightforward concept that an infringer should pay the copyright owner compensation for amounts that the copyright owner lost due to the infringement. Thoroughbred Software Int'l, Inc. v. Dice Corp., 488 F.3d 352, 358 (6th Cir. 2007) (conceptualizing actual damages as "the amount [the copyright owner] would have received but for [the infringer's] unlawful copying") (quoting Thoroughbred Software Int'l, Inc. v. Dice Corp., 439 F. Supp. 2d 758, 772 (E.D. Mich. 2006)). Often this is the value that the infringer would have paid to use the copyrighted material legally—an amount sometimes calculated as a "reasonable license fee" based on a hypothetical "willing-seller/willing-buyer" informed by norms and customs in the relevant industry.² Id. at 359 ("[A]ctual damages may include in appropriate cases the reasonable license fee[.]") (quoting Davis v. Gap, Inc., 246 F.3d 152, 167 (2d Cir. 2001)).

2 Other times, actual damages are equal to the amount of profits that the copyright owner lost as a result of the infringement. For example, a copyright owner might recover actual damages in the amount of lost business or lost market value as a result of consumers purchasing the infringer's product. See, e.g., Robert R. Jones Assoc., Inc. v. Nino Homes, 858 F.2d 274, 280-81 (6th Cir. 1988) (explaining that the measure of actual damages is the profits the copyright owner would have made on houses it would have sold but for its competitor's infringement).

But infringement does not happen in a vacuum. Infringement does not merely harm the copyright owner, but also may benefit the infringer in the infringer's own business to an even greater extent. Profit-based damages, the second form of damages under § 504(b), force the infringer to disgorge the profits it realized from the infringement to the extent that such profits exceed the copyright owner's actual damages. Goldman v. Healthcare Mgmt. Sys., Inc., 559 F. Supp. 2d 853, 865-66 (W.D. Mich. 2008) (explaining that a plaintiff can only recover profit-based damages in excess of actual damages).

The calculation of profit-based damages is a two-step process. First, the copyright owner must identify the portion of an infringer's gross revenues that bears a "reasonable relationship" to the alleged infringement. Balsley v. LFP, Inc., 691 F.3d 747, 767-68 (6th Cir. 2012). Once the copyright owner has identified the appropriate revenues, the burden shifts to the infringer to demonstrate what part, if any, of those revenues are not profits attributable to the infringement. Id. at 769. The infringer can do so in two ways—first by pointing to the expenses it incurred in generating the revenues, and then second, by allocating any remaining amount (the profit) among infringing and non-infringing activity. Id. at 767 (quoting 17 U.S.C. § 504(b)).

The experts at issue all offer opinions as to different damages topics. Jeff Sedlik, the subject of Defendants' only motion here, is Navarro's actual damages expert. (Sedlik Mot., Doc. 214). Sedlik also proposes to testify as to the damages for Navarro's common law fraud claim. The subject of the second motion at issue is Robert Zeithammer, who is Defendants' expert on profit-based damages for Navarro's copyright infringement claim. (Zeithammer Mot., Doc. 218). Zeithammer offers opinions as to the second step of the profit-based damages bipartite framework. That is, his opinions are intended to assist in identifying what portion of P&G's profits are attributable to Navarro's photographs. Rounding out the lineup is Navarro's motion to exclude Stephen Hazel and Doug Bania. (Doc. 217). Hazel is Defendants' omnibus damages expert who offers an opinion both as to actual damages and profit-attribution. Doug Bania, in contrast, offers only a limited rebuttal opinion to Sedlik, Navarro's expert on actual damages.

All three motions in limine directed at these experts are fully briefed and ripe for review.

LAW AND ANALYSIS

Federal Rule of Evidence 702 allows certain witnesses with specialized knowledge to testify as experts at trial. Fed. R. Evid. 702. But admitting expert testimony is not a decision a court should undertake lightly, as juries tend to place extra weight on expert opinions. Daubert v. Merrell Dow Pharm., 509 U.S. 579, 589 (1993) (noting that expert opinions can be "powerful"). Courts serve an important "gatekeeper" role when it comes to expert witnesses, ensuring that the jury does not encounter misleading "junk science." Kumho Tire Co., Ltd. v. Carmichael, 526 U.S. 137, 141 (1999) (acknowledging that district courts have a "gatekeeping" obligation as to all types of expert testimony); Best v. Lowe's Home Ctr., Inc., 563 F.3d 171, 177 (6th Cir. 2009) (stating that district courts are obliged to exclude "junk science").

This gatekeeper role involves three basic inquiries. First, the Court must determine whether a proposed witness is qualified as an expert according to his or her "knowledge, skill, experience, training, or education." In re Scrap Metal Antitrust Litig., 527 F.3d 517, 529 (6th Cir. 2008). Next, the Court must assess whether the expert's testimony is relevant. Daubert, 509 U.S. at 589. Relevancy under the Federal Rules of Evidence is a lenient standard. Id. at 587. Evidence is relevant so long as it relates to a fact at issue and helps the jury determine that fact. Fed. R. Evid. 702; Daubert, 509 U.S. at 590-91.

Finally, the Court must inquire as to whether the expert's testimony is reliable. Id. at 590. The concept of "reliability" implies that an expert's opinion must be based on something "more than subjective belief or unsupported speculation." Id. Rather, a reliable opinion requires that the expert (1) use a reliable methodology, and (2) base his or her opinion on reliable facts or data. In re Scrap Metal Antitrust Litig., 527 F.3d at 529. A. The Court Grants Defendants' Motion To Exclude Sedlik's Fraud Damages Calculations, Legal Opinions, and Model Fee Differential Analysis, But Will Permit Sedlik To Offer His Market Rates and Fee Analysis Models.

Jeffrey Sedlik serves as Navarro's primary expert in support of her claims for actual damages on her copyright claim and fraud damages. Most of Sedlik's opinion is dedicated to calculating Navarro's actual damages on her copyright claim—i.e., the amount P&G would have paid Navarro to purchase a license for each allegedly infringing use of her photographs. Sedlik estimates that Navarro's actual damages range between about $700,000 on the low end to nearly $5 million on the high end. The different estimates are the results of three different methods that Sedlik uses to calculate actual damages: the "Market Rates" model; the "Fee Analysis" model; and the "Model Fee Differential" model. Defendants first argue that Sedlik is not qualified to offer any quantitative analysis at all. Even if Sedlik is qualified, Defendants argue, none of Sedlik's calculations are admissible because each of the three models fails to meet Daubert's reliability threshold. Defendants levy the same arguments against Sedlik's fraud damages calculations, suggesting that they are unreliable, and thus inadmissible, under Daubert.

1. Sedlik Is Qualified To Estimate Reasonable Copyright License Fees.

Rule 702 of the Federal Rules of Evidence permits only those witnesses who are "qualified as an expert" to offer an expert opinion. The qualification requirement is a liberal one, but a court will not accept a witness as an expert simply because the expert identifies as such. Pride v. BIC Corp., 218 F.3d 566, 577 (6th Cir. 2000). Rather, the court must assess the witness's "knowledge, skill, experience, training, or education." Fed. R. Evid. 702. This is a context-specific inquiry. Courts are to ask whether the expert's qualifications "provide a foundation for [the] witness to answer a specific question." Berry v. City of Detroit, 25 F.3d 1342, 1351 (6th Cir. 1994). Accordingly, here, the Court must decide whether Sedlik's training, education, and experience qualify him to opine on fraud damages and actual damages.

Sedlik's proposed testimony consists of his calculations of Navarro's supposed actual damages and fraud damages. Defendants argue that Sedlik is unqualified to act as a damages expert because he is a commercial photographer, who has no professional background in statistics, mathematics, accounting, finance, forensics law or economics. (Sedlik Mot., Doc. 214, #10538). Sedlik concedes that he does not have a degree in any of these quantitative subjects, although he states that he did take classes in economics, statistics, and accounting in college. (Sedlik Dep., Doc. 210, #10179-80). According to Defendants, this admission is a death-knell for Sedlik's expert testimony. With no formal quantitative training, Defendants argue that Sedlik is an "industry expert posing as a damages expert." (Sedlik Mot., Doc. 214, #10539). Therefore, Defendants conclude, Sedlik is unqualified to offer an expert opinion on damages issues.

Defendants are right that Sedlik is an 'industry expert" on the photography industry. On the education front, his photography-centric qualifications include his Bachelor of Fine Arts degree, which required him to complete a curriculum on the business, legal, and technical aspects of photography. (Sedlik Dep., Doc. 210, #10177). Sedlik's formal training, however, accounts for only a small portion of his knowledge. Most of Sedlik's expertise stems from the fact that he has been an advertising photographer for over 30 years. (Sedlik Report ("Sedlik Rep."), Doc. 182-2, #7897).

Even more to the point here, Sedlik's resume demonstrates that his expertise extends beyond how to take pictures, to include expertise about the intricacies of copyright licensing, especially as to commercial photographs. First, Sedlik regularly speaks and teaches about the copyright licensing process. He attends, usually as a featured speaker, the U.S. Patent Office's annual symposium on copyright licensing and events held by the Copyright Society, (Sedlik Dep., Doc. 210, #10177-78), and has testified before Congress about issues surrounding copyright licensing procedures, (id. at #10180). On the academic front, Sedlik teaches some continued learning education courses for attorneys, (id. at #10178), guest taught a course on the practical aspects of photography copyright licensing with Paul Goldstein at Stanford Law School, (id. at #10178-79), and contributed a chapter about licensing photography copyrights for a book by the American Society of Media Photographers. (Id. at #10180). Finally, Sedlik regularly offers his knowledge on copyright licensing to others in his capacity as the president of PLUS Coalition, a non-profit service that helps "define licensing language and provide[s] a foundation for building and managing image rights." (Sedlik Dep., Doc. 210, #10178; Sedlik Rep., Doc. 182-2, #7898).

Defendants argue that this "industry" knowledge does not qualify Sedlik to testify as to damages calculations. But it is not clear why. Sedlik's knowledge and experience seem to include substantial exposure to valuing photography, especially at the licensing and assignment stages. That knowledge is especially pertinent here, where both actual damages under copyright law and fraud damages are directly related to the value of a license, or a full buyout, of Navarro's photographs. Specifically, actual damages likely will depend on the amount to which P&G and Navarro would have agreed as a reasonable licensing fee for the use of Navarro's photographs. Thoroughbred Software, 488 F.3d 358-59. Similarly, fraud damages are arguably equal to the amount that Navarro and P&G would have negotiated for a full buyout of Navarro's copyrights if P&G had been clear that the Supplier Agreement was really an assignment of Navarro's rights.

Sedlik's extensive professional experience likely has given him some idea as to what a reasonable license fee and full buyout price may have looked like for the photographs at issue here. In particular, he has extensive personal experience licensing his own photographs, advising others on how to license their photographs, and engaging in academic and legislative discussions about copyright licensing. These factors, when taken together, qualify Sedlik as a damages expert in suits involving copyrights. Perhaps not surprisingly, a number of courts have reached that same conclusion, and have admitted his testimony on such issues. See, e.g., Karol Western Corp. v. Smith News Co., No. CV12-07695 BRO(VBKx), 2013 WL 12084485, at *3-4 (C.D. Cal. Aug. 23, 2013); Leonard v. Stemtech Health Sci., Inc., No. 08-067-LPS-CJB, 2013 WL 5311295, at *5 (D. Del. Sept. 23, 2013).

2. Sedlik's "Market Rates" Model And "Fee Analysis" Model Are Sufficiently Reliable Under Fed . R. Evid. 702 And Daubert.

Qualifying Sedlik as an expert, however, is just the first hurdle Navarro must clear in order to offer Sedlik's testimony. Navarro must also show that Sedlik's proposed opinions are both sufficiently relevant and reliable under Fed. R. Evid. 702 and Daubert. As evidence regarding licensing fees is clearly relevant, Defendants attack Sedlik's proposed damages testimony on the "reliability" front.

