Rosetta Stone LTD v. Google Inc.

Filing 115

Declaration of Edward A. Blair, Ph.D. re 112 MOTION for Summary Judgment (portions filed under seal) by Google Inc.. (Attachments: # 1 Exhibit A)(Frieden, Jonathan) Modified text on 3/30/2010 re Received under seal copy in black binder on 3/29/10. (UNDER SEAL)(klau, ).

IN THE UNITED STATES DISTRICT COURT
FOR THE EASTERN DISTRICT OF VIRGINIA
(Alexandria Division)
ROSETTA STONE LTD.
Plaintiff,

,

: CIVIL ACTION NO. I :09cv736 (GBL I
: TCB)

v.

GOOGLEINC.
Defendant.
DECLARATION OF EDWARD A. BLAIR, PH.D.

I, Edward A. Blair, declare as follows:
I.

I am over eighteen (18) years of age. The statements made herein are true

and correct and are of my own personal knowledge. I make this declaration in
connection with Defendants' Motion for Summary Judgment and Motion to Exclude the
Expert Report of Kent Van Liere.
Background

2.

I am Michael J. Cemo Professor of Marketing & Entrepreneurship and

Chairman of the Department of Marketing and Entrepreneurship in the C.T. Bauer
College of Business at the University of Houston in Houston, Texas. A professional
resume showing further information is attached as Appendix I to the Expert Report I
submitted in this action.
3.

I was asked to evaluate the survey and report provided in this matter by

Dr. Kent Van Liere and provide my opinion as to the likelihood of confusion in this
matter. In connection with this engagement, in addition to relying upon my experience in
the field generally, I reviewed the Complaint, Defendant's Answer to Plaintiffs
Complaint and Affirmative Defenses, Expert Report of Kent VanLiere and its attached
exhibits, an Excel file of Dr. Van Liere's survey data, and Internet search results.
I

4.

I wrote a report, dated January 20, 20 I 0, detailing my considered opinion

regarding Dr. Van Liere's survey, as well as the basis and reasons therefore, a true and
correct copy of which is attached as Exhibit A. This report accurately reflects my true
opinion.
5.

Since the completion of my report, I have reviewed certain additional

information, including the deposition transcripts of Dr. Vau Liere and Van Leigh, who I
understand to be a Rosetta Stone employee. I have also reviewed a collection of screen
shots that I understaud Rosetta Stone provided to Dr. Van Liere aud from which he chose
the screen shot to base the test and control conditions of his survey, and a Prospectus
filed with the SEC by Rosetta Stone.
6.

In view of this additional information, I submit this declaration to provide

more detail and numeric specificity to the issues I discussed in my deposition and my
report.
Dr. Van Liere erroneously counted respondents who provided disqualifying
answers to a filter question as if they were respondents to the confusion questions.

7.

As noted in my report, the control condition Dr. Van Liere used was

flawed because the control listings did not simply differ from the test listings in being
unpaid vs. paid; they also differed in the nature of the sites listed. Most of the test listings
were for sites that appeared to sell Rosetta Stone products or offer related promotions,
while most of the control listings were for sites that relate to the historical Rosetta Stone
artifact, and none of the control listings except the company website appeared to sell
Rosetta Stone products or offer related promotions. Dr. Liere could have mitigated this
problem by creating a more balanced control. One way he might have done this was by
using search results from a query that was not for "Rosetta Stone" alone, but was more
clearly tied to language learning software, such as "Rosetta Stone software" or "Rosetta
Stone Spanish." I understand that the screen shots provided to him by Rosetta Stone

2

included screen shots of such results. However, Dr. Van Liere chose a screen shot
without any commercial Rosetta Stone links in the organic results except for its own.
8.

In his deposition, Dr. Van Liere indicated that he attempted to address the

imbalance in the links with a preliminary "filter question" that he said was intended to
focus respondents' attention on commercial listings. See Van Liere Deposition 90:9-14.
This filter question was: "Which link or links, if any do you think sells Rosetta Stone
language software products?" See Van Liere Deposition 89:23-90:14. Respondents who
indicated that a link did not sell Rosetta Stone products were filtered out, and were not
asked the confusion questions for that link.
9.

However, in calculating his results, Dr. Van Liere ignored the fact that

respondents were filtered, and many respondents were therefore not even asked the
confusion questions for any of the target links. Not surprisingly, there were more of
these respondents in the control condition. This is because respondents in that
condition-where only two of the eleven links actually referred to Rosetta Stone, the
plaintiff here, or its products-were less likely to identify any of the target links as selling
Rosetta Stone products.
10.

In ignoring the filtering, Dr. Van Liere treated respondents who said "no"

to the filter question as if they said "no" to the confusion questions. In his calculation,
these respondents contribute to the denominator (or base) of the confusion rate, but not
the numerator, exactly as if they had been asked the confusion questions and said "no."
To put it another way, these respondents are treated as if they could have indicated
confusion but did not. This is inappropriate. The filter question was not intended to
provide evidence regarding confusion, but rather to address the imbalance in test and
control links by focusing respondents on commercial listings. The proper base for
calculating the confusion rate is those respondents who were actually asked the confusion
questions for the target links.

