"The Apple iPod iTunes Anti-Trust Litigation"

Filing 788

ORDER DENYING: (1)DEFENDANT'S COMBINED MOTION FOR SUMMARY JUDGMENT AND DAUBERT MOTION; (2)PLAINTIFFS' DAUBERT MOTION; AND (3)PLAINTIFFS' MOTION TO STRIKE EXPERT REPORT. Signed by Judge Yvonne Gonzalez Rogers on 9/26/14. (fs, COURT STAFF) (Filed on 9/26/2014)

1
2
3
4
5
6

UNITED STATES DISTRICT COURT

7

NORTHERN DISTRICT OF CALIFORNIA

8
9
10

United States District Court
Northern District of California

11

THE APPLE IPOD ITUNES ANTITRUST
LITIGATION

Case No.: 05-CV-0037 YGR

13

This Order Relates to:

14

All Actions

ORDER DENYING: (1) DEFENDANT'S
COMBINED MOTION FOR SUMMARY
JUDGMENT AND DAUBERT MOTION; (2)
PLAINTIFFS' DAUBERT MOTION; AND (3)
PLAINTIFFS' MOTION TO STRIKE EXPERT
REPORT

12

15
16
17
18

I.

INTRODUCTION
The remaining plaintiffs in this long-running antitrust case allege that defendant Apple, Inc.

19

("Apple"), after lawfully acquiring monopoly power in the market for portable digital music players

20

with the introduction of the iPod, unlawfully maintained its monopoly power in violation of Section

21

2 of the Sherman Antitrust Act, 15 U.S.C. § 2. These plaintiffs represent a certified class of direct

22

purchasers, specifically, individuals and businesses who purchased certain enumerated models of

23

iPods directly from Apple between September 12, 2006 and March 31, 2009.

24

Now before the Court are three substantive motions, as well as numerous administrative

25

motions to seal the moving papers and supporting evidence. This Order resolves the substantive

26

motions: (1) Apple's Motion for Summary Judgment and to Exclude Expert Testimony of Roger G.

27

Noll (Dkt. No. 740-4 ("MSJ")); (2) plaintiffs' Daubert Motion to Exclude Certain Opinion

28

Testimony of Kevin M. Murphy and Robert H. Topel (Dkt. No. 737-4 ("Daubert Motion")); and (3)

1

plaintiffs' Motion to Strike Supplemental Expert Report of Kevin M. Murphy and Robert H. Topel,

2

Dated December 20, 2013 (Dkt. No. 750-3). The Court addresses the administrative motions to

3

seal in a separate Order.1
The motions are fully briefed and the Court held hearings in connection with them on

4
5

February 7, 2014 (Dkt. No. 775) and August 13, 2014 (Dkt. No. 787 ("Tr.")). For the reasons set

6

forth below, the Court DENIES all three motions.

7

II.

BACKGROUND

8

The lengthy procedural history of this case has been recounted in prior opinions.2 Here the

9

Court sets forth only those background facts necessary to understand the case's present posture and

10

United States District Court
Northern District of California

11

the motions at bar. The facts supplied herein are undisputed unless otherwise noted.
During the class period, Apple provided to iPod owners a software program for loading and

12

managing digital song files on their iPods, as well as for purchasing digital song downloads from

13

Apple. That program is "iTunes" and Apple's online music store is the "iTS."3 One feature of both

14

iTunes and iPods during the class period was their use of a digital rights management ("DRM")

15

system unique to Apple, called "FairPlay." FairPlay made certain iPods distributed during the class

16

period incapable of playing digital songs downloaded from an online music store unless they had

17

been downloaded from the iTS.4

18
1

19
20
21
22
23
24
25
26
27
28

The parties submitted all of their briefs and the bulk of their supporting evidence under
seal pursuant to this Court's Civil Local Rule 79-5, albeit in a procedurally defective manner. The
Court cites the unredacted, nonpublic version of the briefs and evidence, taking pains, however, to
place in the public record no material designated as sealable. This does not suggest that the
materials so designated are in fact sealable.
2

See generally Tucker v. Apple Computer, Inc., 493 F. Supp. 2d 1090 (N.D. Cal. 2006)
("iPod I"); In re Apple iPod iTunes Antitrust Litig., C05-00037 JW, 2008 WL 5574487 (N.D. Cal.
Dec. 22, 2008) amended, C05-00037 JW, 2009 WL 249234 (N.D. Cal. Jan. 15, 2009) ("iPod II");
In re Apple iPod iTunes Antitrust Litig., 796 F. Supp. 2d 1137 (N.D. Cal. 2011) ("iPod III").
3

Apple initially called its online digital-music retailer the iTunes Music Store but later in
the class period rechristened it simply the iTunes Store. The Court refers to both as "iTS."
4

The Court earlier granted Apple judgment on the pleadings as to a Section 1 tying claim
challenging its creation of a so-called "walled garden" through its implementation of FairPlay.
(Dkt. No. 274.)
2

1

In July 2004, an Apple competitor in the online music market, third party Real Networks

2

("Real"), introduced a new version of its own digital-song manager, RealPlayer. RealPlayer

3

included a feature called Harmony. Harmony made songs downloaded from Real's online music

4

store mimic FairPlay, and thus made music purchased from Real playable on iPods.

