|
Evidence - Survey Evidence. Promotion in Motion, Inc. (PIM Brands, Inc.) v. Hershey Chocolate & Confectionery LLC
In Promotion in Motion, Inc. (PIM Brands, Inc.) v. Hershey Chocolate & Confectionery LLC (Fed CA, 2026) the Federal Court of Appeal dismissed an appeal, this brought against the earlier dismissal of a Federal Court appeal, and that against the refusal of the Registrar of Trademarks to register a trademark.
Here the court considers issues of 'survey evidence':[61] The Federal Court was right to point out that consumer surveys have gradually been admitted into evidence as an exception to the hearsay rule. As the Supreme Court stated in Mattel Inc. v. 3894207 Canada Inc., 2006 SCC 22 [Mattel] at paragraph 43:Until comparatively recently, evidence of public opinion polls were routinely held to be inadmissible because it purports to answer the factual component of the very issue before the Board or Court (i.e. the likelihood of confusion), and in its nature consists of an aggregate of the hearsay opinions of the people surveyed who are not made available for cross-examination, see, e.g., Building Products Ltd. v BP Canada Ltd. (1961), 1961 CanLII 869 (CA EXC), 36 CPR 121 (Ex.Ct.); Paulin Chambers co. v. Rowntree Co. (1966), 1966 CanLII 970 (CA EXC), 51 CPR 153 (Ex.Ct.). The more recent practice is to admit evidence of a survey of public opinion, presented through a qualified expert, provided its findings are relevant to the issues and the survey was properly designed and conducted in an impartial manner. [62] When the parties propose to introduce expert evidence, it is the duty of a trial judge to determine the admissibility of that evidence having regard to the criteria developed in Mohan and Masterpiece, especially the necessity and relevance of that evidence. Survey evidence, as a species of expert evidence, does not detract from that rule. Relevance, with its attendant requirements of validity and reliability in the context of survey evidence, is a question of appreciation over which the trier of facts must be left with a wide margin of appreciation.
[63] The Federal Court judge accurately pointed out that the admissibility of survey evidence at issue in Mattel and later discussed in Masterpiece was within the context of in-person surveys. When such surveys are conducted, the interviewer can attest to the survey participant being the participant whom the participant claims to be, and can ensure that the person is answering the particular questions without consulting any other person or device, and is therefore free from external influence. It is also easier to verify that the person interviewed falls within the intended demographic for the study. This is obviously not the case when the survey is conducted online. Is that sufficient to conclude, however, that online Internet-based panels are fundamentally flawed, and that they should not be admitted unless there is a "“person in the loop”" or "“video recoding of the participant filling out the survey”"?
[64] I am not sure that such a bald conclusion was warranted, in the absence of a more fulsome debate and record. After all, surveys conducted by phone are now commonly used in courts, despite being open to the same criticisms levelled by the Federal Court judge with respect to online surveys. The respondent is correct to point out that the concerns raised by the Federal Court judge were also brought up by Dr. Corbin, the appellant’s own survey expert, in an article that was entered into the record before the Federal Court (Himanshu Mishra & Ruth M. Corbin, "“Internet Surveys in Intellectual Property Litigation: Doveryai, No Proveryai”", The Trademark Reporter, 107(5) (September-October 2017) 1097). In that article, however, Dr. Corbin points to other benefits that online surveys yield that are too valuable to ignore, including cost-efficiency, wider coverage of populations than in-person surveys could ever hope to achieve, and avoidance of the potential "“interviewer effects”" that can occur in telephone or in-person surveys. She also presented guidelines and safeguards to address the specific vulnerabilities of Internet surveys and to ensure that they can be trusted as a basis for reliable inferences about consumer perceptions, most of which were adhered to in the surveys designed by Dr. Corbin and Mr. Bourque.
[65] To increase Internet survey reliability, Dr. Corbin suggests three strategies that experts can implement to improve both reliability and validity: 1) choosing a panel company that will strive to deliver accurate information about panel participants and that employs sampling procedures that will increase representativeness of the population to be tested; 2) including questions to check for faking, inattention and, cheating; and 3) engaging in post-survey validation to provide assurance of legitimate respondent identities.
