CRA Holdings US, Inc. and Subsidiaries v. United States Government

Filing 46

DECISION AND ORDER granting in part and denying in part 29 Defendant's Motion to Compel; Plaintiffs' motion 33 is denied. Signed by Hon. Leslie G. Foschio on 1/26/2017. (SDW)

UNITED STATES DISTRICT COURT
WESTERN DISTRICT OF NEW YORK
________________________________________
CRA HOLDINGS US, INC. AND SUBSIDIARIES,
Plaintiffs,

DECISION
and
ORDER

v.
15-CV-239W(F)
UNITED STATES OF AMERICA,
Defendant.
________________________________________
APPEARANCES:

ZERBE FINGERET FRANK & JADAV
Attorneys for Plaintiffs
JEFFERSON H. READ,
ROBERT G. WONISH, II, of Counsel
3009 Post Oak Boulevard, Suite 1700
Houston, Texas 77056
CAROLINE D. CIRAOLO
Principal Deputy Assistant Attorney General
Tax Division, U.S. Department of Justice
Attorney for Defendant
TERESA MARIE ABNEY
JAMES M. STRENDJORD, of Counsel
PO Box 55
Ben Franklin Station
Washington, DC 20044

In this tax refund action, Plaintiffs claimed a refund of $419,924 for Plaintiffs’
2002 tax year and $1,029,402 for Plaintiffs’ 2003 tax year (“the tax years”) based on
Plaintiffs’ assertion of a research and development (“R&D”) tax credit, under 26 U.S.C.
§ 41, allegedly available to Plaintiffs for 6,100 remediation projects in which Plaintiffs
engaged during the tax years. After filing income tax returns for the tax years, Plaintiffs,
a national environmental engineering firm specializing in assisting clients in remediation
projects involving environmental contamination, retained a consulting firm, Axiom
1

Custom Business Solutions (“Axiom”) to review its R&D activities and to produce a
report that would allow Plaintiffs to file amended returns for tax years in which Plaintiffs
would request a credit based on qualified research expenses (“”QRE” or “QREs”) for the
tax years, as defined by 28 U.S.C. § 41(a)(1)(A) and (b)(2)(B), resulting in a claim for a
refund of taxes paid by Plaintiffs for the tax years. Based on the Axiom study of
Plaintiffs’ R&D activities, Plaintiffs filed amended returns for the tax years requesting a
refund. Following an audit, the Internal Revenue Service denied Plaintiffs’ refund
request.
Included in such QREs are labor costs for conducting “qualified research” which
must (1) “represent research and development costs in the experimental or laboratory
sense,” (2) “for all purposes of discovering information which is technological in nature,”
i.e., information that relies “on principles of the physical or biological sciences,
engineering, or computer science,” 26 C.F.R. § 1.41-4(a)(4), (3) the information
discovered must be “useful in the development of a new or improved business
component,” i.e., a product, process, software, technique, formula, or invention the
taxpayer intends to sell, lease, license, or use in its trade or business, 26 U.S.C. §
41(d)(2)(B), which is “new of different” from the taxpayer’s existing products, etc., see
Trinity Indus., Inc. v. United States, 691 F.Supp.2d 688, 696 (N.D. Tex. 2010), aff’d, 757
F.3d 400, 412 (5th Cir. 2014), and (4) substantially all of the research activities must
involve elements of “a process of experimentation,” 26 U.S. §§ 41(d)(1) & (3)(A), i.e.,
“research and development costs in the experimental or laboratory sense” to “discover
information that would eliminate uncertainty concerning the development or
improvement of a product.” 26 C.F.R. § 1.174-2(a)(1). To qualify as a QRE, the related