5

Apple responded to Harmony by taking technological countermeasures to stop Harmony

6

from mimicking FairPlay. First, in October 2004, Apple issued an iTunes update denominated

7

"4.7." The 4.7 update, among other things, thwarted Harmony's ability to mimic FairPlay. The

8

Court previously held 4.7 to be a genuine product improvement and therefore lawful, and entered

9

summary judgment in favor of Apple to the extent plaintiff's Section 2 claim rested on Apple's

10

introduction of 4.7. iPod III, 796 F. Supp. 2d at 1146.

United States District Court
Northern District of California

11

It is Apple's second instance of disabling Harmony that forms the basis of plaintiff's present

12

Section 2 claim. Following Apple's release of 4.7, Real modified Harmony such that it could again

13

mimic FairPlay and make any new songs purchased from Real's online music store playable on

14

iPods. Thereafter, in September 2006, Apple released another iTunes update that introduced a

15

variety of features while also disabling Harmony—namely, "7.0." In an earlier summary judgment

16

order, the Court found a triable issue of fact as to whether 7.0 was a genuine product improvement

17

so as to not be anticompetitive. Id. at 1147. Apple's present motion seeks summary judgment on

18

two different bases: (1) a lack of admissible evidence of antitrust impact, and (2) a lack of

19

admissible evidence as to the definition of the relevant market. To understand these arguments, it

20

is necessary to articulate plaintiff's theory of liability.

21

That theory is intricate, but ultimately it amounts to a charge that Apple's release of 7.0

22

unlawfully maintained Apple's monopoly in the market for portable digital media players by

23

making demand for iPods less elastic. Specifically, plaintiffs claim that 7.0 resulted in an increased

24

"lock-in" effect for iPod owners who purchased songs online. Lock-in, according to plaintiffs'

25

principal economics expert, "is a form of foreclosure that arises from actions that increase the cost

26

to consumers of switching to a product that has better quality and/or a lower price." (Noll Merits

27
28

3

1

Report at 4.)5 Plaintiffs offer expert opinion that Apple, by counteracting Harmony, "raised the

2

cost of switching from iPods to competing portable digital media players by eliminating the ability

3

of consumers to collect a library of downloads that could be played on all players." (Id.) That is,

4

7.0 made iPod owners unable to play songs purchased from iTS competitor Real and thus pushed

5

them to make their online song purchases only on the iTS. As a result, it discouraged iPod owners

6

from buying a competing, non-iPod digital portable music player when it came time to replace their

7

iPods due to loss, breakage, or a desire to upgrade. (Id.) Such owners would have to either forego

8

use of the songs they had purchased through Real (as well as any other online music store besides

9

iTunes, though that is not part of the damages alleged in this case), repurchase such songs through
other, iPod-compatible means (for instance, iTS or physical CDs), or convert music bought from

11

United States District Court
Northern District of California

10

Real into a non-DRM format, for example, by "burning" that music to a CD and then "ripping" the

12

CD onto their computers in a file format with no DRM, from whence the songs could then be

13

loaded on their iPods. These increased "switching costs," plaintiffs argue, locked iPod owners into

14

continuing to purchase iPods, notwithstanding the allegedly similar or better quality of and lower

15

prices of competing products. They also locked out owners of non-iPod portable digital media

16

players who had downloaded songs from the Real store. The effect of both lock-in and lock-out,

17

plaintiffs say, was to reduce competition in the market for digital portable music players and to

18

reduce the price elasticity of iPods, which permitted Apple to charge a supracompetitive price

19

therefor. (Noll Merits Report at 4-5; Noll Merits Rebuttal at 27.) According to plaintiffs' expert,

20

"[t]he damages in this case are the overcharge on iPods during the class period due to the

21

incompatibility that was created by iTunes 7.0." (Noll Merits Report at 5.) Plaintiffs' expert

22
23
24
25
26
27
28

5

The parties submitted numerous expert reports in connection with the motions at bar. In
referring to those reports, the Court adopts the nomenclature generally used by the experts
themselves. The reports are listed here in chronological order: Dkt. Nos. 751-4, Ex. 1 (Noll report
of April 3, 2013 ("Noll Merits Report")); 740-14 (Noll report of May 13, 2013 ("Noll Corrections
Report")); 737-8 (Murphy report of August 19, 2013 ("Murphy Report")); 737-9 (Topel report of
August 19, 2013 ("Topel Report")); 751-4, Ex. 2 (Noll rebuttal report of November 25, 2013 ("Noll
Merits Rebuttal")); 740-23 (joint Murphy and Topel supplemental report of December 20, 2013
("Joint Report")); 751-15, Ex. 54 (Wooldridge report of December 20, 2013 ("Wooldridge
Report")); 751-5 (Noll supplemental rebuttal report of January 13, 2014 ("Noll Supp. Rebuttal"));
and 763-5 (Wooldridge supplemental report of January 31, 2014 ("Wooldridge Supp. Report")).
4

1

estimates damages to the class of $351,631,153, "consisting of $148,947,126 for resellers,

2

$194,655,141 for direct purchasers, and $8,028,886 for additional iPod sales from the additional

3

transactions." (Noll Merits Rebuttal at 51.)