3

II.

If one takes the filtering into account, and calculates confusion rates

among only respondents who were actually asked the confusion questions for the target
links, confusion rates in the test and control conditions are as follows.

llA. Confusiou results for respondents who were asked the
confusion questions for at least one target link
Test

% Confused
12.

Control

Net

75%

73%

2%

Therefore, with the filtering question appropriately treated as a filter, there

is no significant difference in confusion rates between the test and control conditions.
This shows that the difference in confusion rates reported by Dr. Van Liere between the
test and control conditions is attributable to the fact that many respondents in the control
condition were not even asked the confusion questions.
Dr. Van Liere improperly couuted as coufused ou the "endorsement" question
those who explained that their response was based on the belief the advertisers sold
Rosetta Stone products.

13.

As discussed in my report, use of the word "endorsement" in the survey

was problematic. This is due in part to many consumers' belief that a company reselling
a product is endorsed by the provider of that product. This is reflected in the
respondents' narrative responses explaining why they replied that a particular advertiser
was endorsed by Rosetta Stone. Many respondents gave some version of the following
reasons: the site sells Rosetta Stone products, offers discounts/coupons/rebates on Rosetta
Stone products, and/or is a reputable merchant that sells many products. If respondents
who gave these answers are excluded from the confusion calculation, but Dr. Van Liere's
calculations are not otherwise changed, "confusion" rates in the test and control
conditions are as follows.

4

13A. Confusion results excluding respondents who said they
thought a listing was endorsed solely because it sells Rosetta
Stone products, offers discounts/coupons/rebates, and/or is a
reputable merchant
Test

% Confused
14.

Control

Net

34%

29%

5%

Therefore, with "endorsement confusion" responses based on an

affirmative explanation by the respondent that the site was endorsed because it sells
Rosetta Stone products, offers discounts/coupons/rebates, and/or is a reputable merchant,
removed from those counted as confused, leaving intact all of Dr. Van Liere's other
counting methodologies, the "net confusion" rate drops to 5%,
Dr. Van Liere erroneously counted as confused those respondents identifying
Amazon.com and Coupon Cactus as endorsed by Rosetta Stone.

15,

Dr. Van Liere did not count as confused those who answered that they

thought the "Rosetta Stone (software)" Wikipedia article was endorsed by Rosetta Stone,
Van Liere Report ~ 40. Dr. Van Liere testified in his deposition that this exclusion was
because Rosetta Stone "contributes to the content and reviews the content of the
Wikipedia entry about them" and thus "they are endorsing the content in the sense that
they are monitoring it and contributing to it." Van Liere Deposition 84:23-85:7. The
basis for this understanding is unclear from Dr. Van Liere's report or his deposition.
16,

Assuming for the sake of this issue that Dr. Van Liere correctly excluded

from his confusion count those who thought the "Rosetta Stone (software) Wikipedia
article was endorsed by Rosetta Stone, Dr. Van Liere should have treated the respondents
who identified Amazon,com and CouponCactus as endorsed by Rosetta Stone the same
way based on Rosetta Stone's relationships with those companies,
17,

5

Amazon.com is an

reseller for Rosetta Stone-one that Rosetta Stone referred

to in an SEC filing as an "select retailer." See Rosetta Stone Inc. Form S-1 - Prospectus
Summary p. 1. Given that Rosetta Stone had official relationships with these sites, Dr.
Van Liere should have considered them "endorsed" as well. Accordingly, he should not
have counted as confused respondents who identified these two links as endorsed by
Rosetta Stone.
18.

If the Amazon and CouponCactus listings in the test condition are treated

the same as the Wikipedia software listing in the control condition - that is, if
respondents who said that these listings are endorsed by the Rosetta Stone company are
not counted as confused - then confusion rates in the test and control conditions are as
follows.

18A. Results if Amazon and Coupon Cactus "endorsement"
answers are not counted as confused
Test
% Confused
19.

Control

Net

27%

30%

-3%

Therefore, with "endorsement confusion" responses based on an

affirmative answer of the Amazon.com or CouponCactus links explanation removed from
those Dr. Van Liere counted as confused, leaving intact all of his other counting
methodologies, the "net confusion" rate drops to the point that those in the test condition
are less confused than those in the control.

Concluding remarks
20.

In my opinion, analysis of Dr. Van Liere's survey and resulting data

demonstrates that Dr. Van Liere's survey does not show a likelihood of confusion
regarding the source of advertised goods or Rosetta Stone's affiliation or "endorsement"
oflinks. It simply shows that respondents tend to believe that Rosetta Stone endorses
websites that appear to sell its products or offer related promotions such as coupons or
rebates.
6

21.

I reserve the right to supplement these opinions based on further review of

information or any additional arguments or facts that may be presented to me.
I declare under penalty of perjury that the foregoing is true and correct and that

this Declaration was executed on March 26, 2010, at Houston, Texas.

A.

7

Ph.D.