2

research must precede commercial production, may not constitute an adaptation of an
existing business component, may not result from research funded by other than the
taxpayer, and must take place within the United States. 26 U.S.C. § 41(d)(4)(A), (B),
(C), (F), (H). Taxpayers asserting an R&D tax credit are also required to retain the
records necessary to show an entitlement to the credit. 26 C.F.R. § 1.41-2(d). The
court’s determination of the merits of a taxpayer’s claim is de novo based on the facts
presented to the court. See Trinity Indus., 757 F.3d at 415. Tax credits are narrowly
construed, Shami v. Comm’r, 741 F.3d 560, 567 (5th Cir. 2014), and a tax payer has the
burden to demonstrate compliance with each of the statutory requirements. United
Stationers, Inc. v. United States, 163 F.3d 440, 443 (7th Cir. 1998); Research, Inc. v.
United States, 1995 WL 560140 at *3 (D.Minn. June 21, 1995). See also Lenihan v.
Comm’r, 296 Fed.App’x. 160, 162 (2d Cir. Oct. 17, 2008) (disallowing deductions and
capital loss carry over for which taxpayer produced no supporting evidence).
In this case, Plaintiffs initially claimed an R&D tax credit for 6,100 R&D “projects”
listed by Plaintiffs, Dkt. 29, Exh. 8, however, Plaintiffs later produced a revised list of its
R&D projects at issue (“Plaintiffs’ projects”) which included 4,643 projects. Dkt. 35 at 5
n. 1. Following filing of the Case Management Order on May 5, 2016 (Dkt. 20), directed
to whether sampling could be used to narrow the number of projects for purposes of
litigating this case, and the parties’ subsequent disagreement as to the feasibility of
conducting discovery and trial based on a sample of Plaintiffs’ projects, Defendant
served, on July 6, 2016, Defendant’s Second Set of Interrogatories particularly limited to
Superseding Interrogatory No. 1, Dkt. 29, Exh. 5 (“Defendant’s Superseding
Interrogatory”), which requested Plaintiffs to respond to 13 questions seeking factual

3

information including the eight required elements for a QRE under § 41, upon which
Plaintiffs’ R&D tax credit claims are based, specifically, Plaintiffs’ asserted QREs. In
Plaintiffs’ Amended Objections and Responses To Second Interrogatories (Dkt. 29, Exh.
10) served September 12, 2016, Plaintiffs objected to Defendant’s Interrogatories 1(a)(c) based on burdensomeness, lack of relevancy to the need to select a sample of the
projects for purposes of discovery and trial, and referred Defendant to the list of
Plaintiffs’ projects as listed in the Axiom report previously produced to Defendant. As to
Interrogatory No. 1(d), Plaintiffs stated all the projects’ research activity was performed
in the United States.
By papers filed September 19, 2016, Defendant moved to compel Plaintiffs’
responses to Defendant’s Superseding Interrogatory No. 1(Dkt. 29) (“Defendant’s
Motion”) together with the Declaration of Teresa M. Abney, Trial Attorney, United States
Department of Justice (“Defendant’s Motion Exh. A”), Defendant’s Brief in Support of
Defendant’s Motion (Exh. A) and Exhibits 1-13 In Support of Defendant’s Motion
(“Defendant’s Motion Exh(s). __”). On October 3, 2016, Plaintiffs filed Plaintiffs’
Memorandum In Opposition To Motion To Compel (Dkt. 31). Defendant filed, on
October 11, 2016 its Reply To Plaintiffs’ Opposition To Defendant’s Motion (Dkt. 32)
(“Defendant’s Reply”). On October 12, 2016, Plaintiffs filed Plaintiffs’ Motion For Case
Management Order (Dkt. 33) (“Plaintiffs’ Motion”) together with Plaintiffs’ Brief In
Support of Plaintiffs’ Motion attaching exhibits 1-10 (“Plaintiffs’ Motion Exh(s). __”). On
November 3, 2016, Defendant filed its Response In Opposition To Plaintiffs’ Motion
(Dkt. 35) (“Defendant’s Response”). On November 10, 2016, Plaintiffs filed Plaintiffs’
Reply To Defendant’s Response Opposition To Plaintiffs’ Motion (Dkt. 36) (“Plaintiffs’