4

Apple strenuously disputes the sufficiency of plaintiffs' evidence of this theory. Apple

5

contends that it is entitled to summary judgment because plaintiffs lack admissible evidence of

6

either antitrust impact or the relevant product market, both of which are required elements of

7

plaintiffs' Section 2 claim. The linchpin, and Achilles’ heel, of Apple's argument is the word

8

"admissible." Apple disputes the admissibility of the opinions of plaintiffs' principal economics

9

expert, Professor Roger G. Noll. Noll has conducted both (i) a complex statistical analysis that
plaintiffs offer as proof of both the fact and the amount of antitrust damages suffered by the class,

11

United States District Court
Northern District of California

10

and (ii) an analysis of the relevant market. In response, Apple offers its own experts, Professors

12

Kevin M. Murphy and Robert H. Topel, who criticize the design and execution of Noll's statistical

13

analyses and fault his relevant market findings. Plaintiffs counter with rebuttal opinion from Noll,

14

as well as opinion testimony from a second expert with special expertise in statistics, Professor

15

Jeffrey M. Woodridge, whose opinions corroborate those of Noll. All of these opinions are subject

16

to Daubert motions or procedural objections. It is to those matters that the Court now turns.

17

III.

DISCUSSION

18

"A trial court may only consider admissible evidence in ruling on a motion for summary

19

judgment." Ballen v. City of Redmond, 466 F.3d 736, 745 (9th Cir. 2006). Accordingly, before

20

turning to the substance of Apple's summary judgment motion, the Court first resolves the parties'

21

challenges to the admissibility of the proffered expert opinions.
CHALLENGES TO ADMISSIBILITY OF EXPERT OPINIONS

22

A.

23

The principal focus of Apple's Daubert motion is a set of opinions offered by Noll as to

24

both the fact and amount of antitrust damages suffered by the class. These opinions have as their

25

bases econometric analyses Noll performed on a dataset supplied by Apple. The dataset consists of

26

Apple's complete sales records for the models of iPod covered by the class definition and sold

27

during the class period, stripped of obvious outliers (e.g., sales where the price was zero or

28

5

1

negative, or many times the listed retail price) and incomplete records.6 Noll used this dataset to

2

perform a hedonic multiple-regression analysis. Multiple-regression analysis is a statistical tool

3

that "permits the comparison between an outcome (called the dependent variable) and one or more

4

factors (called independent variables) that may be related to that outcome." Manpower, Inc. v. Ins.

5

Co. of Pennsylvania, 732 F.3d 796, 808 (7th Cir. 2013). The term "hedonic" denotes that the

6

analysis sought to "isolate the effect of one or more product attributes on the price of a product." In

7

re ConAgra Foods, Inc., --- F. Supp. 2d ---, 2014 WL 4104405, at *4 (C.D. Cal. 2014). In short,

8

the attributes of particular models of iPod constituted some (not all) of the independent variables in

9

Noll's regression analysis. Noll's regressions purport to isolate the effect on iPod pricing
attributable to Apple's release of 7.0 and the security feature embedded therein. According to

11

United States District Court
Northern District of California

10

plaintiffs and Noll, that pricing effect is the illegal overcharge in this case, and constitutes proof of

12

both the fact of damages and their amount. (See, e.g., Noll Merits Report at 5.) Noll also offers an

13

opinion as to the relevant product markets, identifying two: a market for portable digital media

14

players and a market for digital audio files.

15

In response to Noll's opinions, Apple submits reports from Murphy and Topel, who purport

16

to identify flaws in Noll's damages analysis that render it so unreliable as to be inadmissible under

17

the familiar standard of Daubert v. Merrell Dow Pharm., Inc., 509 U.S. 579 (1993). Their three

18

main criticisms are summarized as follows: Noll (i) failed to account for "clustering" problems in

19

his regression analysis, (ii) omitted from his hedonic model certain variables that measure product

20

attributes affecting iPod prices, and (iii) used the wrong "but-for" world to calculate damages by

21

turning off the variable for 4.7 on the date that 7.0 was introduced.7 As to Noll's market definition,

22
23
24
25
26
27
28

6

The class period in this case covers the time period beginning with Apple's initial release
of 7.0 and ending the date that Apple—following a shift in business strategy by the record
companies—entirely stopped using DRM. See iPod IV, 2011 WL 5864036, at *4 nn. 22-23. The
class is comprised of individuals and businesses who, during the class period, purchased directly
from Apple any of 29 individual iPod models distributed among 4 model types. Id. at *4-5.
7

It is common in antitrust cases to estimate damages by comparing the price actually
charged to an expert economist's estimation of the price that would have been charged "but for" the
asserted anticompetitive conduct. E.g., In re Hydrogen Peroxide Antitrust Litig., 552 F.3d 305, 313
(3d Cir. 2008); In re Elec. Books Antitrust Litig., 11 MD 2293 DLC, 2014 WL 1282298, at *3
(S.D.N.Y. Mar. 28, 2014).
6

1

Apple's expert Murphy opines that because Noll failed to use the proper mode of analysis in

2

deriving his market definition, his opinions are inadmissible as merely "untested, subjective

3

opinions." (MSJ at 24.) In response to these criticisms, plaintiffs offer supplemental rebuttal

4

opinion from Noll on damages and market definition and, with respect to the clustering criticism

5

only, the opinions of Wooldridge.

6

The parties have not challenged any of these experts as unqualified to give their respective

7

opinions.8 Rather, the parties have objected to the opinions themselves. As set forth herein, the

8

Court finds that the objections to the content of the challenged opinions go to weight rather than

9

admissibility, and that the parties' procedural and other technical objections are insufficient to

10

persuade the Court to exercise its discretion to exclude the opinions.
1.