Daubert provides a non-exhaustive list of factors that the Court may consider in assessing the reliability of an expert's opinion. Kumho Tire Co., 526 U.S. at 150-51. Specifically, Daubert advises that a court may consider whether the expert's sources and methods were subject to "testing, peer review, [or] publication" or enjoy "general acceptance in the relevant scientific community." 509 U.S. at 593-94. These factors, though, are merely part of a flexible inquiry "with an overarching goal of assessing the 'scientific validity and thus the evidentiary relevance and reliability' of the principles and methodology underlying the proposed expert testimony." United States v. Langan, 263 F.3d 613, 621 (2001) (quoting Daubert, 509 U.S. at 594-95).

Defendants suggest that all of Sedlik's actual damages calculations and his fraud damages calculations are unreliable under Daubert. As the majority of Sedlik's testimony concerns his actual damages calculations, the Court starts there. Sedlik calculates actual damages using three different models: the "Market Rates" model, the "Fee Analysis" model, and the "Model Fee Differential" model. The first two of these methods, the "Market Rates" model and the "Fee Analysis" model, build upon a "Table of Usages." Sedlik had Navarro produce this "Table of Usages," which outlines Navarro's view of when Defendants exceeded their usage rights. (Ex. to Sedlik Rep. Ex. F, ("Table of Usages"), Doc. 214-3, #10959-96).

To produce the Market Rates model, Sedlik estimates what the "market rate" would be if P&G were to purchase licenses for the usages Navarro outlined in the Table of Usages. He did this by "obtain[ing] reference and fee quotes" from other photographers and then combining that information with his own personal experience to arrive at a final figure. (Sedlik Rep., Doc. 182-2, #7943). The Fee Analysis model similarly estimates a value for each of the "usages" Navarro outlined. But, instead of estimating value through a "market rate," Sedlik asked Navarro to provide a fee schedule detailing what she personally would have charged P&G to purchase licenses for these usages. Sedlik then applied Navarro's fee schedule to the Table of Usages to arrive at an actual damages figure. (Id. at #7934-44).

Defendants claim that the Market Rates and Fee Analysis models are unreliable for two reasons. First, Defendants argue that the Table of Usages, which Navarro manually created in the midst of litigation, is full of erroneous assumptions and biased data. (Sedlik Mot., Doc. 214, #10542). Therefore, Defendants claim, the table is unreliable, and corrupts any models Sedlik predicates on it. (Id.). Second, Defendants argue that both the Market Rates and Fee Analysis models are unreliable because they "are contrary to Navarro's actual licensing practices." (Id. at #10545). Specifically, the models assume that Navarro would negotiate a license for each individual usage when, in reality, she would negotiate bulk licenses that cover multiple usages. (Id.).

Defendants' first argument is essentially a garbage-in-garbage-out theory—i.e., that Navarro's Table of Usages is so flawed that Sedlik's Market Rates model and Fee Analysis model are necessarily flawed as well. Defendants are right that the data set matters. A reliable expert opinion must be "supported by appropriate validation—i.e., 'good grounds,' based on what is known." Daubert, 509 U.S. at 590. The data underlying an expert's opinion does not need to be correct, however, but rather need only provide a "reliable foundation" that is something more than "unsupported speculation." In re Scrap Metal Antitrust Litig., 527 F.3d at 529-30.

Defendants' first argument blurs the distinction between "correctness" and "reliability". If Navarro was inaccurate in her description of P&G's usages then, at least as a general matter, those imperfections go to the weight of Sedlik's calculations, not the reliability (and thus admissibility) of his opinions. As the Sixth Circuit has explained, mistakes in an underlying data set often implicate the credibility and accuracy of the expert's opinion, but that alone does not render it unreliable. Id. at 529. For example, the defendants in In re Scrap Metal Antitrust Litigation moved to exclude a damages expert on the grounds that he used an inaccurate price index and because alterations he made to the price index data set made it even more inaccurate. Id. at 526. Importantly, though, the defendants there did not argue that the "type" of data the expert used was improper for the study, but only that the specific numbers that the expert used were wrong for one reason or another. The Sixth Circuit explained that such objections went to the weight of the evidence, not its admissibility, as least so long as the expert does not pull the underlying data "out of thin air." Id. at 531-32.

Similarly, here, Defendants do not argue that the Table of Usages is an entirely improper, or irrelevant, data set from which to calculate reasonable licensing fees. Rather, they argue only that the specific numbers Navarro provided and Sedlik used were wrong. Thus, the alleged flaws in Sedlik's analysis are of "a character that impugn the accuracy of his results, not the general scientific validity of his methods." Id. at 530 (quoting Quiet Tech. DC-8, Inc. v. Hurel-Dubois UK Ltd., 326 F.3d 1333, 1345 (11th Cir. 2003)). In other words, the basic method—estimating an appropriate licensing fee (more on that below) for the various types of described usages—passes muster. So, the attack on the Table of Usages, at least based on what is currently known, is not grounds to exclude Sedlik's damages calculations. To be sure, if trial testimony shows that the Table of Usages amounts to "pulling data out of thin air," that could change. But based on Navarro's briefing, it appears that she did more than that to assemble the table. A final decision on that front can await further testimony.

That leaves Defendants' argument that the Market Rates model and Fee Analysis model are unreliable because each model assumes that P&G would buy a license for each use instead of bulk licenses for multiple uses. This approach, Defendants claim, is "unmoored from reality," and thus unreliable under Daubert. (Sedlik Mot., Doc. 214, #10546).

This argument, unlike the attack on the Table of Usages, is an attack on Sedlik's methodology, rather than merely the underlying data. In copyright cases, "the touchstone [methodology] for [calculating] hypothetical-license damages is 'the range of the license's reasonable market value.'" Oracle Corp. v. SAP AG, 765 F.3d 1081, 1088 (9th Cir. 2014) (quoting Polar Bear Prods. v. Timex Corp., 384 F.3d 700, 709 (9th Cir. 2004)) (brackets omitted). Experts must tie their estimation of the "reasonable market value" to the economic realities of the relevant market. Concord Boat Corp. v. Brunswick Corp., 207 F.3d 1039, 1057 (11th Cir. 2000). One indicator of the relevant market is the parties' past licensing practices, which may or may not demonstrate a practice of bulk licensing. Bruce v. Weekly World News, Inc., 310 F.3d 25, 29-30 (1st Cir. 2002). But there is "no requirement that actual damages be calculated based on a plaintiff's own history of licensing fees." Leonard v. Stemtech Int'l Inc., 834 F.3d 376, 391 (3d Cir. 2016). In some cases, it is appropriate to calculate a "per-use" licensing fee. In fact, calculating the "fair market value" of a license for each unauthorized use is a "recognized [] approach for calculating damages," id., that is only inappropriate if there is "no realistic prospect whatsoever" that the plaintiff could have negotiated a per-use licensing fee. Bruce, 310 F.3d at 29-30.

Here, although the record demonstrates instances where Navarro negotiated bulk licenses, the Court cannot say, or at least cannot say based on the evidence presented to date, that there is "no realistic prospect whatsoever" that Navarro could (or would) have negotiated certain per-use licenses. Perhaps Navarro would have agreed to different license fees for sales of the packaged product depending on the different retailers who sold the product, or she may have varied fees from one blog post to another based on circulation, timing, and prominence of the image.

In fact, Navarro claims that P&G had previously agreed to exactly these kinds of specific licenses for her photographs. For example, Navarro claims that the "parties entered into three very specific license extensions." (Sedlik Resp., Doc. 223, #12889). In one of these licenses "P&G paid Navarro $5,000 for a license to use her photograph for in-store brochures for one year only and in North America only for ProX." (Id. (citing Pls.' Statement Proposed of Undisputed Facts, ("Pls.' SPUF"), Doc. 204-1, #9426)). In another instance, Navarro claims that "P&G paid [her] $500 for a license to use her photograph in one monthly edition of several Canadian magazines for the facial hair removal Teal Box." (Id. (citing Pls.' SPUF at #9429-30)).

Defendants are free to argue that here, in a hypothetical willing-buyer/willing-seller world, the parties would have negotiated a bulk license, but the Court cannot say that Sedlik's calculations of per-use licenses are completely "unmoored" from the realities of copyright licensing practices. Accordingly, the Court DENIES Defendants' motion to exclude Sedlik's Market Rates and Fee Analysis models.

3. Defendants' Motion To Exclude Sedlik's "Additional Damages" Calculations Is Moot .

After Navarro compiled the "Table of Usages," her trial team scoured the internet looking for additional, as-yet-undiscovered, potential infringements. They then assembled these possible "additional" infringements in a table and provided them to Sedlik. Sedlik reports that he is currently "manually comparing each such usage against the usages included in the Table of Usages, in order to identify potential duplicate usages." (Sedlik Rep., Doc. 182-2, #7953). He does not rely on these "additional" infringements as a separate basis for calculating actual damages, but rather to supplement Navarro's existing Table of Usages, and thus as a supplement to his damages calculations.

In his expert report, Sedlik provides an example illustrating how these additional usages could increase his actual damages estimate. In the hypothetical, Sedlik explains that, if 20% of usages are duplicates or invalid, and the remaining 80% of usages occurred on a website for 5 years, then actual damages for those as-yet-unconsidered usages could be north of $10 million.

Defendants move to exclude any actual damages calculations based on these additional uses. Specifically, Defendants argue that Sedlik's assumptions that only 20% of the additional usages are duplicates for entries on the Table of Usages, or that the additional usages all occurred on a website, find no basis in the record. (Sedlik Mot., Doc. 214, #10547). Navarro contends that Sedlik, however, is not actually offering an actual damages opinion based on these assumptions. Rather, Sedlik was simply offering a "rough estimate that was expressly not intended to be an opinion of actual damages." (Sedlik Resp., Doc. 223, #12905 (quoting Sedlik Rebuttal Rep., Doc. 214-13, #11554)). Thus, Defendants' motion directed to these Additional Unauthorized Uses is moot, Navarro claims, because Sedlik does not intend to offer a damages opinion based on these additional usages.

Based on Navarro's representations about the scope of Sedlik's testimony, the Court agrees the issue is moot. The Court, however, will hold Sedlik to that, and will not allow him to reference such numbers as a basis for buttressing the alleged "reasonableness" of his actual damages calculations (which excludes those additional usages).

4. Sedlik's Model Fee Differential Calculations Are Unreliable.

The final method that Sedlik uses to calculate actual damages is the "Model Fee Differential" analysis. True to its name, this framework estimates Navarro's hypothetical licensing fees by examining the supplemental payments P&G provided to the models whose likeness appears in Navarro's photographs. These models complained to P&G when its use of their likenesses allegedly exceeded the rights Navarro granted to LPK. P&G then paid each model a certain supplemental fee to cover this additional use. Sedlik assumes that the proportionate difference between the original payments to these models, and the supplemental payments they received, tracks "in the same measure" the amount P&G should owe Navarro as "reasonable licensing fee" for the use of her photographs. (Sedlik Rep., Doc. 182-2, #7944). This framework produces two actual damages estimates: one that assumes that Navarro's supplemental fees track the average model fee differential, and one that assumes that Navarro's fees track the highest model fee differential.

Defendants attack this framework on two fronts. First, they claim that the "Model Fee Differential" framework is something that Sedlik created from whole cloth and is not a reliable method of calculating actual damages. (Sedlik Mot., Doc. 214, #10548). Second, Defendants contend that there is no basis for Sedlik to assume that Navarro's fees would track the highest fee differential, or even the average fee differential, across all projects. (Id.).

In response, Navarro fails to provide a single case demonstrating that the Model Fee Differential approach carries any validity with other experts on actual copyright damages. (See generally Sedlik Resp., Doc. 223). This poses a problem for Navarro, who must show that Sedlik's "Model Fee Differential" theory enjoys some level of credibility in the expert community. Navarro could show, for example, that the Model Fee Differential analysis is a "generally accepted" calculation method, or that the theory has been subject to "peer review or testing." Daubert, 509 U.S. at 593-94. But Navarro attempted neither. Sedlik himself also failed to defend the established usage of his Model Fee Differential analysis during his deposition. There, Sedlik explained that in all his academic, practical, and personal experience he has never employed the "Model Fee Differential" theory to price copyright licenses. (Sedlik Dep., Doc. 210, #10248). Likewise, it is not a method that Sedlik advocates in his role at the PLUS Coalition. (Id.). Nor could Sedlik identify any other instances where the "Model Fee Differential" analysis has been accepted as a methodology to determine actual damages. (Id.). In sum, Sedlik says that he "cannot [provide] examples of [this] type of comparison and analysis in any business guide or standard." (Id.).