4

Reply”). Oral argument on Defendant’s and Plaintiffs’ motions was conducted on
November 30, 2016 (Dkt. 38) and decision reserved. On December 5, 2016, Defendant
filed Defendant’s Supplemental Memorandum (Dkt. 39) (“Defendant’s Supplemental
Memorandum”).1 Thereafter, on December 14, 2016, Plaintiffs filed Plaintiffs’
Supplement To Plaintiffs’ Motion (Dkt. 40 (“Plaintiffs’ Supplement”); in accordance with
the court’s request (Dkt. 41), Defendant’s Response to Plaintiffs’ Supplement was filed
December 20, 2016 (Dkt. 42) (“Defendant’s Response to Plaintiffs’ Supplement”).
In Defendant’s Motion, Defendant contends that Defendant did not, contrary to
Plaintiffs’ assertion, agree to proceed with discovery and trial in this case based on a
limited sample of Plaintiffs’ projects, i.e., 10 cases to be selected by each side (“selfselected sample”), or any other sampling procedure and that before agreeing to any
sampling procedure it would be necessary for Defendant to assess, by evaluation of
Plaintiffs’ answers to Defendant’s Superseding Interrogatory, whether the case was
amenable to a particular form of sampling. Dkt. 29 at 10; Dkt. 35 at 9-11. Plaintiffs
objected to Defendant’s Superseding Interrogatory asserting the Interrogatory, which
requested Plaintiffs, inter alia, identify the business component as defined by §
41(d)(2)(B) for each project for which Plaintiffs claims an R&D tax credit, the nature of
the uncertainty requirement under § 41, the particular science, e.g., biology, chemistry,
involved in the related QRE, the process of experimentation used in connection with
Plaintiffs’ projects, how the results of such required experimentation would be useful in
Plaintiffs’ development of a new or improved business product, process, etc., and the

1

A further oral argument was conducted on January 26, 2017 (Dkt. 45) limited to the question of why the
sampling plan devised by the I.R.S. to facilitate the audit of Plaintiffs’ claims, see Dkt. 33-5, would not be
suitable to provide a workable sample of Plaintiffs’ projects as the basis upon which to litigate the merits
of Plaintiffs’ claims.
5

expenses and related amounts for which the client was billed and the project was
performed, was premature absent an agreement to use a sample of Plaintiffs’ projects
as the basis for discovery, rather than requiring Plaintiffs to access the records relevant
to all 6,100 projects. Dkt. 29, Exh. 10 at 1. Plaintiffs thus refused to answer the
Superseding Interrogatory based on excessive burdensomeness, harassment and lack
of relevance to the randomly selected “judgment sample,”10-12, of Plaintiffs’ projects,
and referred Defendant to the original list of 6,100 projects and information, including
the Axiom study, Plaintiffs previously provided to Defendant in digital form, Dkt. 29, Exh.
10 at 3-6, and agreed to supplement such responses to provide documents following
selection of an acceptable sampling technique. Id. at 6.2 As to Defendant’s request for
identification of Plaintiffs’ business component relevant to each claim, Plaintiffs referred
Defendant to the study prepared by Axiom and Plaintiffs’ responses to “IDR requests,”3
which Plaintiffs state had been provided to Defendant and that such components
included “designs, process, and techniques” employed by Plaintiffs “for each project”
without further description. Id. at 4-5. As to Defendant’s request for an explanation of
the degree of uncertainty as to how the relevant business component and final design
would be developed by Plaintiffs as required for a QRE, Plaintiffs stated that such
uncertainty existed “at the outset” of each project which was “overcome” by Plaintiffs’
employees, and that further details responsive to the Superseding Interrogatory would
be forthcoming upon selection of the anticipated “judgment sample” from among
Plaintiffs’ projects. Id. at 5. In response to Defendant’s request for a description of the
scientific principles involved in the QREs for each project, Plaintiffs asserted
2

The record is unclear as to whether Plaintiffs had previously limited Plaintiffs’ claims to 4,643 projects,
see Defendant’s Motion Exhs. 8, 9, 10.
3
Such IDR requests are not defined in the record.
6