United States District Court
Northern District of California

11
12

Applicable Legal Standard

"In federal courts, the admission of expert testimony is governed by Federal Rule of

13

Evidence 702, as elucidated by the Supreme Court in Daubert." Barabin v. AstenJohnson, Inc.,

14

700 F.3d 428, 432 (9th Cir. 2012). Federal Rule of Evidence 702 allows expert testimony only if

15

the expert's "scientific, technical, or other specialized knowledge will help the trier of fact to

16

understand the evidence or to determine a fact in issue." Rule 702 permits experts to testify if their

17

testimony is: (i) based upon sufficient facts or data, (ii) the product of reliable principles and

18

methods, and (iii) the result of applying those principles and methods reliably to the facts of the

19

case. Id. In determining whether an expert's testimony meets the standards of Rule 702, the court

20

acts as a "gatekeeper" that "ensur[es] that [the] expert's testimony both rests on a reliable

21

foundation and is relevant to the task at hand." Daubert, 509 U.S. at 597; see also Kumho Tire Co.

22

v. Carmichael, 526 U.S. 137, 148-49 (1999). In addition, the court may exclude expert testimony

23

that is "otherwise admissible may be excluded under Rule 403 if its probative value is substantially

24
25
26
27
28

8

The Court finds all four professors qualified to render expert opinions on economics and
econometrics. All four hold distinguished positions as economics professors—Noll at Stanford
University, Murphy and Topel at the University of Chicago, and Wooldridge at Michigan State
University. All four have also written and taught extensively concerning the subjects upon which
they have opined. See Noll Merits Report, Appendix A; Murphy Report, Appendix A; Topel
Report, Appendix A; Wooldridge Report, Appendix A.
7

1

outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury." U.S.

2

v. Rincon, 28 F.3d 921, 923 (9th Cir. 1994) (citing Daubert, 509 U.S. at 595).

3

The Ninth Circuit has reiterated that the "test of reliability is ‘flexible’ . . . . When an expert

4

meets the threshold established by Rule 702 as explained in Daubert, the expert may testify and the

5

jury decides how much weight to give that testimony." Primiano v. Cook, 598 F.3d 558, 564-65

6

(9th Cir. 2010). "Shaky but admissible evidence is to be attacked by cross examination, contrary

7

evidence, and attention to the burden of proof, not exclusion." Id. at 564.

8

2.

Apple's Daubert Motion to Exclude Opinions of Noll

The Court first addresses Apple's challenges to Noll's opinions, specifically: (i) his opinions

10

regarding antitrust impact and damages, as based on his regression model, and (ii) his opinion as to

11

United States District Court
Northern District of California

9

which products comprise the relevant market in which iPods reside.

12
13

a.

Regression Model

Proof of an injury caused by Apple's alleged antitrust conduct is a required element of

14

plaintiffs' Section 2 claim. Rebel Oil Co. v. Atl. Richfield Co., 51 F.3d 1421, 1432-33 (9th Cir.

15

1995). In this case, plaintiffs seek to demonstrate the required antitrust injury with Noll's

16

regression model. The model purports to isolate the effect on iPod pricing during the class period

17

attributable to the alleged anticompetitive conduct, that is, the effect of Apple's introduction of 7.0

18

and the consequent inability of iPods loaded with 7.0 to play songs downloaded from Real. As a

19

threshold matter, the Court finds that the dataset relied upon by Noll, as the complete record of

20

sales transactions for covered models of iPods during the class period, constitutes sufficient data

21

upon which to base expert testimony. The Court also finds that hedonic multiple-regression

22

analysis is a sound and, indeed, commonplace method for isolating the pricing effects of alleged

23

anticompetitive conduct. E.g., In re High Fructose Corn Syrup Antitrust Litig., 295 F.3d 651, 660-

24

61 (7th Cir. 2002) ("Corn Syrup"); Conwood Co., L.P. v. U.S. Tobacco Co., 290 F.3d 768, 793 (6th

25

Cir. 2002); Petruzzi's IGA Supermarkets, Inc. v. Darling-Delaware Co., 998 F.2d 1224, 1238 (3d

26

Cir. 1993). Accordingly, Noll's opinions are admissible under Rule 702 unless they are the result

27

of an unreliable application of otherwise sound methods or would be unhelpful to the trier of fact.

28

8

1

Apple contends that Noll's regression model is inadmissible because (i) it lacks a sufficient

2

"fit" with the facts of the case, (ii) it does not account for all the relevant factors that affect iPod

3

pricing, and (iii) it does not supply statistically significant results (once one modifies the model in

4

the manner suggested by Apple's experts). The Court has weighed Apple's objections carefully and

5

finds that none establish such a level of unreliability or unhelpfulness that would justify wholesale

6

exclusion of Noll's opinions. Rather, they go to the weight of Noll's opinions.

7

The Court turns first to Apple's arguments as to fit. As a threshold matter, the Court notes

8

that the challenged opinions rest on a regression analysis conducted on data consisting of Apple's

9

actual sales records for iPods during the class period. Given that Noll's regressions, and his
opinions based thereon, purport to explain the prices charged in those sales, there is a sufficient fit

11

United States District Court
Northern District of California

10

between the facts and the opinion.

12

Apple offers four arguments as to fit, none of which persuade. These arguments posit that

13

Noll's regression "rests on unsupported assumptions that conflict with the real world." (MSJ at 14.)

