Roberta Jacobs-Meadway
Ballard Spahr Andrews & Ingersoll LLP
United States
Survey evidence in trademark litigation
The likelihood of confusion – the test for trademark and litigation. The Federal Judicial Center’s Reference Guide on
unfair competition infringement – is subjective. However, Survey Research describes the use of surveys in trademark
objective evidence of the likelihood of confusion may be infringement cases as routine.
secured through the conduct of a survey. Similarly, Admissibility is restricted by Daubert v Merrill Dow
whether an alleged mark is descriptive or generic (which Pharmaceuticals (509 US 579 (1993)) and Kumho Tire v
depends on the perceptions of the relevant purchasing Carmichael (526 US 137 (1999)) and governed by Federal
class) can be established through objective testing by Rule of Evidence 702. It depends on:
survey. Just as it is claimed that advertising materials carry
a false implication rather than literally being false, in the • the qualifications of the survey expert;
same way survey evidence may provide an objective view • the helpfulness of the testimony to the facts; and
of the message conveyed by the challenged claim. • the reliability and fitness of the testimony.
At first, it appears that surveys should be conducted
by both parties in trademark litigation, where the Reliability and fitness depend on the trustworthiness
question generally turns on the principles of the relevant of the survey itself, which in turn depends primarily on
public and trade as to source or affiliation. whether the right questions were put in the right way to
the right respondents.
Survey evidence conundrum Before Daubert and Kumho, perceived deficiencies in
Lawyers in trademark litigations who contemplate survey research were generally seen as coming from
offering survey evidence face something of a dilemma. weight rather than admissibility, although survey
On the one hand, some courts expect a financially evidence could be and was excluded on various grounds,
able party to produce a survey and will draw an adverse primarily based on Federal Rule of Evidence 403 (that the
inference from the failure to offer a survey. Some courts danger of prejudice and confusion outweighs any
give less weight to a critique of a survey by a party that probative value). Subsequently, the courts have continued
does not offer its own survey showing divergent results. to admit into evidence surveys that have a perceived
It is easy to point out small errors, but a failure to show methodological deficiency, but are sufficiently probative
that the results change with a better protocol or universe of a matter at issue. However, when the deficiency
can render such an attack less credible. significantly reduces the weight given to the survey, the
On the other hand, about half of all surveys offered admission of the survey is decidedly Pyrrhic.
are accorded little or no weight, even if they are not
excluded. Using a previously approved methodology is Avoiding a survey
no assurance of acceptance and using a previously Inevitably, the other side will point out some
lauded expert does not guarantee that the opinion will methodological or other deficiency. It seems impossible
not be savaged. to design, conduct and report on a survey and not leave
This chapter addresses this conundrum: how to open some element which the other side’s survey expert
avoid doing a survey and, if it cannot be avoided, what will seize and trumpet as the ‘fatal flaw’ undermining the
to do to avoid what survey opponents will term the fatal value of the entire project. Therefore, the first question to
flaws in the research. consider is when you can avoid doing a survey.
A defendant can avoid carrying out a survey to
Legal history establish likelihood of confusion if the plaintiff has not
Survey evidence is frequently used in trademark offered one. A plaintiff can avoid doing a survey if it moves
92 Building and enforcing intellectual property value 2007
Ballard Spahr Andrews & Ingersoll LLP United States
quickly, before the defendant is established in the market. questions asked attack the issue, the universe is
A plaintiff can also avoid doing a survey if the registration appropriate, the control makes sense and there are no
on which it relies is more than five years old, meaning that obvious gaffes in question wording or order. Once a
the mark cannot be challenged as merely descriptive. survey expert is engaged, the types of research to be used
However, there are times when it is difficult to avoid must be chosen.
offering a survey, such as:
Types of research
• when the parties have co-existed in the marketplace Survey research may be qualitative or quantitative.