Defendant’s request was vague and ambiguous but that for all Plaintiffs’ projects,
Plaintiffs nevertheless employed “principals [sic] of engineering, science, mathematics,
physics . . . chemistry and biology.” Id. at 5. As to Defendant’s request that Plaintiffs
describe the process of experimentation required by § 41 for the QRE, Plaintiffs stated
such process was comprised of “a systematic iterative process” whereby technical
issues and desired outcomes and potential solutions and alternatives were identified,
i.e., by trial and error, a methodology to analyze and test such potential solutions and,
where such potential solutions were proven to be unavailing, repeating this process until
a viable solution was identified. Dkt. 29, Exh. 6 to Defendant’s Motion at 9. Plaintiffs
also responded by stating that once the appropriate “judgment sample” was selected,
Plaintiffs would “provide more specific detail on the technical challenges” mentioned by
Plaintiffs and “how a process of experimentation was employed to resolve the technical
challenges.” Id. Similar responses were provided by Plaintiffs to Defendant’s other
Superseding Interrogatory requests including information as to Plaintiffs’ projects’
clients, managers, source of funding and expenses. As can be seen from the foregoing
summarization of Plaintiffs’ responses, Plaintiffs’ responses were highly generalized,
essentially parroting the statutory elements required for an R&D tax credit without
providing sufficiently specific information from which Defendant could determine the
factual parameters upon which Plaintiffs’ refund claim for any of the 6,100 (or 4,643)
projects is based, despite Plaintiffs’ acknowledgment that such responsive information
is in fact available to Plaintiffs and that Plaintiffs would make it available to Defendant
but only as to an extremely small sample, 10-12 projects, to selected by the parties at a
later time. See Dkt. 29, Defendant’s Motion Exh. 10 at 5 (“Once a judgment sample list

7

of projects has been selected for discovery and to be tried, the project documents for
each project . . . will be made available [to Defendants].”4 (underlining added5))
Defendant rejects this limitation on Defendant’s access to the facts underlying
Plaintiffs’ claim as unworkable given the absence of any facts upon which it could be
inferred that such limited self-selected samples would be sufficiently representative of
the entirety of Plaintiffs’ projects upon which Plaintiffs’ refund claim is based. Dkt. 29,
Exh. B at 13 (“Using test cases in this litigation is only appropriate if it will produce a
valid estimate of [Plaintiffs’] qualified research expenses.”); Dkt. 35 at 11 (suggesting
that if Plaintiffs had been forthcoming in segmenting the 6,100 projects into groupings
with common characteristics such responsive details could have facilitated a stipulated
sampling procedure). Defendant also argues Plaintiffs’ burdensomeness objection
should be rejected as Plaintiffs have failed to sufficiently particularize, including by
presenting affidavits from persons with direct knowledge of Plaintiffs’ record keeping
system, demonstrating such excessive burdensomeness. Dkt. 29, Exh. B at 13-14
(citing Swinton v. Livingston County, 2016 WL 1056608, at *6 (W.D.N.Y. Mar. 17, 2016)
(quoting In re Priceline.com, Inc. Sec. Litig., 233 F.R.D. 83, 85 (D.Conn. 2005))); Burns
v. Imagine Films Entm’t, Inc., 164 F.R.D. 589, 593 (W.D.N.Y. 1996) (objections to
interrogatories required to be specific with detailed explanations supporting an
objection). Plaintiffs’ burdensomeness objection is further undercut by the fact that
Plaintiffs are required to maintain records sufficient to support Plaintiffs’ tax credit and
refund claims, a requirement Plaintiffs do not dispute. See 26 C.F.R. § 1.41-2(d); see
also Bayer Corp. and Subsidiaries v. United States, 850 F.Supp.2d 522, 539 (W.D.Pa.