Sedlik's inability to identify a single instance where he, or any other expert, has used a "Model Fee Differential" approach takes on specific significance given Sedlik's long expert career. Sedlik estimates that he has provided expert opinions by report, testimony, or both, in more than 100 cases in the last 20 years. (Id. at #10192-93). In all that time, Sedlik admits he has never come across the "Model Fee Differential" theory. Nor is this an instance where the absence of support for the "Model Fee Differential" theory in literature or cases simply reflects that this is a newly-emerging calculation method. Sedlik's own experience undercuts this hypothesis. Many of the cases in which he offered testimony were quite recent. In fact, he has testified in 33 cases in the last 4 years, 23 of which were in federal district court. (Sedlik Rep., Doc. 182-2, #7901-05). If the "Model Fee Differential" theory was a recently-emerging method, one would expect to find at least some mention of it during these last four years.

This raises questions as to whether the "Model Fee Differential" method is sufficiently "reliable" under Daubert. Techniques that "attract only minimal support in the scientific community may properly be viewed with skepticism." Daubert, 509 U.S. at 594 (internal quotation omitted). The "Model Fee Differential" method has extremely minimal support—Sedlik is the only expert who uses this method, and he has used it exactly once, in this case.

That said, "neither newness nor lack of absolute certainty in a test suffices to render it inadmissible in court. Every useful new development must have its first day in court." United States v. Bonds, 12 F.3d 540, 561 (6th Cir. 1993) (quoting United States v. Stifel, 433 F.2d 431, 438 (6th Cir. 1970)). And here, Sedlik explains that the theory is "justified in this instance because of the profound similarity that occurred in the unauthorized use[s]." (Sedlik Dep., Doc. 210, #10248).

But even a new theory needs verifiable standards and a reliable foundation in the underlying field. See Bonds, 12 F.3d at 555. Sedlik provides no grounding for his "Model Fee Differential" theory in the methods and procedures of copyright damages calculations or industry licensing standards. Instead, Sedlik simply assumes, with no support (so far as the Court can tell), that the differential between the model's original pay and her supplemental pay is a good predictor for Navarro's reasonable licensing fees. He does not explain why the model's fee differential should track the photographer's fee differential. Rather, he rests on the assumption that the two differentials are "in the same measure." (Sedlik Rep., Doc. 182-2, #7944). Here, Sedlik runs afoul of Daubert. See Nelson v. Tenn. Gas Pipeline Co., 243 F.3d 244, 254 (6th Cir. 2001) (noting that Daubert requires an expert to do more than simply draw mere assumptions from the facts in the record).

At bottom, Sedlik's "Model Fee Differential" theory stands on nothing more than Sedlik's own assumption that P&G exceeded the rights of the models and Navarro in "the same measure." (Sedlik Rep., Doc. 182-2, #7944). Accordingly, Sedlik's Model Fee Differential is "mere speculation or conjecture" that does not pass muster under Daubert. Bonds, 12 F.3d at 562 (quoting United States v. Brown, 557 F.2d 541, 559 (6th Cir. 1977)).

5. Sedlik's Opinion About The Extent Of Navarro's Fraud Damages Is Not Reliable.

Sedlik also offers an opinion as to Navarro's fraud damages, which is his calculation of the amount Navarro and P&G would have negotiated for a full buyout of her rights in her photographs. In formulating his damages calculation, Sedlik first asked Navarro to provide an estimate for what she would charge P&G for a full buyout of one of her photographs. (Sedlik Rep., Doc. 182-2, #7951). Sedlik "reviewed the information provided by Navarro" in light of his "personal knowledge and experience" and found it "to be sound." (Id.). Specifically, Sedlik referenced Navarro's estimates against her 2019 fee schedule. (Sedlik Dep., Doc. 210, #10207). Sedlik concluded that his "careful review" of Navarro's prior pricing schedules coupled with his "knowledge and expertise" indicated that this was a reasonable estimate of Navarro's fraud damages. (Id. at #10206).

In short, Sedlik wants to convey Navarro's own estimates of the value of her photographs to the jury. He will then further assure the jury that Navarro's estimates are a good indicator of the value Navarro's photographs because he checked her estimates using his "personal knowledge and experience" and found "Navarro's methodology in attempting to calculate fraud damages to be sound." (Sedlik Rep., Doc. 182-2, #7951).

Thus understood, Sedlik's opinion falls short of Daubert's requirements. The issue is not that Sedlik used Navarro's own pricing schedules and her advice as a base for his estimate. To be sure, experts can, and often should, rely on record facts to support their opinions. See Jahn v. Equine Servs., PSC, 233 F.3d 382, 390 (6th Cir. 2000) (permitting veterinarian to testify to cause of race-horse's death because his opinion had a factual basis in the record). But Sedlik's opinion fails because it is not at all clear what methodology led him to conclude that her estimates were "sound."

Rather, Sedlik repeatedly notes Navarro's perception of the value of her work, and then concludes that Navarro's photographs indeed could command that price. Sure, Sedlik claims that he checked the amounts she suggested against his "knowledge and expertise," (Sedlik Dep., Doc. 210, #10207), but without a better understanding of the methodology he employed based on that "knowledge and expertise," there is nothing to ensure that Sedlik's opinion is a reliable interpretation of the facts, as opposed to his mere subjective speculation. This is a problem. As the Sixth Circuit has repeatedly instructed, "[r]ed flags that caution against certifying an expert include reliance on anecdotal evidence, improper extrapolation, failure to consider other possible causes, lack of testing, and subjectivity." Newell Rubbermaid, Inc. v. Raymond Corp., 676 F.3d 521, 527 (6th Cir. 2012) (citing Best, 563 F.3d at 177).

The bottom line is that Sedlik's opinion on fraud damages amounts to little more than parroting back Navarro's own estimates, coupled with his stamp of approval, with no clear basis for explaining why that stamp was warranted. That is not an expert opinion. Because this portion of Sedlik's opinion lacks methodology demonstrating an expertised valuation of Navarro's photographs, it is not reliable as that term is used in Daubert.

6. Sedlik's High Damages Estimates Are Not Grounds To Exclude His Testimony.

Defendants ask the Court to exclude all of Sedlik's calculations because his damages estimates are "excessive" to such a degree that they "defy common sense." (Sedlik Mot., Doc. 214, #10551). Sedlik's damages estimates certainly are significant, amounting to several million dollars. (Sedlik Rep., Doc. 182-2, #7943-44). Defendants argue that these estimates are untethered from the facts, as Navarro's photographs are worth, at best, only several thousand dollars. For example, Defendants allege that P&G paid only "$2,000 (for all worldwide web and print of the model on the pack) to replace Navarro's photograph on the Daily Facials packaging." (Sedlik Mot., Doc. 214, #10551). Defendants further allege that Navarro valued her photographs at a similar rate. As evidence, Defendants say that Navarro "offered a retroactive license of $14,000 for the alleged unauthorized usages." (Id.).

When measured against these figures, Defendants argue, Sedlik's damages estimates are "unreasonable." The Court should exclude these opinions, Defendants explain, because "[c]ourts routinely reject copyright holders' claims for actual damages that are excessive on their face." (Id.). In support of that proposition, Defendants cite Baker v. Urban Outfitters, Inc., in which the court "order[ed] the plaintiff to pay $424,185.04 in fees and sanctions because he pursued a meritless claim for $260,000 in actual damages." (Id. (citing 431 F. Supp. 2d 351, 365 (S.D.N.Y. 2006)).

That argument misreads Baker. To start, Baker addressed whether to sanction a party for "vexatious" litigation conduct. Baker, 431 F. Supp. 2d at 362. It did not speak to whether a court should, or could, exclude an expert's testimony because his actual damages estimates allegedly were excessive. And even putting that difference aside, Sedlik's damages calculations are not "excessive" by Baker's standards. The Baker Court sanctioned the plaintiff because he sought $260,000 in actual damages when early discovery revealed that the defendant's profits only amounted to $3,896. Id. at 365. Here, Sedlik's actual damages estimates are a small fraction of Defendants' revenues (and thus, likely fall far short of their profits on these products, too). Perhaps a jury will determine that Sedlik's estimates are off base, but at this stage, they are not so clearly excessive as the plaintiff's damages estimate at issue in Baker.

This argument also faces a larger problem. On a motion in limine, the Court focuses on assessing the reliability of the expert's methods, not scrutinizing the results those methods produce. As the Supreme Court explained in Daubert, "[t]he focus, of course, must be solely on [the expert's] principles and methodology, not on the conclusions that they generate." Daubert, 509 U.S. at 595. Navarro complains that Sedlik's damages estimates are too high, and thus must be unreliable. But this is a challenge to Sedlik's results, not his methods. The Court cannot exclude Sedlik's opinions solely because his damages estimates are too high in Defendants' eyes.

7. The Court Excludes Sedlik's Opinion To The Extent That It Offers Opinions As To Legal Conclusions Or Simply Restates Navarro's Version Of The Facts.

Finally, Defendants object to the balance of Sedlik's report on the grounds that he (1) offers legal conclusions; or (2) simply restates Navarro's version of the facts. (Sedlik Mot., Doc. 214, #10552-54). The Court agrees that Sedlik may not offer opinion testimony that falls into either of these categories.

First, "[t]he principle that an expert may not make legal conclusions is indeed well established." Jones v. Pramstaller, 874 F. Supp. 2d 713, 720 (W.D. Mich. 2012) (citing Berry v. City of Detroit, 25 F.3d 1342, 1353 (6th Cir. 1994)). Defendants are correct that several of Sedlik's "opinions" are actually poorly disguised legal conclusions. For example, Sedlik says that "Navarro's invoices speak for themselves, clearly supporting Navarro's assertion that she licensed limited rights to LPK and that she retained copyrights in her photographs." (Sedlik Rep., Doc. 182-2, #7926). At another point, Sedlik states that "P&G and its customers made significant, unlicensed, and unauthorized use of the Photographs." (Id. at #7934).

Navarro asserts that these statements are not actually legal conclusions. For example, she notes that Defendants take issue with Sedlik's opinion that "the 2007 Supplier Agreement has no bearing on the copyright ownership in any photographs created by Navarro for use by P&G." (Sedlik Rep., Doc. 182-2, #7939). Navarro claims that, when put in context, this statement is not a legal conclusion, but rather an expert opinion. (Sedlik Resp., Doc. 223, #12904). But here is Sedlik's full sentence:

Given the behavior of LPK, P&G, and Navarro in repeatedly offering, accepting, and completing limited usage licenses; and based on my knowledge and experience in the negotiation, application, and use of

service agreements, purchase orders, estimates, and invoices in the photography and advertising industries, the 2007 Supplier Agreement has no bearing on the copyright ownership in any photographs created by Navarro for use by P&G.

(Sedlik Rep., Doc. 182-2, #7939). This context doesn't help Navarro. Sedlik's interpretation of the Supplier Agreement is still a legal conclusion, regardless of whether he comes to that conclusion based on his "knowledge and experience." In short, statements such as those above, and several others like them in Sedlik's report, are not "factual opinions" but rather bald conclusions about the assignment and joint authorship issues in this case. The Court will not allow Sedlik to opine on these ultimate legal issues, nor may he offer any other legal conclusion in this case.

Secondly, Defendants contend that Sedlik parrots Navarro's version of the facts throughout his opinion. As stated above, an expert can, and often should, draw from the facts in the record. See Jahn, 233 F.3d at 390. But the expert must add something in the nature of expertise to those facts. There are several instances in Sedlik's report where he simply repeats Navarro's rendition of the facts, without adding any expert evaluation. For example, Sedlik opines that "P&G provided minimal creative input to Navarro" and that "P&G repeatedly hired Navarro to create photographs for P&G packaging, demonstrating that P&G was pleased with Navarro's creative vision." (Sedlik Rep., Doc. 182-2, #7917).