for long enough that the absence of actual confusion Qualitative research (eg, focus groups) may be relied on
evidence may raise a real question for the finder of by businesses to justify or confirm decisions, but cannot
fact as to whether confusion could be likely; be relied on in court in an infringement case. Therefore,
• when the mark asserted is a single colour and the for the purpose of litigation the practitioner should
question is whether the colour has acquired consider quantitative research or observational studies
distinctiveness; and (which are a separate type of research, and not really a
• when the element allegedly copied is presented survey). Quantitative research is research using a
generally with other, uncopied elements, and the probability sample where the reliability of the members
question is whether, in context, the similarity can be determined at some confidence level. This
outweighs the differences. involves surveying respondents by phone, online or by
post. Questioning people in shopping malls cannot be
Of course, this list is not exclusive. Trademark used for a probability sample, as this method permits
infringement cases are invariably fact sensitive. The only a convenience sample. However, the courts accept
point is that there are times when the expense and risk the use of mall intercepts as the interviewer is more likely
involved in offering a survey are worth undertaking. to find people in a buying mood, so it is easier to
replicate market conditions. In fact, while it is good to
Timing have people in market conditions, any survey is artificial
When should the survey be carried out? A strong case and not a real reflection of the marketplace. For that
can be made that the survey should be done before suit is reason, survey results may indicate the likelihood of
filed. What may be perceived as a problem may turn out confusion, but show nothing as to actual confusion of
not to be, or at least not to be enough of a problem to purchasers at the point of sale.
warrant the expense of litigation. If research shows a low
level of confusion or no service recognition, measures Proper universe
other than filing suit may be appropriate. Is it only purchasers at the point of sale who are relevant
If the survey is not considered until after suit is to the issue to be surveyed? Precedent and the statutory
brought, it may be best to delay designing the survey language and legislative history suggest the relevant
until after some initial discovery in order to see: universe is not limited to purchasers or the point of
purchase. Notwithstanding this, it has become almost a
• whether the other party has done research and, if so, convention to restrict the proper universe for surveying
what it shows; research to purchasers and potential purchasers of the
• how the other party defines its market; and parties’ products.
• where the other party has distribution. Any arbitrary exclusion of a segment of the relevant
market will cast doubts on the validity of the results – for
This information may help to avoid mistakes in example, if the purchasers include men and women, a
survey design that will be exploited by the other party. universe that includes only women is suspicious. If the
product is marketed primarily to people under 24, the
Carrying out a survey universe should be selected to reflect that limit. The
Once it has been decided to go ahead with a survey, the closer the selected sample to the demographic of the
first step is to establish the issue the survey is to address. relevant public or trade, the more likely it is that the
The next is to engage a survey expert. The expert chosen universe will be accepted by the court as proper.
may depend on availability, geographic proximity, prior Assuming that the proper universe is the segment of
experience or a recent case. The expert must also work the population whose perceptions are relevant to the issues
well with the counsel; while the expert designs and in the case, that universe is then narrowed by the selection
conducts the survey, the attorney must ensure the of the sample from that universe to be questioned.
Building and enforcing intellectual property value 2007 93
United States Ballard Spahr Andrews & Ingersoll LLP
Conducting interviews never been marketed. Noise is a related problem. Stimuli
Once the universe and the sample size are set (generally in the market are more diverse than the stimuli presented
at least 200 respondents in the test cell and an equal in a controlled research setting. The respondent brings to
number in any control cell), there is the question of how bear a vast array of experience in the market. This past
to conduct the interviews. In a trade dress case, the experience, whatever the limitations placed upon the
stimulus may well need to be shown. Where the issue is stimuli presented in the interview context, may filter
genericness or descriptiveness of a word, a verbal through and affect the results of this study.
stimulus may well suffice. There are practical Both spurious awareness and noise problems may be
implications. Telephone surveys of the general reduced or eliminated by proper controls. These controls
population can be carried out quickly and efficiently. If may be questions designed to identify and take account
there is a need for personal interviews of a specific of past experience, or the presentation of additional
population (eg, cardiologists), the process becomes more stimuli to respondents to see whether the same answers
time consuming and costly. are irrespective of the specific stimuli presented. For
The stimulus shown or given to respondents is example, where one company is highly dominant in a
another point where an attack may be made. If a picture field a respondent may think that company is responsible
of a product is shown, are irrelevant features redacted? It for all products of the same type, irrespective of
is safest to show the actual product – but what if that trademark and/or trade dress, because that is the
makes the survey a reading test? And what if the alleged company associated with that product category. In this
word appears on multiple products? Which products instance, a survey result may show generic use of a mark
should be shown? It may be better for the party to or name for the product category.
describe in general terms what it wants to know about It is important to ask only what is essential. To seek
rather than show a stimulus, product or picture that does interesting but non-essential information risks:
not represent what respondents may have seen.