4
5

Unless indicated otherwise bracketed material has been added.
Unless indicated otherwise underlining has been added.
8

2012) (“Bayer”) (rejecting use of plaintiffs’ proposed restrictive sampling plan without
particularizing nature of plaintiffs’ business components related to plaintiffs’ QREs as
basis for plaintiffs’ R&D tax refund claim of approximately $50 million so as to not give
“a reward to [plaintiffs] for failing to keep evidence regarding research expenses in
‘sufficiently’ useable form and detail.”). Given Plaintiffs are seeking a refund of
approximately $1.499 million, for similar reasons Plaintiffs’ burdensomeness objection
fails on its face.
The force of Plaintiffs’ further contentions with regard to the availability of some
form of sampling as typically utilized in R&D tax credit refund litigation in recognition of
the prohibitively extensive amount of potential trial time that would be imposed on the
court if all of Plaintiffs’ projects were to require trial, Plaintiffs’ Brief, Dkt. 33, at 5, is
significantly diluted by Plaintiffs’ recent and unexpected unilateral decision to reduce to
159 the number of projects for which Plaintiffs seeks R&D tax credits and a resultant
refund. See Plaintiffs’ Supplement (Dkt. 40) at 1 (“Plaintiff [sic] submits an amended list
of projects which Plaintiff [sic] claims as part of the calculation of qualified research
expenses.”); Defendant’s Response to Plaintiffs’ Supplement (Dkt. 42) at 1 (“This
amended list contained only 159 projects, a substantial reduction from the 6,100
projects originally claimed.”).6 Despite such a dramatic reduction in the Plaintiffs’
projects at issue, Plaintiffs nevertheless reiterate their request that discovery and trial be
limited to 10-12 sample projects to be selected from among the 159 projects now relied

6

Although in its Supplement Plaintiffs state “will not seek to claim qualified research expenses from
projects not included in the list of 159 projects attached as ‘Exhibit A,’” Plaintiffs have not filed an
amended complaint consistent with this representation. Plaintiffs are therefore directed to file, promptly,
an amended complaint consistent with Plaintiffs’ Supplement particularizing the tax years to which the
159 projects relate as well as the amount of QRE credit and tax refund Plaintiffs now seek based on the
159 projects correctly asserting the status, as Plaintiffs and Defendant, of the respective parties.
9

on by Plaintiffs in support of its claims, to be selected either randomly by the court or by
the selection of five to six projects by each party. Plaintiffs’ Supplement (Dkt. 40) at 2.
Unsurprisingly, Defendant rejects Plaintiffs’ proposal, Defendant’s Response to
Plaintiffs’ Supplement (Dkt. 42) at 2-3 (court should order “discovery . . . on all 159
claimed projects” as the results of such discovery may support disposition of the matter
on summary judgment or Plaintiffs’ entire refund claim (and the case) may be rendered
moot by the loss of approximately two-thirds of Plaintiffs’ original claimed QREs based
on the 6,100 projects initially asserted by Plaintiffs as supporting Plaintiffs’ original
refund claims. See Defendant’s Response to Plaintiffs’ Supplement at 2 (calculating
that reducing to 159 the relevant projects upon which Plaintiffs’ claims are now based
lowers to approximately $546 thousand Plaintiffs’ claimed tax refund from approximately
$1.4 million, resulting from a reduction in Plaintiffs’ QREs to approximately $5.5 million
from $14.5 million, see Defendant’s Response to Plaintiffs’ Supplement (Dkt. 42) at 2,
thereby potentially “eliminat[ing] Plaintiff’s [sic] claimed entitlement to a refund”).
Defendant further notes that despite Plaintiffs’ 97% reduction in the number of claimed
projects, Plaintiffs have failed to sufficiently support that providing discovery as to all
159 projects would nevertheless impose an excessive amount of burdensomeness
upon Plaintiffs. Id. at 5-6. Defendant also questions whether Plaintiffs had any good
faith basis for its refund claim based on the original assertion of 6,100 projects given
that Plaintiffs have acknowledged it failed to do so. Id. at 6 (citing Plaintiffs’
Memorandum (Dkt. 33) at 11 (admitting Plaintiffs’ QRE allegations were based on
employee “interviews,” a “review” of Plaintiffs’ financial records, and “estimations” of
expenses attributable to the projects by the Axiom study) or provide sufficient responses