But Sedlik does not provide a reliable, expert basis to conclude, for example, that "P&G provided minimal creative input to Navarro," or that P&G was "pleased with" her "vision." Rather, Navarro and Defendants should submit their accounts of the photoshoots to the jury, who will then decide whether P&G's creative input was "minimal." In other words, Sedlik makes several statements that infringe on the province of the jury to make factual determinations. This is not to say that Sedlik cannot reference portions of the record to support his opinions, but Sedlik must then offer an expert interpretation of those facts. In sum, as noted in the context of damages above, Sedlik may not relay Navarro's version of the facts to the jury and then give that version the imprimatur of expert approval, unless, of course, he provides some reliable expert basis for his agreement with Navarro's account, which he has failed to do.

In sum, the Court GRANTS IN PART AND DENIES IN PART Defendants' Motion to Exclude Jeffrey Sedlik (Doc. 214). More specifically, the Court strikes Sedlik's: (1) damages calculations that use the "Model Fee Differential" approach; (2) calculations of fraud damages; and (3) any opinions that purport to offer legal conclusions or that merely recite Navarro's version of the facts. Conversely, the Court will allow Sedlik to opine about his actual damages calculations using the "Market Rates" model and the "Fee Analysis" model. B. Robert Zeithammer's Opinion Meets Daubert's Reliability Standards.

Robert Zeithammer, Defendants' expert, offers six online consumer surveys to address Navarro's claim for profit-based damages. (Zeithammer Report ("Zeithammer Rep."), Doc. 199-1, #8912). These surveys fall into one of two categories: "Product Surveys" and "Coupon Surveys." Four of the six surveys are Product Surveys, which address the effect of Navarro's photographs when they appeared on the packaging of in-store Olay products. The Coupon Surveys, as the name suggests, supposedly detect the effect of using coupons that display Navarro's photographs on the sales of the underlying products. Navarro alleges that both types of surveys are unreliable under Daubert.

1. Zeithammer's Product Surveys Are Reliable.

Zeithammer's four Product Surveys include a: (1) Daily Facials no-photo counterfactual; (2) Daily Facials different-photo counterfactual; (3) FHR no-photo counterfactual; and (4) FHR different-photo counterfactual.

Zeithammer employed several different types of questions within these surveys, two of which are at issue here. First, Zeithammer used "conjoint analysis questions." The conjoint analysis proceeded in three steps. First, Zeithammer asked participants to make a purchasing decision from an array of possible products with varying attributes, including the products at issue here. Zeithammer then took that data and estimated the percentage of market share attributable to each product attribute, including Navarro's photographs. At this point, Zeithammer concluded that Navarro's photographs had no statistically significant effect on Olay's sales. But, to be sure, Zeithammer then took the third and final step: he applied an "Aggregate Share Adjustment Factor" which supposedly adjusted his market-share predictions for real-world factors like market structure, product category, and level of competition.

The second form of Zeithammer's questions with which Navarro takes issue is called a "semantic-differential." This is a more straightforward form of analysis than conjoint analysis. In each of the four surveys, Zeithammer presented participants with two photographs: one of the product with Navarro's photograph on the packaging, and one without. He then asked participants to indicate whether they had a preference for either product.

In order for Zeithammer's surveys to qualify under Daubert, his questions and methods must meet Daubert's reliability standards. These standards require a survey expert to: (1) properly define the "universe" of respondents; (2) select a representative sample of that universe; (3) frame the questions in a clear, precise and non-leading manner; (4) use competent interviewers who follow sound interview procedures and have no knowledge of the litigation or the purpose of the survey; (5) accurately report the data; (6) analyze the data in accordance with accepted statistical principles; and (7) assure the objectivity of the process. Safe Auto Ins. Co. v. State Auto. Mut. Ins. Co., No. 2:07-CV-1121, 2009 WL 3150328, at *2 (S.D. Ohio Sept. 30, 2009) (quoting Leelanau Wine Cellars v. Black & Red, Inc., 452 F. Supp. 2d 772, 778 (W.D. Mich. 2006) aff'd, 502 F.3d 504 (6th Cir. 2007)).

According to Navarro, both Zeithammer's conjoint analysis and his semantic-differential questions fail to meet these standards. Navarro levels criticisms against each of the three steps of the conjoint analysis. First, she argues that Zeithammer failed to vary Navarro's photographs as an independent attribute between product options, and thus the survey could not properly gauge the effect of her photographs (prongs 3 and 7). (Zeithammer Mot., Doc. 218, #11858). Next, Navarro argues that Zeithammer improperly calculated the statistical significance of changes in market share (prongs 6 and 7). (Id. at #11875). Finally, Navarro suggests that Zeithammer used the Aggregate Share Adjustment factor as a "fudge factor," thus skewing the results (prongs 6 and 7). (Id. at #11876).

In contrast to Navarro's web of objections to the conjoint analysis, she offers only a single objection to Zeithammer's semantic-differential questions. That is, Navarro argues that the semantic-differential questions were infected with bias because Zeithammer failed to "rotate" the order in which the actual and counterfactual packaging was presented to survey participants (prong 3). (Id. at #11862).

Last, but not least, Navarro generally objects to all of Zeithammer's product surveys on the grounds that Zeithammer sampled an improper universe of respondents (prong 1). The Court addresses, and rejects, each of these arguments in turn.

a. Navarro's Criticism That Zeithammer's Conjoint Analysis Failed To Isolate And Highlight Navarro's Photographs As An Independent Attribute Goes To Weight, Not Admissibility.

Zeithammer's conjoint analysis asked respondents to choose between a variety of facial wipes or female hair removal products, including either the P&G Daily Facials product or the Teal Box/Purple Box. In each iteration of the question, Zeithammer varied several aspects of these products, including package size, package design, brand, and price. (Zeithammer Rep., Doc. 199-1, #8926-27, 8940). By varying these elements, Zeithammer sought to isolate the effect of Navarro's photographs on P&G's sales of Olay products. See Hadley v. Kellogg Sales Co., 324 F. Supp. 3d 1084, 1103 (N.D. Cal. 2018) (describing conjoint analysis).

Navarro argues that Zeithammer could not have properly grasped the effect of her photographs because Zeithammer's survey displayed her photographs only on the P&G packages at issue, and never on the packaging of other brands. (Zeithammer Mot, Doc. 218, #11860). Therefore, Navarro contends, Zeithammer could not properly discern whether the consumer selected P&G's product because of the product's own attributes, or because of Navarro's photograph. (Zeithammer Reply, Doc. 230, #13185).

Navarro's argument is essentially that Zeithammer's report is unreliable because he did not sufficiently isolate her photographs. But "there is an important difference between what is unreliable support and what a trier of fact may conclude is insufficient support for an expert's conclusion." In re Dial Complete Mktg. & Sales Practicing Litig., 320 F.R.D. 326, 332 (D.N.H. 2017) (brackets omitted) (quoting Milward v. Acuity Specialty Prods. Grp., Inc., 639 F.3d 11, 22 (1st Cir. 2011)). Failure to properly isolate or highlight specific attributes in conjoint analysis might provide a promising line of attack on cross-examination, but those critiques go to the weight, not the admissibility, of a conjoint analysis. Hadley, 324 F. Supp. 3d at 1108-09. Thus, even if Navarro is right that Zeithammer failed to sufficiently isolate the effect of her photographs, it is not grounds to exclude the survey under Daubert. And, on a related note, that is a big "if." As P&G notes, the point of the consumer surveys was not to isolate the impact of Navarro's photographs on the sale of these types of products generally, but rather to isolate, to the extent possible, the impact that Navarro's photographs had on the sale of Olay products. That is, the counterfactual that Zeithammer sought to determine was how Olay's sales would have fared against competitors without photographs on the Olay boxes, as opposed to how much P&G had in sales of Olay with the photos. In both of those worlds, the competitors' offerings did not (or would not) have displayed Navarro's photographs, so it is not entirely clear to the Court why Zeithammer should have included those options in his survey.

Next, Navarro says that Zeithammer could not have properly identified the effects of Navarro's photographs because he did not indicate to respondents that they should pay attention to Navarro's photographs when making a purchasing decision. That is an odd objection—because, if Zeithammer had indicated to respondents that they should pay attention to Navarro's photographs, that could create a "focalism bias," which presents a recognized Daubert problem. In other words, if Zeithammer alerted respondents to the importance of Navarro's photographs, that might have led them to think that picking the box with (or without) the photograph was the "right answer." Such a survey technique runs afoul of Daubert because it tends to create the participants' opinions, rather than elicit them. See Wells Fargo & Co. v. WhenU.com, Inc., 293 F. Supp. 2d 734, 767-68 (E.D. Mich. 2003) ("A survey is not reliable if it suggests to the respondents an answer that would not otherwise have occurred to them.").

Zeithammer seems to be aware of this "focalism" pitfall and it appears he purposefully sought avoid it. In fact, Zeithammer states in his report that he "did not want to risk revealing the purpose of my survey." (Zeithammer Rep., Doc. 199-1, #8938). Therefore, Zeithammer purposefully tried not to overwhelm respondents with options featuring Navarro's photographs. (Id.). As an extra precaution, Zeithammer "included an open-ended question at the end of each survey, asking respondents what they thought the survey's intent was." (Id.). And from the responses to that question, Zeithammer concluded that no respondents inferred the purpose of the surveys. (Id.). Navarro criticizes Zeithammer's diligence in this regard, but Navarro's logic is backwards. The Court will not exclude Zeithammer's report because of what appear to be appropriate steps to avoid a focalism issue.

b. Zeithammer's "Market Analysis" Is Relevant And Reliable.

Navarro next contends that Zeithammer's survey is unhelpful because "his analysis cannot detect any downward change in Olay market share." (Zeithammer Mot., Doc. 218, #11875). This is so, Navarro contends, because of two flaws in Zeithammer's analysis. First, Navarro says that Zeithammer's analysis "does not have the power to detect changes in market share smaller than 5% points." (Id.). And, as P&G has only cornered 2-6 percent of the market share (depending on the product), Zeithammer's analysis cannot discern practically any effect of Navarro's photographs even if P&G's market share dropped to zero. (Id.). Next, Navarro takes issue with the fact that "in order for [Zeithammer's] survey to be able to register a statistically significant effect from the use of Navarro's images sales would need to decline by approximately 50 percent." (Id. at #11844) (emphasis omitted). These two "flaws," Navarro reasons, strip Zeithammer's survey of any probative value.

It is not clear whether Navarro is suggesting that Zeithammer's survey is irrelevant, unreliable, or both. The Court understands, however, that Navarro's objection sounds both in reliability, as she alleges Zeithammer's analysis is biased, and in relevance, as she claims Zeithammer's analysis is so unreasonable that it is unhelpful to a jury. Either way, Navarro's arguments are not persuasive.

As a general matter, Navarro's objections primarily concern Zeithammer's results, not his methodology. In short, Navarro claims that Zeithammer's report is "imprecisely calibrated" such that he should have found statistical significance when he did not. This is not an objection to the model itself, but rather to Zeithammer's interpretations of his data, namely his findings of statistical significance. As such, Navarro's objections, just like other "objection[s] as to [an expert's] interpretation of the data," "go[] to its weight, not its admissibility." In re E.I. du Pont de Nemours & Co. C-8 Pers. Injury Litig., 337 F. Supp. 3d 728, 746 (S.D. Ohio July 20, 2015) (citing In re Scrap Metal, 527 F.3d at 531-32).

Even so, it is not entirely clear that Navarro has actually identified flaws in Zeithammer's analysis. Start with Navarro's argument that Zeithammer "set parameters" so that it would take a significant drop in market share—i.e., about 50%—to indicate statistical significance. (Zeithammer Dep., Doc. 199, #8866). But, as Zeithammer explained in his deposition, the allegedly high threshold for demonstrating statistical significance is not a part of his survey design. Rather it is simply a function of the data from the surveys. (Id.).³ In other words, once Zeithammer selected the significance level (here, a level of two standard errors, which is a recognized standard for significance), the actual percentage decrease needed to demonstrate that level of significance turned on the nature of the data in the sample.

3 In other words, Zeithammer's data has a large standard deviation and so it is not unusual to see fluctuations in market share. Accordingly, it would require a decently large drop or increase in market share, i.e., about 50%, to indicate that Navarro's photographs had some effect on the market.