14

First, Apple contends that the requisite fit is lacking because plaintiffs have marshaled no evidence

15

of particular persons who used Harmony or suffered lock-in or lock-out. The lack of direct

16

evidence of named individuals who used Harmony does not disprove their existence. The Court

17

finds nothing unreasonable about an assumption that, among the millions of persons who used

18

iPods during the class period, some may have purchased songs from Real. Under plaintiffs' theory,

19

such purchases give rise to the lock-in effect and reduced elasticity of demand. No "indisputable

20

record facts contradict or otherwise render the opinion unreasonable," Brooke Grp. Ltd. v. Brown &

21

Williamson Tobacco Corp., 509 U.S. 209, 242 (1993), nor are Noll's opinions "connected to

22

existing data only by the ipse dixit of the expert," Kumho Tire Co. v. Carmichael, 526 U.S. 137,

23

157 (1999). Rather, Noll's opinions derive from a dataset of transactions supplied by Apple itself.9

24
25
26
27
28

9

Apple analogizes this case to American Booksellers Association, but that case is
distinguishable. Am. Booksellers Ass'n, Inc. v. Barnes & Noble, Inc., 135 F. Supp. 2d 1031 (N.D.
Cal. 2001). In that case, plaintiffs' expert on antitrust damages "concededly made no effort to base
his model on actual purchasing data" and "his model ma[de] no attempt to determine the actual
prices paid . . . ." Id. at 1038. Not so here, where it is uncontested that Noll based his model on
actual purchasing data and the actual prices paid.
9

1

Second, Apple argues that plaintiffs present no direct evidence of Apple's pricing committee

2

taking into account any effects of the Harmony-disabling countermeasures contained in 7.0 when

3

setting iPod prices. (MSJ at 13.) The Court is not persuaded that this is a valid reason to exclude

4

Noll's testimony. The record contains testimony and documentary evidence sufficient to support an

5

inference that the Apple executives on the pricing committee knew of and were concerned by

6

Harmony. (Dkt. No. 751-13, Ex. 36 (submitted under seal); Dkt. No. 751-15, Ex. 48 at 45:9-19

7

(submitted under seal).)

8

Third, the Court rejects Apple's related argument that Noll's opinions must be excluded

9

because his prediction of a constant, immediate lock-in effect resulting in overcharge "is flatly
contrary to how Apple set its prices." (MSJ at 13; see also Dkt. No. 762-5 ("MSJ Reply") at 6, 7-

11

United States District Court
Northern District of California

10

8.) Apple explains that its prices are set at particular times pursuant to a uniform pricing policy.

12

However, the record contains non-trivial evidence that the actual prices charged were not in fact

13

uniform and that pricing decisions may have incorporated factors above and beyond Apple's

14

preference for so-called "aesthetic" prices. Apple's argument does not sufficiently undermine the

15

reliability of Noll's model to warrant exclusion.

16

Finally, the Court rejects Apple's argument that the analysis predicts a constant, immediate

17

overcharge that Apple claims is not consistent with the notion of a gradual lock-in over time.

18

Apple purports to demonstrate that Noll's own admissions "are irreconcilable with the single,

19

unchanging overcharge amount predicted by his damages model." (MSJ Reply at 7-8.) That

20

argument ultimately is one of weight, not evidence of the unreliability of the regression analyses

21

themselves.

22

Apple's second proffered basis for excluding Noll's opinions centers on his supposed failure

23

to account properly for relevant pricing factors. (MSJ at 14-20.) Generally, such deficiencies in a

24

statistical analysis raise issues of weight rather than admissibility. E.g., Bazemore v. Friday, 478

25

U.S. 385, 400 (1986) (reversing lower court's exclusion of regression analysis based on its view

26

that the analysis did not include proper selection of variables); Hemmings v. Tidyman's Inc., 285

27

F.3d 1174, 1188 (9th Cir. 2002) ("[I]n most cases, objections to the inadequacies of a study are

28

more appropriately considered an objection going to the weight of the evidence rather than its

10

1

admissibility."); Obrey v. Johnson, 400 F.3d 691, 695 (9th Cir. 2005) (same principle); Rudebusch

2

v. Hughes, 313 F.3d 506, 516 (9th Cir. 2002) (same); Maitland v. Univ. of Minnesota, 155 F.3d

3

1013, 1017 (8th Cir. 1998) (same); see also Manpower, 732 F.3d at 808 (collecting cases). Here,

4

Apple faults Noll for leaving "on" throughout the entire class period a variable representing Apple's

5

introduction of 4.7, for purportedly failing to account for the impact on price of aspects of 7.0

6

besides the allegedly anticompetitive security feature that disabled Harmony, and for failing to

7

account for certain, though by no means all, of the product characteristics of iPods used in Noll's

8

hedonic model. (MSJ at 14-20.) These criticisms do not persuade that Noll's regression analysis so

9

fundamentally unreliable as to warrant exclusion. Noll supplied cogent reasons for his inclusions
and exclusions. Apple's criticisms reflect mere disagreement with those reasons. As such, they go

11

United States District Court
Northern District of California

10

to the weight that should be afforded Noll's opinions, not their admissibility.

12

Apple's third main reason for excluding Noll's opinions is that his regression, once

13

"corrected" in the manner urged by Apple's experts, does not supply statistically significant results.

14

(See MSJ at 20 (arguing Noll's results are statistically insignificant "[w]hen properly calculated").)

15

The Court notes that Apple's argument relies on the Court's acceptance of Apple's experts'

16

criticisms of Noll's methodology. While some of those criticisms are compelling, the Court is not

17

persuaded, in light of Noll's rebuttal opinions and Wooldridge's opinions, that the battle between

18

the economists of the University of Chicago school, on the one hand, and those from Stanford and

19

Michigan State University, on the other, is properly resolved here. It is not lost on the Court that

20

econometrics is in the family of social sciences and therefore necessarily contains certain value

21

judgments and hypotheses that are tested and used in conjunction with statistics. Unless beyond the

22

realms of reason and reliability, these issues are fully within the province of experts to debate and a

23

jury to resolve.