10

to Defendant’s Superseding Interrogatory seeking the factual bases for Plaintiffs’ claims,
particularly Plaintiffs’ QREs, as required by § 41. Id. at 6. The court therefore turns to
the question of whether discovery in this case should proceed, at this stage of the
litigation, on the basis of a sample of the 159 projects as Plaintiffs now request or
whether discovery should be directed against all 159 cases as Defendant argues.
The use of statistical sampling techniques to resolve questions of liability and
damages in federal court during discovery and for trial in cases involving numerous
parties or large quantities of documents is now well-established. See Hallmark v.
Cohen & Slamowitz, 304 F.R.D. 165, 169 (W.D.N.Y. 2015) (“The use of sampling in
connection with discovery in class actions as a practical and cost-effective means by
which to obtain relevant discovery is well-recognized.” (citing Hinterberger v. Catholic
Health Sys., 284 F.R.D. 94, 103 (W.D.N.Y. 2012) (citing MANUAL FOR COMPLEX
LITIGATION, Fourth, § 21.14 (2009) (acknowledging use of sampling methodology for
discovery in class actions may be appropriate)))); Schwab v. Philip Morris USA, Inc.,
449 F.Supp.2d 999, 1244 (E.D.N.Y. 2006), rev’d sub nom on other grds., McLaughlin v.
American Tobacco Co., 522 F.3d 215 (2d Cir. 2008) (citing authorities and cases).
“[W]hen it is necessary to limit discovery, ‘statistical sampling techniques [may be used]
to measure whether the results of the discovery fairly represent what unrestricted
discovery would have been expected to produce.’” Id. (quoting MANUAL FOR COMPLEX
LITIGATION (FOURTH) § 11.422 (bracketed material in original)). “‘The use of acceptable
sampling techniques in lieu of discovery and presentation of voluminous data from the
entire population, may produce substantial savings in time and expense.’” Id. (quoting
MANUAL FOR COMPLEX LITIGATION (FOURTH) § 11.493). Such sampling has been found

11

useful in the litigation of R&D tax cases including the extent to which a taxpayer has
incurred QREs meeting the criteria of § 41. See, e.g., United States v. McFerrin, 570
F.3d 672, 679 (5th Cir. 2009) (“If McFerrin can show activities were “qualified research,”
then the court should estimate the expenses associated with those activities.” (citing
Cohan v. Comm’r of Internal Revenue Service, 39 F.2d 540 (2d Cir. 1930))); see also
Suder v. Comm’r of Internal Revenue, 108 T.C.M. (CCH) 354, 2014 WL 4920724, at *6
(2014) (basing court’s rejection of respondent’s deficiency and penalty determination for
two of petitioners’ tax years upon petitioner’s alleged excessive R&D tax credit claims
based on the parties’ stipulated selection of 12 of the 76 projects upon which petitioners
relied for computation of the claimed QREs related to each project and upon which
petitioners’ returns were prepared); Union Carbide Corp. and Subsidiaries v.
Commissioner of Internal Revenue, T.C.M. (CCH) 50, 2009 WL 605161, at *2 (2009)
(deficiency for improper R&D tax credits resolved based on parties’ agreement to try
case on five of largest of 106 claimed projects); contra Bayer Corporation and
Subsidiaries, 850 F.Supp.2d at 546 (rejecting use, in R&D tax refund case, of plaintiffs’
proposed sampling technique where, in response to defendant’s interrogatory, plaintiff
failed to particularize nature of plaintiff’s business components underlying plaintiff’s
alleged QRE credits and citing cases). 7 Despite the use of limited self-selected sample
projects in Tax Court tax deficiency litigation, as Plaintiffs have proposed for this case,
courts are reluctant to approve the use of such self-selected exemplars that could
7