Navarro attempts to impugn Zeithammer's survey design in another way, as well, pointing to Zeithammer's statement that his survey is "designed to detect proportion differences as small as 5 percent in more than 4 out of 5 replications." (Zeithammer Mot., Doc. 218, #11875). Navarro appears to be under the impression that Zeithammer's methodology cannot detect the effect of Navarro's photographs unless P&G acquires or loses 5% market share. But Zeithammer explains that the "5 percent" is not referring to market share. (Zeithammer Dep., Doc. 199, #8825). Rather, it is a measure of the "statistical power" of the survey. Statistical power is a term of art that refers to the probability that a test will correctly reject a false null hypothesis at a given significance level. According to Zeithammer, then, 80% of the time, Zeithammer's model should correctly reject a false null hypothesis at 5% significance.⁴ That is, if the proportion of customers who select the Olay packaging is at least 5% different when the packaging has a Navarro image on it versus not having a Navarro image on it, Zeithammer's survey should detect that at least 80% of the time. Thus, Navarro's argument, which is based on a misunderstanding of Zeithammer's claim, fails.

4 In other words, Zeithammer claims that, 80% of the time, his test will pick up on the effect of Navarro's photographs, so long as the effect causes at least a 5% change in the proportion of customers who pick that product. See Kuhn v. Wyeth, Inc., 686 F.3d 618, 622 n.5 (8th Cir. 2012).

c. Zeithammer's Aggregate Share Adjustment Calculations Are Not Unreliable "Fudge Factors."

Daubert's reliability threshold requires an expert to analyze survey data in accordance with accepted statistical principles. Safe Auto Ins, 2009 WL 3150328, at *2 (quoting Leelanau Wine Cellars, 452 F. Supp. 2d at 778). The dispute here is whether the third step of Zeithammer's conjoint analysis, namely his application of the "Aggregate Share Adjustment" method, contravenes accepted statistical principles.

As a result of his conjoint analysis, Zeithammer arrived at a "Predicted Share of Preference," which is the "average [] probability that a respondent in the conjoint study purchases the product at issue in the simulated market." (Zeithammer Rep., Doc. 199-1, #8918). But this "Predicted Share of Preference" exists only in theory and rests on a series of assumptions, including the assumptions that all products are distributed evenly and that customers have equal awareness about each product. (Zeithammer Resp., Doc. 220 Ex. 2, Bryan Orme & Rich Johnson, External Effects Adjustments in Conjoint Analysis, Sawtooth Software (2006), ("Orme & Johnson")). Zeithammer says that he thus applied the "Aggregate Share Adjustment" method to tune his model to more realistic market conditions. (Zeithammer Rep., Doc. 199-1, #8919).

Navarro argues that this "Aggregate Share Adjustment" is nothing more than a "fudge factor" Zeithammer employed to arrive at his preferred result. (Zeithammer Mot., Doc. 218, #11877). Fudge factors fail under Daubert for the same reason that ipse dixit expert opinions fail, namely because there is no methodology to connect the expert's conclusion to the facts. See United States v. Rothberg, No. 00 CR 85, 2002 WL 171963, at *3 (N.D. Ill. Feb. 4, 2002).

Dow Corning Corp. v. Jie Xiao, 11-10008-BC, 2013 WL 992773 (E.D. Mich. Mar. 13, 2013), provides a good illustration of the kind of "fudge factor" that Daubert prohibits. That case concerned the alleged appropriation of trade secrets about the size and characteristics of Dow's first-generation fluid bed reactors, a type of reactor used to manufacture polysilicon products for the semiconductor industry. Id. at *7. It was undisputed that the actual dimensions of the defendant's fluid bed reactors were almost identical to those employed by Dow Corning. But the question was whether that was due to the defendant copying Dow Corning's dimensions (one of the key people at the defendant's company had previously worked at Dow Corning with fluid bed reactors for many years), or instead whether the necessary dimensions were independently derived from publicly available information.

In opining that it was the latter, the defendant's expert explained that if one took the estimated diameter for the reactor based on publicly available information, then added 10%, and then rounded to the nearest foot, the resulting answer matched the actual dimensions of the defendant's reactor. Id. at *8. The plaintiffs then asked the expert if he did "any calculations to determine whether 10 percent was an appropriate estimate?" Id. The expert replied: "No. And that's not typically done. You know it's not zero. And if you don't put something in you're going to be wrong for sure. So 10 percent is [] reasonable." Id. The plaintiffs' counsel then asked the expert if he had "any literature" to support his decision to "round to the nearest foot." Id. The expert again gave a non-answer, saying only that "I don't know—I certainly can't point to it sitting here." Id. In short, the expert jumped from one figure to another, without any explanation or support for that calculation. The Court ultimately concluded that the expert's method of calculating the dimensions failed to provide a reliable basis for asserting that the actual dimensions used were the result of calculation rather than copying, and that the expert's opinions instead consisted of inadmissible "fudge factors." Id. at *15-16.

Zeithammer's "Aggregate Share Adjustment," in contrast, has a basis in accepted statistical practice, and cannot be dismissed as a mere fudge factor. Far from lacking in academic support, Zeithammer indicates that he draws "Aggregate Share Adjustment" methodology directly from Bryan Orme and Rich Johnson's research paper on adjusting for external effects in conjoint analysis. (Zeithammer Rep., Doc. 199-1, #8919). Navarro, on the other hand, claims that Orme and Johnson oppose, or least do not encourage, such adjustments. (Zeithammer Mot., Doc. 218, #11877). But that is an odd claim, considering that Orme and Johnson's work specifically instructs researchers on how to perform such adjustments. Moreover, Orme and Johnson acknowledge that "there are cases where researchers must make adjustments." (Orme & Johnson at #12208). When adjustments are necessary, Orme and Johnson outline the procedures to do so through the Aggregate Shares Adjustment approach used here. (Id. at #12214).

For his part, Zeithammer is transparent about exactly how, and why, he adjusted the "predicted share" values. Put briefly, Zeithammer explains that the adjustment simply fit his model data into actual reported market data. So, for example, in one of Zeithammer's models, P&G's market share with Navarro's photograph was 11.6%, while it was 10% without Navarro's photograph—about a 14% drop. (Zeithammer Rep., Doc. 199-1, #8931). In reality, though, P&G did not control 11.6% of the market, but rather controlled about 2.9%. Accordingly, Zeithammer calculated what a 14% drop would look like in the actual market, which equaled about 0.4%. (Id.).

In short, not only is Zeithammer's Aggregate Share Adjustment a recognized methodology, but Zeithammer also explains its application in this case. This is a methodology, not a fudge factor. Navarro might disagree with the results of Zeithammer's adjustment, or the merits of such adjustment, and those criticisms may even have some merit, but they go to the weight, not the admissibility, of the evidence.

d. Any "Order Bias" In Zeithammer's Study Goes To Weight Not Admissibility.

In order for a survey to meet Daubert's reliability requirement, an expert must frame each question in "a clear, precise and non-leading manner," so as not to improperly influence respondents. Safe Auto Ins., 2009 WL 3150328, at *2 (quoting Leelanau Wine Cellars, 452 F. Supp. 2d at 778). Navarro claims that Zeithammer's semantic-differential questions fall short of this neutrality requirement. (Zeithammer Mot., Doc. 218, #11863). The semantic-differential questions at issue showed respondents two side-by-side photos of P&G products: one with the Navarro image, and one without (either with no photo or with a different photo). Zeithammer presented these two formats, the no-photo differential and different-photo differential, for each of the three products at issue (FHR-Teal Box, FHR- Purple Box, and the Daily Facials product).⁵ Each respondent encountered the question in the exact same format—that is, for a given question, the packaging featuring Navarro's photograph was always on the same side (left or right, depending on the question) for all respondents.

5 Zeithammer Mot. Ex. 4, Doc. 218-1, #11916 (Daily Facials "no-photo" survey); Zeithammer Mot. Ex. 5, Doc. 218-2, #11954 (Daily Facials "different-photo" survey); Zeithammer Mot. Ex. 6, Doc. 218-3, #11991 (FHR "no-photo" survey—Purple Box); id. at #11993 (FHR "no-photo" survey—Teal Box); Zeithammer Mot. Ex. 7, Doc. 218-4, #12031 (FHR "different-photo" survey—Purple Box); id. at #12033 (FHR "different-photo" survey—Teal Box).

Navarro says that Zeithammer's failure to rotate the order of Navarro's photographs from one respondent's survey to the next infected his survey with "order bias." (Zeithammer Mot. at #11863). In other words, Navarro complains that everyone who took the Daily Facial no-photo differential, for instance, saw Navarro's photograph on the left, and the packaging without her photo on the right. Some of these respondents, Navarro claims, should instead have seen the Navarro photograph on the right, and the other package on the left. Navarro particularly objects to four out of the six questions where Navarro's photographs appear on the left side of the page. This presentation infected the survey with order bias, Navarro reasons, because most people are right-handed and, as such, "tend to judge objects on the left as negative." (Zeithammer Mot., Doc 218, #11863 (citing id. Ex. K, Doc. 218-11)).

First, it is not entirely clear that this right-side bias exists. In fact, Zeithammer suggests that the opposite is true—that people prefer objects on the left because most people read left to right. (Zeithammer Dep., Doc. 199, #8846). Further, Zeithammer's results show that, sometimes, respondents preferred the image on the left. Navarro's images appeared on the left in four out of six questions, and in one instance, respondents preferred the Navarro image on the left to the non-Navarro image on the right. (Zeithammer Resp., Doc. 220, #12192 (citing Zeithammer Rep., Doc. 199-1, #8930, 8932, 8943-44.)). The fact that Navarro has failed to demonstrate that Zeithammer's semantic differential questions resulted in order bias is enough in and of itself to deny her motion in limine on this issue. Frosty Treats, Inc. v. Sony Comput. Ent. Am., Inc., No. 03-0378-CV-W-SOW, 2004 WL 5500075, at *4 (Mar. 3, 2004) (denying a motion in limine because "Defendant has failed to demonstrate that the questioning utilized by [Plaintiff's expert] resulted in order bias").

But even if "order bias" does exist, Navarro has not demonstrated that the bias corrupts Zeithammer's survey to a such an extent that the survey fails to satisfy Daubert. The source Navarro cites to support her claim of right-handed bias found that it happens only 3% of the time. (Zeithammer Mot. Ex. K, Doc. 218-11, #12144-45). Moreover, other courts, including the ones Navarro cites, have held that perceived order bias, standing alone, is not a sufficient basis to exclude a survey. See, e.g., Alzheimer's Disease & Related Disorders Ass'n, Inc. v. Alzheimer's Found. of Am., Inc., 307 F. Supp. 3d 260, 296 (S.D.N.Y. 2018) (explaining that order bias alone does not make a survey inadmissible, but does so only when considered with other well-founded criticisms).

For example, Navarro points to Procter & Gamble Pharm., Inc. v. Hoffmann-La Roche Inc., No. 06 Civ. 0034, 2006 WL 2588002, at *23 (S.D.N.Y. Sept. 6, 2006). There, the Court excluded a survey that may have involved order bias, noting that "the questions ought to be rotated." Id. But that observation was hardly the driving force behind the court's decision. Rather the court decided to exclude the survey at issue because it suffered from numerous problems including leading and suggestive closed-ended questions, an inappropriate control question, and serious data collection flaws. Id. at *23-24. The order-bias issue, to the extent it existed at all, comprised a mere three sentences of the court's multi-page reliability analysis. In fact, the court concluded that the survey was unreliable only because of its "combined" structural and design inadequacies. Id. at *24.

Order bias, like most other flaws in survey methodology, does not automatically disqualify an expert opinion unless the flaw is so substantial that it undermines the very foundation for the expert's conclusions. See In re Whirlpool Corp. Front-Loading Washer Prods. Liab. Litig., 45 F. Supp. 3d 724, 753-54 (N.D. Ohio 2014). Thus, here, as in other instances, the Court will allow a jury to weigh Zeithammer's survey and discount its value for order bias as the jury sees fit. See, e.g., Estes Park Taffy Co., LLC v. Original Taffy Shop, Inc., No. 15-cv-01697-CBS, 2017 WL 2472149, at *5-6 (D. Colo. June 8, 2017).

e. Zeithammer's Universe Of Respondents Is Not So Overbroad As To Be Unreliable.