24

Most significant are Apple's criticisms regarding clustering. Murphy and Topel argue that

25

Noll vastly overstates the precision and reliability of his model by failing to account for clustering

26

and thus generates an artificially high "t-statistic." By way of background, a t-statistic is one

27

measure of confidence that a statistical result is statistically significant, that is, not the byproduct of

28

chance. A t-statistic of 1.96 translates to a 95 percent confidence level that the result is statistically

11

1

significant, REFERENCE MANUAL ON SCIENTIFIC EVIDENCE 303, 343 (3d ed. 2011), or, inversely, a 5

2

percent chance that the figure is statistically insignificant. Using the inverse measure, confidence

3

levels of 1, 5, or 10 percent have all been treated as benchmark measures of reliability by statistics

4

experts. See In re High-Tech Employee Antitrust Litig., 11-CV-02509-LHK, 2014 WL 1351040, at

5

*15 (N.D. Cal. Apr. 4, 2014) ("High-Tech") (collecting authorities). Thus far though, courts have

6

been unwilling to mandate a particular t-statistic as a prerequisite for purposes of admissibility. Id.

7

(finding no cases legally requiring 10 percent or greater confidence to admit statistical evidence)

8

(citing Cook v. Rockwell Int'l Corp., 580 F. Supp. 2d 1071, 1102 (D. Colo. 2006); Kadas v. MCI

9

Systemhouse Corp., 255 F.3d 359, 362 (7th Cir. 2001)).

10

The Court is not persuaded that Apple's criticisms of Noll's approach to clustering merit

United States District Court
Northern District of California

11

exclusion of Noll's opinions. While Murphy and Topel's modifications to Noll's regression to

12

account for clustering "properly" would, if accepted, dramatically change the t-statistics for the

13

iTunes 7.0 coefficient, Noll and Wooldridge make at least a colorable case that no clustering

14

problems exist where, as here, the data set constitutes the entire population of observations, as

15

opposed to a sampling. (See Noll Merits Rebuttal at 39 ("[C]luster analysis is irrelevant if the data

16

set is either representative of the entire population of observations or is not a sample at all, but in

17

fact is the entire population."); Noll Supp. Rebuttal at 3-7 (same principle); Wooldridge Report at

18

10 (same).) The question presented here is not whether Noll's analyses are correct, but whether

19

they are the product of a generally accepted method for demonstrating both the fact and the amount

20

of antitrust damages. In light of the opinions of Noll and Wooldridge, the Court finds that the

21

results of Noll's regression analyses do meet the threshold of reliability necessary for admissibility,

22

even if the proffered claim of accuracy strains credulity. Obrey, 400 F.3d at 696 ("As a general

23

matter, so long as the evidence is relevant and the methods employed are sound, neither the

24

usefulness nor the strength of statistical proof determines admissibility under Rule 702."). Given

25

the general acceptance of multiple-regression analysis, Noll's obvious qualifications for conducting

26

such analyses, and the unresolved threshold question of whether Noll's figures should be

27

"corrected" in the precise manner urged by Murphy and Topel, the Court finds Noll's multiple-

28

regression analyses admissible under Federal Rule of Evidence 702 and Daubert. Under these

12

1

circumstances, the issue is more appropriately one of weight and credibility. Accordingly, the

2

Court DENIES Apple's Daubert motion insofar as it challenges the admissibility of Noll's regression

3

analyses.

4
5

b.

Relevant Market

Plaintiffs' Section 2 claim requires them to prove the contours of the "relevant market" in
which Apple allegedly used unlawful means to maintain its monopoly power. See Eastman Kodak

7

Co. v. Image Technical Servs., Inc., 504 U.S. 451, 481 (1992); Bhan v. NME Hospitals, Inc., 929

8

F.2d 1404, 1413 (9th Cir. 1991); Newcal Indus., Inc. v. Ikon Office Solution, 513 F.3d 1038, 1044

9

(9th Cir. 2008). "[A] 'market' is the group of sellers or producers who have the actual or potential

10

ability to deprive each other of significant levels of business." Rebel Oil Co. v. Atl. Richfield Co.,

11

United States District Court
Northern District of California

6

51 F.3d 1421, 1434 (9th Cir. 1995) (internal quotation marks omitted). "The outer boundaries of a

12

product market are determined by the reasonable interchangeability of use or the cross-elasticity of

13

demand between the product itself and substitutes for it." Newcal Indus., 513 F.3d at 1045 (quoting

14

Brown Shoe Co. v. United States, 370 U.S. 294, 325 (1962)).

15

To prove their relevant market, plaintiffs rely exclusively on the opinions of Noll, who

16

undertook a lengthy analysis of both relevant markets asserted by plaintiffs in this case,

17

respectively, portable digital media players and digital audio files. (Noll Merits Report at 25-32

18

(portable digital media players), 32-42 (digital audio files).) Noll opined that there are multiple

19

ways of identifying a relevant market. One such method is to estimate cross-elasticity of demand

20

using a formal econometric study. However, Noll further opined that "[i]n most cases data

21

limitations preclude econometric estimation of cross-elasticity of demand" and that such estimation

22

"is usually impossible for products that have extensive product differentiation and that are rapidly

23

evolving, as was the case of portable digital media players during the class period." (Id. at 23-24.)

24

Noll looked instead to internal Apple documents, employee testimony, and discovery responses,

25

third-party information such as contemporaneous financial analysis and press coverage of the

26

development of the portable digital media player market, and his own experience.