The court notes that the I.R.S. formulated a statistical sample plan using, in part, random sampling as
the “most practical and efficient examination method,” Dkt. 33-5 at 3, to conduct an audit of Plaintiffs’
R&D tax credit claims. See Dkt. 33-5. At the further oral argument conducted January 26, 2017,
Defendant stated that because the I.R.S. sampling plan was based on groupings of Plaintiffs’ projects
based on the amount of claimed QREs and not on the various types of remediation projects such as
water and ground contamination represented in Plaintiffs’ QREs, it was unacceptable to Defendant;
Plaintiffs indicated the I.R.S. sampling plan was not preferable but acceptable.
12

create the possibility that such self-selection will not result in a sample representative of
all claims at issue in the case. See In re Chevron U.S.A., 109 F.3d 1016, 1020-21 (5th
Cir. 1997) (disapproving use of 30 of plaintiffs’ 3,000 cases, self-selected by the parties,
to extrapolate defendant’s toxic tort liability for the class as this procedure did not
support that such limited sample would be representative of all of plaintiffs’ claims in the
case and that such failure to assure such representativeness also violated due
process). “Typically, a finding [of liability] must be based on competent, scientific,
statistical evidence that identifies the variables involved and that provides a sample of
sufficient size so as to permit a finding that there is a sufficient level of confidence that
the results obtained reflect results that would be obtained from trials of the whole.” Id.
at 1020.
Here, similar to the dispute in Bayer, Defendant refuses to agree to any form of
sampling even of the reduced number (159) of Plaintiffs’ projects drawn from the
original 6,100 projects which Plaintiffs now claim form the basis of Plaintiffs’ refund
claim without benefit of some indication that Plaintiffs can or cannot establish there is
any factual basis for Plaintiffs’ claims, for example, the exact nature of Plaintiffs’
asserted business components, new products, techniques, etc., to which any of
Plaintiffs’ alleged QREs relate as required by § 41. To permit Plaintiffs to proceed
based on a self-selected sample of 10-12 cases as Plaintiffs propose, Dkt. 40, Plaintiffs’
Supplement, at 2, to be randomly selected by the court or self-selected by the parties,
will not allow Defendant to understand which, if any, of the remaining cases may or may
not meet the requirements of § 41, and would require Defendant to litigate the merits of
the case based on a sample which, depending on the actual characteristics of the 148-

13

150 remaining cases with respect to QRE requirements, may well be too small to be
fairly representative of the population of 159 projects, albeit a significantly reduced
number. In particular, “[a] sample is said to be representative with respect to a variable
[or relevant characteristic of the population to be sampled] if its relative distribution in
the sample is equal to its relative distribution in the population.” Jake Bethlehem,
APPLIED SURVEY METHODS – A STATISTICAL PERSPECTIVE, Wiley (2009) (italics in original)
at 24. As relevant, it is also accepted that “[b]igger samples give estimates that are
more precise,” and where the population to be “sampled is heterogeneous, random
error8 will be large; a larger sample will be needed to offset the heterogeneity.”
REFERENCE MANUAL ON SCIENTIFIC EVIDENCE (THIRD ED.), Federal Judicial Center, at 246.
Moreover, “[a] pilot sample may be useful to estimate heterogeneity and determine the
final sample size.” Id.
In the instant case, given that there are at least eight relevant, i.e., required,
factors for any valid QRE in accordance with § 41, it is likely that among the population
of 159 projects, a significant number will not meet all applicable requirements thereby
rendering the population sufficiently heterogeneous with regard to the relevant
variables, i.e., § 41 requirements, and requiring a larger sample in order to reasonably
assure the resultant sample is reliably representative. More specifically, the apparent
flaw in Plaintiffs’ self-sampling proposal is that, absent some indication as to the QRE
variations among the 159 projects put forth by Plaintiffs, it is not possible to determine
whether the population of 159 projects is amenable to the use of any sampling
technique for purposes of discovery and trial, including use of a random sample. Thus,
8