• conditioning respondents by the order and nature of
Questionnaire questions;
Next the questionnaire has to be written. Care must be • securing information that contradicts or otherwise
taken to avoid leading questions or asking the wrong calls into question the data sought;
questions. Which kind of questions are faulty? Examples • prolonging the interview, confusing the respondents
are legion. In Trouble v Wet Seal respondents were asked and/or the interviewers; and
if they thought that the company making Arden B • raising the issue of relevance.
clothing was affiliated, associated or connected with any
other women’s clothing company. This was found not to Both the design and analysis should be kept simple.
test confusion (eg, whether people would be confused as A complicated questionnaire provides a larger target for
to affiliation if they encountered both marks in issue) attack. In addition, the issue to be determined may be
unless a respondent spontaneously named the plaintiff; lost in a welter of questions, responses and cross-
rather, it tested awareness of other clothing lines and tabulations. The object of the exercise is to secure clear,
company affiliations. objective evidence persuasive to the court, not to
Where the issue is secondary meaning, a survey produce a dissertation.
showing that people associate the element at issue with If the issue turns on alleged similarities between the
the plaintiff does not establish that they believe that all marks in visual appearance, respondents can provide no
products having such an element originate from the meaningful response to descriptions of the marks in a
plaintiff. telephone survey. Instead, respondents need to be able to
It is also necessary to avoid a question order that see the marks or accurate reproductions.
leads to a particular result. To the greatest extent possible the data secured
should speak for itself without interpolation,
Other issues interpretation or argument. In this connection, closed
There are other points to bear in mind with respect to questions avoid problems with codification and
survey design for trademark litigation. A control should interpretation. The risk of using closed questions to the
be employed to address the related issues of spurious exclusion of open questions, however, is that the
awareness and noise. Spurious awareness is a well- question cannot suggest the answer and the answer
documented phenomenon. People will, without categories must be fairly chosen. Otherwise the
deceptive intent, claim awareness of brands that have respondent is compelled to select the answer that comes
94 Building and enforcing intellectual property value 2007
Ballard Spahr Andrews & Ingersoll LLP United States
closest to the answer he or she would actually choose to difficulty of locating respondents and securing their
give if presented with other alternatives. cooperation. (How often do you want to stand around
The data must be presented accurately and fairly. What answering survey questions?) Offering survey results is
if the data resulting from the survey or a pre-test of it is no guarantee that the court will accept the survey or its
unfavourable? What must be disclosed? If the expert is a conclusions. There is, however, no substitute for a well-
non-testifying expert and the work has been conducted by conducted survey to ascertain the perceptions of the
counsel, the research is a product of the attorney’s. In such relevant public. Why be relegated to speculation as to the
cases the expert may not be used as an expert witness to significance of a term or the likelihood that purchasers
critique a survey by the other side, although he or she may will assume a common source when the right questions
provide some guidance to counsel. can be posed to provide objective germane evidence?
Consider the cost. Consider the difficulty of
Conclusion designing an appropriate survey. Consider the level of
There is no requirement that a survey be offered in certainty in the result. Discuss with the expert and read
trademark litigation. Offering survey research adds a the decisions. Surveys are yet another legal tool –
layer of expense which may be significant, depending on sometimes a good idea, sometimes not, but always to be
the number of respondents, the number of sites and the approached with caution.
Roberta is a member of the litigation department and the licensing, Roberta Jacobs-Meadway
biotechnology and IP group. Her practice focuses on trademarks, copyright Partner, Philadelphia
and unfair competition issues, and includes litigation in the federal courts Tel +1 215 864 8201
and before the Trademark Trial and Appeal Board. Roberta holds a BA from Email
Bryn Mawr College (1972) and a JD from Rutgers University School of Law jacobsmeadway@ballardspahr.com
(1975). She has written and lectured extensively on IP law. Roberta co-chairs Ballard Spahr Andrews & Ingersoll LLP
the ALI-ABA programmes on litigating trademark and unfair competition United States
cases, on internet law for the practical lawyer and on IP licensing.
Building and enforcing intellectual property value 2007 95