27

Apple argues that Noll's opinion on these matters is inadmissible because he failed to

28

conduct either a formal econometric analysis of cross-elasticity of demand or a "hypothetical

13

1

monopolist" test. While such analyses may be possible or even desirable, Apple’s assertion that

2

such formal tests are required lacks legal support. Moreover, as plaintiffs point out, Apple's own

3

expert did not measure the cross-elasticity of demand or engage in the hypothetical monopolist test

4

in their own discussions of market definition. (See generally Murphy Report at 60-64.) Apple's

5

attacks on Noll's opinion comprise mere disagreement with his conclusions.

6

For purposes of Rule 702, the district judge serves only as "a gatekeeper, not a fact finder."

7

Primiano, 598 F.3d at 564-65 (internal quotation marks omitted). Apple has failed to persuade that

8

Noll's opinion as to the relevant market must be excluded for lack of an accepted methodology.

9

Accordingly, Apple's Daubert motion is DENIED on this ground.

10

3.

Plaintiffs' Daubert Motion to Exclude Opinions of Murphy and Topel

United States District Court
Northern District of California

11

Plaintiffs seek to exclude only those opinions of Murphy and Topel bearing on clustering

12

problems. (Daubert Motion at 1.) As was the case with Murphy and Topel's criticisms of Noll's

13

regression analysis, plaintiffs' criticisms of Murphy and Topel's clustering opinions also go to

14

weight, not admissibility. Plaintiffs' motion is premised on a mere disagreement in expert opinion,

15

not a showing that Murphy and Topel used unreliable methods, relied upon facts too far removed

16

from those of this case to be relevant, or are unqualified to render their opinions.

17
18
19
20
21

The Court DENIES plaintiffs' Daubert motion to exclude the opinions of Murphy and Topel.
4.

Apple's Objection to Opinions of Wooldridge

Apple argues in its opposition that Wooldridge's opinions must be excluded as having been
untimely disclosed and inadmissible under Daubert. The Court disagrees on both counts.
First, as to timeliness, even assuming plaintiffs disclosed Wooldridge late, such late

22

disclosure would be harmless under Federal Rule of Civil Procedure 37(c) because Apple had

23

ample time not only to respond to Wooldridge's report (prompting, in turn, a supplemental report

24

by Wooldridge), but also to depose him (which deposition Apple took, see Dkt. No. 754-10, Ex. 11

25

("Wooldridge Dep.")).

26

Second, as to Apple's Daubert objection to Wooldridge's opinions, Apple argues that

27

Wooldridge's opinions must be excluded because they "are contrary to generally accepted

28

econometrics," "have not been peer reviewed," and "were manufactured for this litigation." (Dkt.

14

No. 754-6 at 18-19.) The Court finds none of those points persuasive. The purported lack of peer

2

review or common acceptance of Wooldridge's views does not necessarily justify exclusion of

3

Wooldridge's opinions, given Wooldridge's own apparent status as a leading authority and his

4

specific testimony as to the development of his theories. See Daubert, 509 U.S. at 593-94 (peer

5

review and publication of an expert's views are relevant to, but not necessarily dispositive of,

6

reliability determination). "[I]n some instances well-grounded but innovative theories will not have

7

been published." Id. at 593. Here, Wooldridge testified, and Apple has not disputed, that

8

Woodridge himself is a leading authority on clustering theory, that he has been revisiting and

9

refining his theories for at least two years, sometimes in collaboration with colleagues at Harvard

10

and Stanford, and that he has of late modified his views such that he now disagrees with certainly

11

United States District Court
Northern District of California

1

commonly accepted views, including views encompassed in recommendations contained in an

12

American Bar Association guide. (Wooldridge Dep. at 62:14-65-2, 91:7-92:10, 117:6-118:24,

13

158:16-159:13.)10 Nothing in these statements or elsewhere in the record before the Court

14

establishes that Wooldridge's theories are unreliable, as opposed to merely new.

15

The same testimony undercuts Apple's contention that Wooldridge's rethinking of his

16

theories predated his engagement as an expert witness in this case. (See Wooldridge Dep. at 8:10-

17

9:2 (Wooldridge first contacted about and began working on case in December 2013); see also id.

18

at 7:21-24 (Wooldridge never before served as expert witness).) Apple's argument that

19

Wooldridge generated his opinions for the purposes of litigation amounts to an invitation to

20

discredit Wooldridge's contrary testimony. The Court is unwilling to do so on the cold record now

21

before it. Any concerns with Wooldridge's opinions are best addressed through contrary evidence

22

and cross-examination and, if applicable, impeachment of Wooldridge. Should the Court determine

23

at trial, following examination of Wooldridge or for any other reason, that his opinions are

24

unreliable or ersatz, the Court has the option of giving the jury a limiting instruction. Cf.

25

Hemmings, 285 F.3d at 1183 (affirming district court's denial of Daubert motion where district

26
27
28

10

It is not clear that the matter on which Wooldridge's views have changed is the same
matter at issue here. (See Tr. at 95:5-12.)
15

1

court stated that "if it determined that portions of [the expert's] analysis were improper, it would

2

give a limiting instruction to the jury").

3

The Court OVERRULES Apple's objections to the opinions of Wooldridge.
5.

4
5

Plaintiffs' Motion to Strike

Plaintiffs move to strike the Joint Report, that is, the supplemental report submitted jointly

6

by Murphy and Topel in response to the Noll Merits Rebuttal. Plaintiffs contend that the Joint

7

Report is untimely and, because the Noll Merits Rebuttal contained no new opinion, unjustified.