The selection of a proper universe is one of the most important factors in ensuring the reliability of a survey. Leelanau Wine Cellars, 452 F. Supp. 2d at 781 (citing Amstar Corp. v. Domino's Pizza, Inc., 615 F.2d 252, 264 (5th Cir. 1980)). Selection of a proper universe is critical because "even if the proper questions are asked in a proper manner, if the wrong persons are asked, the results are likely to be irrelevant." Wells Fargo & Co. v. WhenU.com, Inc., 293 F. Supp. 2d 734, 767 (E.D. Mich. 2003) (quoting 5 McCarthy, § 32:159 at 32-250.3). A proper universe of respondents includes the people whose opinions are relevant to the question at hand. Leelanau Wine Cellars, 452 F. Supp. 2d at 781-82. The goal is to survey a group of people that is representative of this population. See Rocky Brands, Inc. v. Red Wing Shoe Co., No. 2:06-cv-00275, 2009 WL 5125475, at *4-5 (S.D. Ohio Dec. 28, 2009) (asking whether surveyed patrons adequately represented likely purchasers of the product at issue).

The parties in this case dispute the scope of the relevant universe of respondents. Navarro insists that Zeithammer's survey universe is overinclusive for three reasons: (1) it included men; (2) it included "irrational" respondents; (3) it included smartphone users. The inclusion of these groups does not corrupt Zeithammer's survey to such an extent as to render it unreliable.

i. Zeithammer's FHR Survey Was Only Slightly Overbroad.

Navarro argues that Zeithammer's FHR survey is unreliable because the data collection company made a mistake and included men in the sample population. (Zeithammer Resp., Doc. 220, #12193). First, by Navarro's own expert's estimate, men accidently included in the facial hair removal survey accounted for only 7.7% of respondents. (Harvey Report, Doc. 180-1, #7543). This relatively slight over-inclusiveness does not render the survey unreliable. An overbroad universe renders a survey inadmissible only when it is so "significantly" overinclusive that it no longer provides any insight into the relevant population. In other words, overbroad surveys are unreliable when the effect of the additional respondents obscures the relevant population.

A good example of a "significantly overbroad" survey appears in Leelanau Wine Cellars, which was a trademark action between two wineries on Michigan's Leelanau Peninsula. 452 F. Supp. 2d at 782. There, an expert surveyed Michigan customers over twenty-one years of age who had either purchased, or intended to purchase, a bottle of wine in the $5 to $14 price range within a three-month period of the interview. Id. This universe was "significantly overbroad" because "a survey participant who purchases wine only at grocery or discount retail store such as a Meijer or Sam's Club and who does not intend to visit and/or is unaware of Defendants' tasting rooms, or even of wineries in the Leelanau Peninsula, would not be a potential purchaser of Defendants' wine." Id. Yet, the court held that, while it was "a close call[,]" this survey was admissible because the survey still provided some, if slight, insight into the relevant population. Id. at 785-86.

The Leelanau court did not discuss the extent to which the survey was overbroad. In other words, the court did not (and, potentially, could not) evaluate how many of the respondents were irrelevant big-box-store-shopping-non-customers. But the court admitted the survey anyway, even considering the risk that the survey was "significantly overbroad." Id. at 782. Here, by contrast, the overbreadth of Zeithammer's survey pool is both known and relatively small. Put another way, Navarro has confidence that at least 92.3% of the people surveyed in Zeithammer's FHR surveys were women, who she does not dispute are part of the relevant universe. Navarro has not demonstrated that the minimal level of overinclusion at issue here somehow obscured, or obstructed, Zeithammer's ability to glean insight into the preferences of 90-plus percent of his survey respondents. To the contrary, Zeithammer says that he re-ran his analysis without men and found that their inclusion did not bias the survey results. (Zeithammer Dep., Doc. 199, #8854). In sum, Zeithammer's FHR survey is, at most, slightly overbroad. Accordingly, Navarro's objection to that effect "merely goes to the weight given the survey results, not to the very admissibility of the survey." Whirlpool Prop., 2006 WL 62846, at *4 (citing Piper Aircraft Corp. v. Wag-Aero, Inc., 741 F.2d 925, 930-31 (7th Cir. 1984)).

ii. A Genuine Dispute Over The Relevant Universe Of Respondents Goes To Weight, Not Admissibility.

The parties also dispute the relevant universe for several of Zeithammer's other surveys. First, Navarro complains that Zeithammer intentionally included men in his Daily Facial surveys. Specifically, Navarro argues that men are not part of the relevant universe of Daily Facials consumers. (Zeithammer Mot., Doc. 218, #11868). Zeithammer disagrees. He claims that Daily Facials may be a unisex product, and as such, it makes sense to include men in the universe of survey respondents. (Zeithammer Rep., Doc. 199-1, #8921).

A similar dispute exists about Zeithammer's inclusion of certain "irrational" respondents in all of his conjoint analysis surveys. In addition to his probative conjoint analysis questions, Zeithammer also presented respondents with two "fixed task" questions designed to filter out irrational respondents. (Id. at #8927). These questions provided respondents three options, two of which were the exact the same product, but one had a higher price than the other. (Id.). The idea is that no rational respondent would choose to pay more for the same product. If a respondent preferred the duplicate product with the higher price, Zeithammer excluded that respondent as "irrational," and therefore not probative of the relevant universe of actual consumers.

Zeithammer discovered, however, that respondents found one of the two fixed-task questions much harder than other. When Zeithammer positioned the two identical products right next to one another, respondents were often able to discern, and avoid, the irrational option. (Zeithammer Dep., Doc. 199, #8830). In the other question, however, the two identical products were not adjacent to one another. More respondents selected the "irrational" option in that circumstance. (Id.). Zeithammer concluded that the respondents who answered this latter question incorrectly were not "irrational," but rather just confused. (Id.). Accordingly, Zeithammer included their responses in his survey data. Navarro believes that this was a mistake.

Last, but not least, the parties dispute the propriety of including smart phone users in the universe of Zeithammer's survey respondents for the FHR different-photo alternative. Originally, Zeithammer sought to exclude all smart phone users from his pool of respondents because it is "difficult to represent multiple complex products with complex descriptions on a small phone screen." (Id. at #8822-23). And Zeithammer succeeded in excluding smart phone users from all of his surveys except one—his FHR different-photo survey included smart phone respondents due to the survey platform company's error. (Id. at #8822). Zeithammer attempted to correct this problem by removing all respondents who used an iPhone to take the survey. Navarro points out that this is not enough, as a significant portion of Zeithammer's remaining respondents, namely about 48 percent (or 169 out of 349), used another type of smart phone to take the survey. (Zeithammer Mot., Doc. 218, #11870). Zeithammer's failure to exclude these respondents, Navarro argues, renders his survey overbroad, and thus inadmissible under Daubert.

Zeithammer disagrees. Although he originally sought to exclude respondents who used a smart phone to take the survey, Zeithammer later clarified that it was only a precaution. (Zeithammer Dep., Doc. 199, #8823). The presence of smart phone users could actually lead to a sample that is more representative of the relevant universe, Zeithammer explains, because "more and more consumers actually use phones in their shopping." (Id.).

At bottom, these three objections (inclusion of men, "irrational consumers," and smart phone users) boil down to disagreements about the scope of the relevant universe. The Court cannot say as a matter of law which party has the better of this argument. Neither party presented any evidence to demonstrate that, for example, men are, or are not, consumers of Daily Facials. Nor have they submitted evidence as to whether smart-phone users are inappropriate respondents. Navarro, for her part, claims that Zeithammer had no grounds beyond his ipse dixit to admit certain "irrational" consumers. But conversely, Navarro fails to provide evidence or citations to cases that establish why Zeithammer was wrong to do so.

At the end of the day, the Court falls back on the proposition that a court should exclude survey evidence only when the universe of respondents is so over (or under) inclusive as to compromise the entire study's reliability. Whirlpool Prop., 2006 WL 62846, at *4 (citing Piper Aircraft, 741 F.2d at 930-31). When two parties dispute the scope of the relevant universe, but neither has proven its case, the appropriate path forward is to permit a jury to assess the credibility of the survey in light of any objections. See, e.g., Calista Enters. Ltd. v. Tenza Trading Ltd., 43 F. Supp. 3d 1099 (D. Or. 2014). As Navarro has not proven that the universe is overbroad, let alone that it is so overinclusive that it renders the survey wholly irrelevant, her objections go to weight, rather than admissibility. See Whirlpool Prop., 2006 WL 62846, at *4.

iii. The Possibility Of "Irrational" Responses To Zeithammer's Random Choice Questions Does Not Render His Survey Unreliable.

Most of Zeithammer's conjoint analysis consisted of random-choice questions, where respondents were free to choose between a series of potential products. When the software shuffled the product options, however, it sometimes included two identical products at different price points. Navarro says that Zeithammer's survey is unreliable because he failed to exclude the "irrational" respondents who picked the higher-priced option. (Zeithammer Mot., Doc. 218, #11873). As support, Navarro cites to Oracle America, Inc. v. Google Inc., No. C 10-03561, 2012 WL 850705, at *11 (N.D. Cal. Mar. 13, 2012). In that case, the Court held that a conjoint analysis was unreliable because the expert failed to exclude the 25% of respondents who preferred or were ambivalent between a smartphone costing $200 to a theoretically identical smartphone costing $100. Id.

As Oracle demonstrates, a survey might be unreliable if it is so tainted with irrational respondents that it no longer reflects the relevant universe. But such an error undermines the reliability of a survey only in the rare circumstances where an expert has failed to exclude a significant number of irrational respondents. For example, in MacDougall v. American Honda Motor Co., the court found that a survey was unreliable because the expert failed to exclude "a remarkable 55.4% of individual responses [that] reflected economically irrational choices as to either price levels." No. SACV 17-1079, 2020 WL 5583534, at *8 (C.D. Cal. Sept. 11, 2020).

Zeithammer's survey does not suffer from a flaw of comparable magnitude. To start, Navarro presents only one instance, where one random-choice question accidently included the two identical products. (Zeithammer Mot., Doc. 218, #11873). She does not allege that a significant percentage of Zeithammer's survey is tainted with these "irrational respondents." Further, Zeithammer's other controls may have already accounted for any of the respondents who answered this question in an "irrational" way. For example, he excluded both the fastest 10% of respondents and those respondents who answered the fixed-choice tasks incorrectly. (Zeithammer Rep., Doc. 199-1, #8927, 8938).

Moreover, even if there is the occasional "irritational" respondent still lurking in Zeithammer's survey pool, Zeithammer insists that the conjoint analysis model is designed to control for such "irrational" respondents. (Zeithammer Dep., Doc. 199, #8835). Orme and Johnston, who provide the scholarly foundation for Zeithammer's study, say conjoint analysis permits irrational responses to a random-choice question, as those answer still provide some insight and are helpful to calculate an error rate. (Zeithammer Resp., Doc. 220, #12196).

In sum, Zeithammer's survey minimized the effect of irrational respondents. To the extent irrational respondents remain in the surveyed population, they do not overshadow the responses of the relevant universe. Accordingly, Navarro's objection to the inclusion of these respondents is not grounds to exclude the survey under Daubert.

2. Zeithammer's Coupon Surveys Are Reliable.

Separately, Navarro takes issue with Zeithammer's coupon surveys for two reasons. First, she alleges that Zeithammer surveyed an overbroad universe because he included men in his pool of respondents. (Zeithammer Mot., Doc. 218, #11868). This argument is unpersuasive for the same reasons as Navarro's objections to the inclusion of men in Zeithammer's conjoint analysis. Second, Navarro argues that Zeithammer's coupon surveys failed to test for the effect of Navarro images. (Id. at #11865). The coupon surveys essentially consisted of two questions. First, Zeithammer presented respondents with a coupon featuring Navarro's photograph, and then showed them the actual product, and asked them how likely they would be to purchase it. (Zeithammer Rep., Doc. 199-1, #8952). Zeithammer then repeated the process using a coupon without Navarro's photograph. (Id.). Navarro claims that this survey could not possibly detect the effect of the coupons featuring Navarro's photographs because, each time, the respondents saw the actual product with Navarro's photograph. This, Navarro contends, "removed the focus from the coupons and placed it on the actual packaging." (Zeithammer Mot., Doc. 218, #11865).