[66] In the case at bar, the Federal Court judge accepted that both Dr. Corbin and Mr. Bourque were properly qualified experts in survey market research and analytics, and the surveys were run in accordance with industry standards and subject to randomized verification after completion. While none of these safeguards can provide complete assurance that the responses were given by the participants themselves and that they were free of external influence and information, I am not convinced that these flaws were so critical as to be a fundamental issue going to the admissibility, as opposed to the weight, of these surveys. As imperfect as they may be, Internet surveys, just like other types of surveys, may sometimes provide valid evidence to inform the confusion analysis when properly designed and implemented. I would be reluctant, at this early stage, to unequivocally rule out this possibility irrespective of the particular facts of each case and of each survey. The admissibility of online evidence, just as any other type of evidence, is an issue that is to be determined on a case-by-case basis; in my view, the Federal Court judge erred in concluding that the absence of a "“person in the loop”" (as is the case, by definition, in any online survey) was a factor to exclude the Corbin affidavit and the Bourque affidavit without any further analysis of the circumstances in which the surveys were conducted. That flaw, however, is not determinative and does not amount to an overriding error.
[67] Finally, the appellant has failed to convince me that the Federal Court judge misapplied the case law with respect to the rate of confusion that is sufficient to establish a likelihood of confusion. The appellant claimed that no Canadian court has held in the last 20 years that a "“net confusion”" rate of 6.1% rises to a level of actionable confusion, and that the Federal Court judge erred in relying on a thirty-year-old Alberta Court of Appeal’s decision dealing with identical marks. There are several counter-arguments to that position.
[68] First of all, the Brigley affidavit filed by the respondent actually showed confusion among approximately ten percent of survey participants. The Federal Court accepted, for argument’s sake, the corrections to the percentages of the test group participants and the control group participants suggested by the appellant’s counsel, as well as the notion that Ms. Brigley should have subtracted the control group from the test group. A control test is designed to measure the level of confusion that would exist in the marketplace in general, unrelated to the factors in dispute. It is on the basis of these assumptions that the Federal Court judge arrived at the 6.1% figure. Yet, according to one of the appellant’s own experts (Mr. Bourque), this notion of "“net level of confusion”" is not universally used in the industry, and serves to establish a minimum level of confusion. The 6.1% figure must therefore be considered with these caveats in mind.
[69] Moreover, there is no basis for the appellant’s assertion that a 10% threshold (net or control) is typically considered necessary to show actionable confusion in Canada and is the prevalent standard in Canada. The mere fact that the Alberta Court of Appeal decision in Triple Five Corporation v. Walt Disney Productions, 1994 ABCA 120 was rendered thirty years ago is obviously not sufficient to discard it. It has, indeed, been cited with approval both by the Trade Mark Opposition Board and the Federal Court on numerous occasions: see for example, Canada Post Corp. v. Paxton Developments Inc., 2000 CanLII 16762 (FC); Novopharm Limited v. Eli Lilly and Company, 2004 CanLII 71728 (CA TMOB); Diageo Canada Inc. v. Heaven Hill Distilleries, Inc., 2017 FC 571; Imperial Tobacco Canada Limited v. Philip Morris Brands SARL, 2018 FC 503. I further agree with the Federal Court judge that whatever level of confusion is considered as sufficient to establish a likelihood of confusion in the United States, this jurisprudence does not bind Canadian courts.
[70] It is also worth adding that the threshold issue in the context of the present case is somewhat of a red herring. Surveys where respondents are asked to name the source of a product, commonly referred to as "“named-source test”", are not designed to show an absence of confusion, but merely to measure the level of confusion. This is precisely why some experts, among which Dr. Corbin in one of her articles (Ruth M. Corbin & Arthur Renaud, "“When Confusion Surveys Collide: Poor Designs or Good Science”"(2004) 94:4 Trademark Rep at 792), have advised defendants in trademark opposition proceedings to stay away from such surveys because they cannot prove that there is no likelihood of confusion. Bearing in mind the broad scope of options defendants have in designing a survey to lower the chances of finding confusion in a named-source test (for example, by narrowing or broadening the population identified for a survey, or creating conditions deviating from a realistic simulation of how the properties of a product are encountered by the actual relevant population), a failure to find confusion in such a survey cannot prove the opposite. The percentage of confusion in a name-source test is therefore irrelevant to the question ultimately to be decided by the Board or the Federal Court.
[71] For all of the above reasons, I am therefore of the view that the Federal Court did not make a palpable and overriding error of fact in rejecting the survey evidence filed by the appellant and the respondent on the appeal from the Trademarks Opposition Board. While the Federal Court judge seemingly went too far in excluding online Internet-based panel surveys to the extent that they do not involve a "“person in the loop”", the other reasons that she gave to come to her conclusion are based on factual findings that are grounded in the record and that are entitled to deference.
|