8

Plaintiffs claim they will be prejudiced if the Joint Report is admitted and ask that, if it is, the Court

9

also admit Noll's response to the Joint Report, the Noll Supplemental Rebuttal. Apple raises
several arguments in opposition to plaintiffs' motion to strike, but essentially acquiesces in the entry

11

United States District Court
Northern District of California

10

of plaintiffs' requested alternative relief, i.e., allowing both the Joint Report and the Noll

12

Supplemental Rebuttal. (Dkt. No. 758-5 at 1.)11

13

The Court finds that permitting the Joint Report to stand, along with the Noll Supplemental

14

Rebuttal, is harmless, given that Noll has responded to the Joint Report and plaintiffs deposed both

15

Murphy and Topel after they issued the Joint Report. Additionally, in light of the new regression

16

analysis contained in the Noll Merits Rebuttal, the Court cannot say that the Joint Report is

17

unjustified. Noll modified several attributes of his regression analysis following the Murphy

18

Report and Topel Report. The regression analyses presented in the Noll Merits Rebuttal cover the

19

same subject matter as those presented in the earlier Noll Merits Report, but they are not the same

20

analyses. Plaintiffs' argument that there was nothing new in the Noll Merits Rebuttal to which

21
22
23
24
25
26
27
28

11

Apple argues that plaintiffs' motion to strike should be denied on procedural grounds as
an evidentiary objection not contained in the body of plaintiffs' summary judgment opposition,
which objections are prohibited by Civil Local Rule 7-3(a). (Dkt. No. 758-5 at 1.) Plaintiffs
respond that the applicable rule is Local Rule 7-8 because plaintiffs move not for exclusion under
Daubert but for evidentiary sanctions under Federal Rule of Civil Procedure 37, and Local Rule 7-8
requires "[a]ny motion for sanctions, regardless of the sources of authority invoked," to be
separately filed. The Court need not address this potential ambiguity in its Local Rules because
Apple's acquiescence to plaintiffs' requested alternative relief moots the issue. The Court notes,
however, that Apple's citation to Apple, Inc. v. Samsung Electronics Co., 11-CV-01846-LHK, 2011
WL 7036077, at *3 (N.D. Cal. Dec. 2, 2011), is inapposite because, in that case, Samsung moved
for Daubert exclusion, not, as here, Rule 37 sanctions.
16

1

Murphy and Topel could respond does not persuade in light of Noll's changes to the model of his

2

regression analyses.

3
4

The Court DENIES plaintiffs' motion to strike the Joint Report. Both the Joint Report and
the Noll Supplemental Rebuttal are allowed.

5

B.

APPLE'S MOTION FOR SUMMARY JUDGMENT

6

Apple contends that: (1) Real's "insignificant" share of less than 3 percent of the online

7

music market in 2006, when Apple released 7.0, makes it "implausible" that Harmony could have

8

the effect ascribed to it by plaintiffs (MSJ at 9); (2) "plaintiffs have no evidence regarding what

9

portion of Real's small sales was to iPod owners or potential iPod purchasers" (id. at 9-10); (3)
"plaintiffs have no proof of the number of people who became locked in or locked out" after 7.0

11

United States District Court
Northern District of California

10

(id. at 10); (4) plaintiffs have not identified evidence showing that Apple's pricing committee "took

12

into account the amount of sales from [Real] or any other online store in setting iPod prices" (id. at

13

10-11); and (5) Apple always abides by an "aesthetic" pricing policy, by which Apple appears to

14

mean a policy of setting prices in fifty-dollar increments, less one dollar (e.g., $199, $249, $399)

15

(see id. at 11-12).

16

The Court rejects these arguments because the admission of Noll's opinions alone supplies a

17

triable issue of fact regarding the fact and amount of antitrust damages, as well as the definition of

18

the relevant market. Apple's five grounds for entering summary judgment are insufficient bases

19

upon which to enter summary judgment in the face of Noll's opinions. Those opinions constitute

20

relevant circumstantial evidence of both the fact and amount of damages upon which a jury

21

applying a preponderance standard reasonably could find for plaintiffs. Noll's opinions also supply

22

non-trivial evidence of the relevant market. Apple's asserted bases for summary judgment merely

23

point out possible gaps in plaintiffs' case.

24

Given that the Court may not intrude upon the province of the jury by weighing the

25

conflicting evidence now before it, Noll's opinions preclude entry of summary judgment. Cf. Corn

26

Syrup, 295 F.3d at 660-61 (after surveying each blow in a "battle of the statistical experts",

27

reversing district court's entry of summary judgment in favor of defendant because plaintiffs had

28

17

1

presented statistical evidence of price-fixing). The Court DENIES Apple's motion for summary

2

judgment.

3

IV.

CONCLUSION

4

For the foregoing reasons, the Court DENIES Apple's Daubert motion to exclude the

5

opinions of plaintiffs' expert Noll, DENIES plaintiffs' Daubert motion to exclude certain opinions of

6

Apple's experts Murphy and Topel, OVERRULES Apple's objections to the opinions of plaintiffs'

7

expert Wooldridge, DENIES plaintiffs' motion to strike the December 20, 2013 joint report of

8

Murphy and Topel, and DENIES Apple's motion for summary judgment.

9

IT IS SO ORDERED.

10

United States District Court
Northern District of California

11
12

Date: September 26, 2014

_______________________________________
YVONNE GONZALEZ ROGERS
UNITED STATES DISTRICT COURT JUDGE

13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28

18