Contrary to Navarro's argument, Zeithammer's decision to hold all variables constant except for Navarro's photographs on the coupons is a sound attempt to isolate the effect of the coupons. In other words, Navarro is separately seeking damages for use of the photographs on the product packaging itself. The question Zeithammer is addressing here is whether the use of Navarro's photographs on the coupons had an additional marginal impact beyond that caused by the product packaging. Thus, the Court can understand why Zeithammer chose to test this by varying only the coupons and not the product packaging. Indeed, in many ways, this survey methodology may better reflect real market conditions, as when consumers took their coupons to the store, they likely would have seen the Olay products with Navarro's photographs. Accordingly, neither of Navarro's challenges to Zeithammer's surveys warrant exclusion.

Therefore, the Court DENIES Defendants' Motion to Exclude Zeithammer. C. Doug Bania's Remaining Expert Opinions Are Admissible, But Much Of Hazel's Sales Trend Opinions, Along With His Opinions On Zeithammer's Surveys, Are Unreliable.

Navarro moves to fully exclude the expert opinion of Steven Hazel, Defendants' omnibus damages expert, and to partially exclude the expert opinion of Doug Bania, Defendants' actual damages rebuttal expert. Navarro argues that both witnesses are unqualified to offer an expert opinion, while Hazel's opinion, Navarro argues, also fails to meet Daubert's reliability threshold.

1. Hazel Is Generally Qualified To Testify As A Damages Expert , But May Not Parrot Zeithammer's Conclusions.

In order to testify as an expert witness, Hazel must be "qualified as an expert by knowledge, skill, experience, training, or education." Fed R. Evid. 702. Navarro argues that Hazel is unqualified to offer four of his opinions, two of which are related to actual damages, and two which are related to profit-based damages. The actual damages opinions include Hazel's: (1) actual damages estimates; and (2) his rebuttal of Navarro's actual damages expert, Jeff Sedlik. On the profit-based damages front, Hazel offers (1) a sales trend analysis purporting to estimate the profits attributable to Navarro's photographs; and (2) a rebuttal of Larry Chiagouris, Navarro's expert on profit-based damages. (Hazel & Bania Mot., Doc. 217, #11633). The Court addresses each of Navarro's challenges in turn.

a. Hazel Is Qualified To Testify As To Actual Damages And To Rebut Sedlik.

Hazel offers two opinions that concern actual damages. First, Hazel provides estimates of Navarro's damages by crunching various numbers relating to (1) the fees that Navarro actually charged P&G under various invoices for the photographs at issue in this case, (2) the revenues that P&G generated on the products associated with those invoices, and (3) information about market rates for photographs that Hazel collected from the Masterfile website. (Hazel Report ("Hazel Rep."), Doc 145-2, #4535-39). Second, Hazel rebuts the opinions of Jeffrey Sedlik, Navarro's actual damages expert. (Id. at #4545). Navarro claims that Hazel is unqualified to opine as to either topic because both concern actual damages, which require Hazel to analyze hypothetical licensing negotiations. (Hazel & Bania Mot., Doc. 217, #11634). Hazel is unqualified to testify as to these licensing terms for photographs, Navarro reasons, because he "is not a photographer," "has never provided an expert opinion on the licensing of photographs and does not consider himself to be a licensing expert." (Id.).

What Navarro's argument overlooks is that Hazel does not have to be an expert in "photography licensing" in order to calculate actual damages in the manner that he did here. Rather, Hazel must show only that some combination of his education, experience, and knowledge equip him to properly estimate a damages figure based on the methodology that he employed. There are a variety of skills that could enable an expert to valuate a hypothetical license. For example, some experts are qualified to estimate the value of a hypothetical license because of their extensive experience in actually licensing photography. Navarro's actual damages expert, Jeffrey Sedlik, is qualified for this reason. But an expert like Hazel checks this box a different way, namely through his quantitative ability. His methodology does not rely on his own assessment of a "proper" licensing rate (a topic on which he admittedly has no expertise), but rather consists of calculating an implied licensing rates based on Navarro's own license terms, the P&G revenues at issue, and "market-based" terms that he did not create, but is merely using. For that task, Hazel's relevant qualifications include his degree in accounting, his financial forensics certification, and his status as a certified public accountant. (Hazel Rep., Doc. 145-2, #4533). These quantitative skills equip Hazel to estimate a damages figure in the manner that he did.

Contrary to Navarro's apparent argument, there is no general requirement that a damages expert have subject-matter expertise in the underlying industry. For example, in Cooper Tire & Rubber Co. v. Farese, the court permitted a certified public accountant to estimate the change in value of a tire company even though he had no experience with the tire industry. 2008 WL 5188235, at *2-3 (N.D. Miss. Dec. 9, 2008). In so holding, the court explained that "the fact that [the expert] may not have specialized expertise with the tire industry does not preclude him from analyzing and offering expert testimony as to the value of alleged damages suffered by Cooper Tire." Id. at *3. The same principle holds true in intellectual property cases. Take for instance, Iconics, Inc. v. Massaro, where the court held that a certified public accountant/forensic services specialist was qualified to estimate copyright damages even though he had no experience "providing valuations of intellectual property in the context of litigation." 266 F. Supp. 3d 461, 467 (D. Mass. 2017).

The point is, quantitative experts have the requisite skills to evaluate damages in a quantitative fashion, even if they do not have industry-intensive experience that would allow them to rely on more subjective assessments of the "value" of something. To hold otherwise would suggest that a forensic financial analyst or accountant could not calculate any damages, except perhaps in cases concerning an accounting firm or consulting firm that does forensic accounting. Leon v. Kelly, No. CIV 07-0467, 2009 WL 1300936, at *17 (D.N.M. Jan. 1, 2009).

Navarro does not dispute that Hazel is a quantitative expert. As such, Hazel is qualified to testify in a quantitative fashion as to Navarro's actual damages, and to rebut Navarro's actual damages expert, Jeff Sedlik.

b. Hazel Is Unqualified To Offer Testimony As To Zeithammer's Surveys.

Hazel also offers his opinion that, "after reviewing the results of Zeithammer's survey conjoint analysis, [Hazel] concludes that the survey results for product packaging show no statistically significant evidence that the photos contributed in a positive manner to the sales." (Hazel Rep., Doc. 145-2, #4541). Navarro argues that Hazel is "unqualified" to render such an opinion because he "is not an expert on surveys and is not offering any expert opinions in this case on surveys." (Hazel & Bania Mot., Doc. 217, #11635). The Court agrees.

Whether an expert is qualified to offer an opinion is an opinion-by-opinion inquiry. An actuarial expert may well be qualified to opine on how long someone would have survived if they had not died in a fatal accident, but likely would not be qualified to offer opinions on accident reconstruction. That is not to suggest that experts cannot rely on opinions offered by other experts. In other words, a second expert can sometimes accept as an input to her analysis something that is an output from another expert's analysis. Such opinions could take the form—"Assuming [fact reported by Expert A] is true, I conclude that this result follows." Opinions of this variety do not necessarily offend Daubert's reliability requirement. In re E.I. du Pont de Nemours & Co. C-8 Pers. Injury Litig., 337 F. Supp. 3d at 743 (citing Walker v. Soo Line R. Co., 208 F.3d 581, 588 (7th Cir. 2000)) ("An expert is able to base an opinion on another expert witness for a point of expert knowledge not personally possessed.").

In such circumstances, the second expert is not verifying or confirming, the first expert's result, but merely relying on it. And, at least so long as it is the type of information on which experts in the second expert's field typically rely, Daubert, 509 U.S. at 598 (citing Fed. R. Evid. 703) (experts can rely on information that is considered reliable in their field), that is fine. See Jackson v. E-Z GO Div. of Textron, Inc., 326 F. Supp. 3d 375, 396 (W.D. Ky. 2018) (holding that an expert may use another expert's conclusions to support his own opinions). But, if the second expert lacks expertise in the first expert's field, she has no basis for opining on the correctness of the first expert's opinion.

Hazel is not a survey expert, and so he is unqualified to voice agreement with Zeithammer's surveys or the results of those surveys. Yet, Hazel's opinion does just that. Specifically, half of Hazel's "profit-apportionment" opinion consists of him describing Zeithammer's surveys, saying that he reviewed Zeithammer's report, and then agreeing that, based off those surveys, Navarro's photographs did not affect Defendants' sales. (Hazel Rep., Doc. 145-2, #4541). Hazel sums up his "opinion" on the matter by noting that Zeithammer's "results also provide reliable, quantifiable data to support the conclusion that the portion of the profits associated with any sales ... that is attributable to any alleged infringement is zero." (Id.).

Defendants claim that this opinion is admissible because it represents Hazel's own independent analysis of Zeithammer's survey. But that just highlights the problem. What are Hazel's qualifications for "independently analyzing" a survey? So far as the Court can tell, he has none. Indeed, he admitted as much at his deposition. (Hazel Dep., Doc. 217-1, #11690).

In sum, Hazel lacks the necessary qualifications to offer a "me too" opinion to buttress the validity of Zeithammer's surveys. Accordingly, Hazel's opinion as to the validity and correctness of Zeithammer's studies is inadmissible under Daubert.⁶

6 Although Hazel cannot offer his stamp of approval on Zeithammer's report, he can still reference Zeithammer's opinions, and may note that Zeithammer's findings dovetail with Hazel's own findings (so long as those findings are based on an area in which Hazel has expertise). Whirlpool Corp. Liab. Litig., 45 F. Supp. 3d at 741.

c. Hazel's Rebuttal To Larry Chiagouris Is Moot.

Hazel offers a two-paragraph rebuttal to Larry Chiagouris, Navarro's marketing expert. (Hazel Rep., Doc. 145-2, #4544). Specifically, Hazel takes issue with Chiagouris's conclusion that "Ms. Navarro's photographs contributed in a positive manner to the sales of each of the five alleged infringing products at issue." (Id. (citing Chiagouris Report, Doc. 182-1, #7843)). Navarro argues that Hazel is unqualified to rebut a marketing expert because "Hazel does not consider himself to be a marketing expert and has never offered an opinion as a marketing expert." (Hazel & Bania Mot., Doc. 217, #11636).

But, on November 20, 2020, the Court excluded the portion of Chiagouris's report where he discusses the particular impact of Navarro's photographs on P&G's sales because it was "entirely divorced from any cognizable methodology that connects his conclusions about Navarro's photographs to the facts in this case." (Op. & Order Partially Excluding Larry Chiagouris, Doc. 245, #13597). Hazel's rebuttal report is entirely focused on this, now excluded, expert opinion. Accordingly, Hazel's rebuttal opinion, and Navarro's motion in limine directed at that opinion, are now moot.

2. Hazel's Actual Damages Calculations Are Reliable , But His Sales Trend Calculations Are Not.

Hazel offers opinions both as to Navarro's actual and profit-based damages. Navarro objects to both opinions on the ground that they are unreliable. The Court rejects Navarro's objections to Hazel's actual damages calculations, but finds merit in some of Navarro's objections to Hazel's profit-based damages calculations (i.e., what Hazel calls his sales-trend analysis).

a. Navarro's Objections To Hazel's Calculations Of Actual Damages Go To Weight Not Admissibility.

Hazel calculates actual damages using three different models. In each, he purports to calculate a hypothetical daily use license fee for P&G's use of Navarro's photographs.

In his first model, Hazel started with the amount that Navarro invoiced P&G⁷ for the use of her photographs for certain time periods in certain markets. From that, he calculated a range of implied daily licensing fees. He then applied those daily licensing fees to what Navarro claims were the time periods and areas of the non-authorized usages, to estimate a range of the fees to which the parties would have agreed. (Hazel Rep., Doc 145-2, #4536-37). In the second model, he instead used the existing invoices to calculate a license rate as a percentage of the revenues P&G earned on the products through the licensed uses. (Id. at #4537-38). That is, he took the fee that Navarro charged and divided it by the associated revenues that P&G obtained on the covered product during the relevant time period. In that manner, he was able to derive a P&G-revenue-based estimate of the likely licensing rate, which he then likewise applied to the revenues that P&G earned through the allegedly unlawful usages. Finally, he estimated damages using three scenarios based on a comparison of licensing rates derived from rates reflected in Masterfile, which he describes as a "global visual content licensor," and Navarro's rates from her invoices for these photographs. (Id. at #4539).