The calculated “impact of random chance on study results.” REFERENCE MANUAL ON SCIENTIFIC
EVIDENCE at 240.
14

in the absence of an affidavit from one qualified to advise the court on a competent
sampling technique for this case, the court is required to makes its own judgment to
assure any sample is reliably representative. See Hallmark v. Cohen & Slamowitz, 307
F.R.D. 102. 106 (W.D.N.Y. Jan. 8, 2015), aff’d sub nom, Hallmark v. Cohen &
Slamowitz, LLP, 2015 WL 4483959 (W.D.N.Y. July 22, 2015) (“Hallmark”) (failure by
defendant to proffer technical support for defendant’s sampling proposal leads court to
impose 10% sample of disputed documents, standard credit card contract terms and
conditions to discern clauses supporting defendant’s contractual authorization defense).
In contrast to Hallmark, which involved a population of more than 1,000
documents – standard preprinted credit card form contracts – with a high degree of
likely homogeneity, here the exact nature and details of the 159 projects, i.e., the types
of asserted remediation involved, is presently unknown as Exhibit A to Plaintiffs’
Supplement indicates only that all 159 projects involved remediation of various
contaminations resulting in a “new process of remediation” not otherwise specified,
alleged to comprise the requisite business component, along with identification of the
project manager and project location. Nor is this representation by Plaintiffs submitted
under oath. Significantly, Plaintiffs’ labeling of the projects as involving remediation
provides no indication as to whether the other prerequisites for a QRE, such as the
nature of experimentation, perceived uncertainties, why the process is “new,” compared
to Plaintiffs’ previous “processes,” and applied scientific principles, are present in any of
the 159 projects. As a safeguard against possible reliance on an unrepresentative
sample it is therefore necessary to enlarge the size of the random sample of the 159
projects, as a “pilot” sample, to provide preliminary insights as to whether discovery

15

(and possible trial) of Plaintiffs’ claims can be confidently predicated based on any
sample of the 159 projects, or whether full discovery and trial must be conducted as to
all 159 projects in order to provide “assurance” to the court against awarding any relief
to Plaintiffs amounting to “‘unguided largesse.’” Bayer Corp. and Subsidiaries, 850
F.Supp.2d at 542 n. 40 (quoting Williams v. United States, 245 F.2d 559, 560 (5th Cir.
1957)).
Thus, in the instant case, to determine whether the case is amenable to use of a
sampling technique such as random9 sampling, either by stipulation or one imposed by
the court, the parties shall within 10 days of this Decision and Order randomly select 40
projects (approximately 25%) of the Plaintiffs’ 159 projects as a pilot sample. The
parties shall, prior to taking the sample, establish a protocol which assures that all
project files to be sampled are marked with a permanent identification number and
indexed by project name and identification number so as to prevent any possible
manipulation of the sample project files after the sample is completed by substituting
files preferred by Plaintiffs. Defendant shall have reasonable access to Plaintiffs’
project files to assure the integrity of the sampling procedure. The Plaintiffs shall then
serve answers to Defendant’s Superseding Interrogatory for each project identified by
the random sample within 30 days thereafter. Defendant’s review of Plaintiffs’ answers
shall be completed within 30 days of the service of Plaintiffs’ answers; Defendant shall
advise Plaintiffs in writing when Defendant’s review is complete. The parties, within 10
days after notice of completion of Defendant’s review of Plaintiffs’ responses and after
evaluating the results of the pilot sample, shall jointly as a proposed stipulation or

9

“A random sampling is one in which every item in the population is equally likely to be chosen.” Alan
Graham, STATISTICS – A COMPLETE INTRODUCTION, Hodder & Stroughton (2013) at 129.
16

individually advise the court whether some, if any, form of sampling upon which to
proceed with this case, including discovery and further proceedings, acceptable to the
parties, together with proposed scheduling orders governing such further proceedings.
The parties may also request a conference with the court to discuss the scheduling of
further proceedings following completion and review of the results of the pilot sample
and Plaintiffs’ answers to the Superseding Interrogatory.

CONCLUSION
Based on the foregoing, Defendant’s motion (Dkt. 29) is GRANTED in part and
DENIED in part; Plaintiffs’ motion (Dkt. 33) is DENIED.
SO ORDERED.
/s/ Leslie G. Foschio
________________________________
LESLIE G. FOSCHIO
UNITED STATES MAGISTRATE JUDGE

Dated: January 26, 2017
Buffalo, New York

17