7 In fact, as Hazel notes, the invoices went from Navarro to LPK, rather than P&G, but these were nonetheless the invoices for the photographs supplied from P&G. --------

Navarro claims that these opinions "lack reliability because they do not consist of any sort of methodology." (Hazel & Bania Mot., Doc. 217, #11647). The Court is not clear on what Navarro means by this argument. The methodology is straightforward and is explained in general terms above. Navarro may be correct that Hazel was unable to identify "any reported case that has approved this methodology, nor any expert who had relied upon it." But while that may be problematic (see discussion above regarding Sedlik's opinions), it is by no means dispositive on the matter. To be sure, if case law explicitly rejected such methods, or, more importantly, if the generally accepted view among intellectual-property damages experts was that such approaches are inherently flawed, that would be one thing. But the mere fact that this exact method has not been employed before (or, more accurately, that Hazel has not substantiated that it was), standing alone, does not require exclusion. How to reliably calculate damages in a given case depends on the particulars of the case, such as the types of comparators available. Here, Hazel had access to invoices for the same photographs, by the same photographer, sold (at least for some uses) to the same customer. It does not strike the Court as wholly unreliable to believe that information derived from such invoices may be useful in estimating the damages associated with other uses of those same photographs, by the same customer, not addressed in the invoices.

Take for example the expert in Oracle America, Inc. v. Google Inc., who valued a hypothetical license for the Java programming language "by starting with the real-world negotiations between [the parties] ... and then adjusting that amount ... based on ... revenue projections at the time." 847 F. Supp. 2d 1178, 1182-83 (N.D. Cal. 2012). The court held that these calculations were "sufficiently based on real-world facts" and "not speculative under Daubert." Id. at 1183. That is a version of one of Hazel's methods here. And Hazel's reliance on a proportion-of-revenues approach to estimating licensing rates also finds support in caselaw. For example, when Pandora, the music streaming service, and a composer were struggling to agree on a copyright licensing fee, they proposed to the court several different licensing rates, each expressed as a percentage of Pandora's revenue. In re Pandora Media, Inc., 6 F. Supp. 3d 317, 320 (S.D.N.Y. 2014). Accordingly, the Court concludes that Hazel's invoice-based and revenue-based methods are not "unreliable" under Daubert.

Navarro's attack on Hazel's third methodology fares no better. She argues that Hazel's methods are unreliable because his market rates came from a vendor that sells stock photos as opposed to made-to-order photos like Navarro's. (Hazel & Bania Reply, Doc. 226, #13092). The argument essentially suggests that Hazel failed to use "sufficient facts or data" to support his testimony. Fed R. Evid. 702. But flaws in the underlying data set justify excluding expert testimony under Daubert only when "the underlying data are so lacking in probative force and reliability that no reasonable expert could base an opinion on them." Ellipsis, Inc. v. The Color Works, Inc., 428 F. Supp. 2d 752, 760 (W.D. Tenn. 2006) (quoting Mohney v. U.S. Hockey, Inc., 300 F.Supp.2d 556, 565 (N.D. Ohio, 2004)).

Here, Hazel used "similar images" from a large vendor of commercial photography. (Hazel Rep., Doc. 145-2, #4539). Although prices for these images may not be exactly the same as Navarro's made-to-order commercial photography, that is not the question. Rather, the question here is whether those prices may provide some insight into what the market-rates would have looked like for Navarro's photographs. The Court cannot say that such data is entirely "lacking in probative force" or so "fundamentally flawed" as to warrant exclusion under Daubert. See Ellipsis, 428 F. Supp. 2d at 760. This objection, as with most objections to the factual basis of an expert opinion, "goes to the credibility of the testimony and not to its admissibility." Id.

Navarro's final objection to Hazel's actual damages analysis argues that, when Hazel applied the price/revenue ratio to P&G's future revenues, Hazel contemplated that P&G would compensate Navarro for all the future revenues it earned while her photograph was on the packaging. This resulted in Hazel calculating hypothetical licenses based on partial years, not full years. Navarro says that Hazel's calculations are unreliable because it was not Navarro and P&G's typical practice to negotiate partial-year licenses. (Hazel & Bania Mot., Doc. 217, #11647-48). This does not undermine the reliability of Hazel's report for two reasons. First, if Hazel estimated whole-year licenses, he may have over or under-calculated the value of a hypothetical license. It may actually be more accurate for Hazel to calculate Navarro's actual damages according to the precise dates of the alleged infringement. Next, as Defendants point out, any over or under-estimate can be "easily adjusted" to allow for full-year license. (Hazel & Bania Resp., Doc. 221, #12469). Accordingly, this objection, too, goes only to weight and not admissibility.

b. Hazel's Opinion Identifying Alleged "Trends" Is Unreliable Under Daubert.

Hazel estimated the profits attributable to Navarro's images using what he refers to as sales trend analysis. (Hazel Rep., Doc. 145-2, #4541-42). In essence, Hazel compared sales trends between the period when P&G used Navarro's images on the packaging and the period when P&G removed Navarro's images from the packaging. (Id. at #4542). Hazel then observed that the patterns of decline and growth followed what he opined was a "similar trend." (Id.). From this, Hazel opined that Navarro's images did not have an impact on P&G's sales. (Id.).

Navarro argues that this opinion is not based on a reliable methodology because (1) Hazel did not produce a statistically accurate trend line, and (2) Hazel should have used regression. (Hazel & Bania Mot., Doc. 217, #11641-44). In a sense, Navarro's first objection is encompassed within her second objection, because the way to get a statistically accurate trend line is to perform a regression. Therefore, the Court takes both arguments as objections to Hazel's use of what he calls a sales trend model, as opposed to a regression model.

A multiple regression analysis is a well-established statistical method that measures the effect of multiple independent variables on a dependent variable. It can be a useful tool in copyright infringement cases to estimate the effect of the infringing material on sales. The Iams Co. v. Nutro Prods., Inc., No. 3:00-CV-566, 2004 WL 5496244, at *2-3 (S.D. Ohio June 30, 2004) (describing regression analysis).

Sales trend analysis, at least as Hazel describes it, is different. Apparently, it attempts to undertake a qualitative analysis of the data to see whether trends "appear similar." The Court is concerned, though, that there does not appear to be any set of metrics, or at least Hazel did not identify any, for determining what constitutes "similar." The extended colloquy at Hazel's deposition regarding the "trend line" was interesting in that regard. Hazel was quick to concede that the trend line did not reflect any type of regression analysis, but he appeared to have a much more difficult time describing the methodology that he used in determining where the line should be placed on the revenue graph. At the end of the day, it appears that he just chose the high point at the end of the product "ramp-up" period, and then drew a straight line to another point (or, in the case of one chart, near another point) at the other end of the revenue timeline. Indeed, he conceded at his deposition that little, if any, expertise was involved in drawing the trend line, and that a ten-year-old likely could have done the job. (Hazel Dep., Doc. 217-1, #11670).

That gives the Court pause. To be sure, Hazel's underlying revenue data appears reliable, and no one seems to dispute that. But, for an expert to claim that revenue trends are "similar" over different time periods seems like it would require at least some basis for defining what constitute "trends" and what makes those trends "similar." Hazel has offered remarkably little on that front.

On a related note, Hazel has provided little if anything to tie the alleged "similarity" in before-and-after trend lines to the issue of causation. That is, as he concedes, there are a number of reasons why sales of a given product may change over time. For example, income changes among the relevant customer group, emergence of competitors, and reduction in advertising, just to name a few, are all factors that could change consumer behavior. Hazel says he took many such things into account in his "trend analysis," but the specifics around that are remarkably sparse. And, without specifics, it is difficult at best to ascertain whether "similar trends," even if Hazel could have identified how he was using that term, in turn also implies an absence of causation.

Accordingly, while the Court concludes that Hazel is free to present, and testify as to, the revenue timelines on the charts in his report, he must remove the unreliable dotted "trend line." Moreover, Hazel cannot opine as to the similarity of the two revenue curves, nor may he testify as to whether the presence or absence of the photographs on the product packaging "caused" or did not "cause" any change in consumption patterns for the Olay products. As a result, he also cannot testify as to whether "the profits associated with the sales of the Allegedly Infringing Products that is attributable to any alleged infringement is zero." (Hazel Rep., Doc. 145-2, #4543). Hazel has simply not identified any meaningful basis for an expert opinion on that issue.

3. Doug Bania Is Qualified To Offer Expert Testimony.

Navarro argues that Doug Bania, Defendants' rebuttal expert on actual damages, is unqualified for many of the same reasons that she argues that Hazel is unqualified. Namely, Navarro contends that Bania is not an expert on photography licensing, and as such, cannot opine on actual damages based on hypothetical licensing negotiations. (Hazel & Bania Mot., Doc. 217, #11636).

Navarro also takes issue with specific portions of Bania's testimony where he opines as to what may or may not be reasonably included in a license. For example, Bania's expert report lists several components of "traditional intellectual property license transactions," like quality control-provisions or provisions to cure breach of license, and then opines that Navarro's invoices lack these key elements. (Bania Report ("Bania Rep."), Doc. 217-2, #11784-75). At another point, Bania says that "[i]n my professional experience I have never encountered a situation where the seller did not allow the buyer to use a picture in advertisements for the products." (Id. at #11788). Bania is not qualified to offer these opinions, Navarro argues, because he is not a photography licensing expert and because these "opinions also inappropriately cross the line into rendering an opinion on the ultimate legal issue in the case." (Hazel & Bania Mot., Doc. 217, #11637).

Navarro's argument that Bania is not qualified to opine on hypothetical licenses or licensing procedures is unpersuasive. Bania is an accepted intellectual property licensing expert who has testified in seventy intellectual property cases, several of which required him to calculate copyright damages. (Bania Rep., Doc. 217-2, #11792-97). He also is a certified licensing professional and principal of a consulting firm that specializes in licensing and valuation expert services. (Id. at #11797-98). Navarro argues that this IP licensing experience is irrelevant because Bania does not have expertise negotiating photography licensing rights. (Hazen & Bania Mot., Doc. 217, #11636). But Bania does not need photography-specific IP experience. His long list of licensing experience is more than sufficient to qualify him to opine on the hypothetical licenses in this case.

Finally, Bania, like all other expert witness, is not permitted to offer "legal conclusions" in the form of expert opinions. Alvarado v. Oakland Cnty., 809 F. Supp. 2d 680, 688 (E.D. Mich. 2011). Hazel's report, however, stops short of expressing an opinion on the ultimate issues in the case. Rather, Hazel opines on factual issues, (such as whether a willing seller/buyer would negotiate for rights to advertise a product), that may have a bearing on an ultimate legal issue, (such as whether Navarro is entitled to damages). These opinions are permissible under Daubert. Numatics, Inc. v. Balluff, Inc., 66 F. Supp. 3d 934, 944 (E.D. Mich. 2014).

Accordingly, the Court DENIES Navarro's Motion to Partially Exclude Doug Bania, and GRANTS IN PART AND DENIES IN PART Navarro's Motion to Exclude Steven Hazel. Specifically, the Hazel may not add his "stamp of approval" on Zeithammer's report, nor may he offer opinions as to the existence or similarity of "trends" in Defendants' sales in various time periods. Hazel may, however, offer all of his other opinions, including his actual damages estimates and rebuttal to Sedlik.

CONCLUSION

For the reasons discussed above, the Court GRANTS IN PART AND DENIES IN PART Defendants' Motion to Exclude the Expert Testimony of Jeffrey Sedlik, (Doc. 214), DENIES Navarro's Motion to Exclude the Expert Testimony of Robert Zeithammer, (Doc. 218), and GRANTS IN PART AND DENIES IN PART Navarro's Motion to Exclude the Expert Testimony of Steven Hazel and to Partially Exclude the Expert Testimony of Doug Bania, (Doc. 217).

SO ORDERED. March 8, 2021
DATE

/s/ _________

DOUGLAS R. COLE

UNITED STATES DISTRICT JUDGE