IN THE UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF OHIO EASTERN DIVISION
TREVEN JOHNSON, CASE NO. 4:20-cv-2584
Plaintiff, JUDGE BENITA Y. PEARSON
vs. MAGISTRATE JUDGE AMANDA M. KNAPP
INTERNATIONAL STEEL AND MEMORANDUM OPINION & ORDER COUNTERWEIGHTS LLC,
Defendant.
This civil action is before the undersigned on a referral pursuant to 28 U.S.C. 636, and Local Rules 72.1 and 72.2(a), to resolve the parties’ discovery dispute. For the reasons set forth below, the Court GRANTS in part and DENIES in part, the Plaintiffs’ Motion for a Protective Order. Plaintiffs’ request that Defendant be limited to representative discovery from the Opt-in Plaintiffs is GRANTED. Plaintiffs’ request that discovery be specifically limited to no more than six Plaintiffs is DENIED. Representative discovery is permitted as follows: (1) Plaintiff Treven Johnson shall respond to pending written discovery requests consistent with Federal Rules of Civil Procedure 33(a), 34, and 36(a); (2) A randomly selected group of thirty-four Opt-in Plaintiffs shall respond to limited written discovery requests that do not exceed ten (10) interrogatories and five (5) requests for production of documents, with the parties to agree on the process for selection; and (3) Defendant may conduct depositions of up to ten (10) Plaintiffs, as selected by Defendant. The referral is hereby terminated. I. PROCEDURAL BACKGROUND Plaintiff Treven Johnson brought this collective action against Defendant International Steel and Counterweights LLC under the Fair Labor Standards Act (“FLSA”), and was joined by fifty-two individuals who timely opted to participate as Opt-in Plaintiffs. Plaintiffs collectively
assert that Defendant only paid them for work performed between their scheduled start and stop times, and did not pay them for work performed before or after that time period. (ECF Doc. 37, p. 2; ECF Doc. 1, p. 3.) Although Plaintiffs reportedly performed a range of job duties that fell into twelve different job categories, they all worked at the same facility in Ohio. (ECF. Doc. 38, p.6.) Some Opt-in Plaintiffs performed tasks from more than one job category over the course of their employment with Defendant. (Id., at p. 6.) Defendant has served interrogatories, requests for production of documents, and notices of deposition on the named Plaintiff and all fifty-two Opt-in Plaintiffs. (ECF Doc. 38, p. 2.) Plaintiffs objected and the parties filed a Notice of Discovery Dispute with the Court on October 25, 2021. (ECF Doc. 32.) The dispute was referred to the undersigned. (ECF Doc. 33.) The
undersigned held a telephone conference with counsel for both parties regarding the discovery dispute on October 29, 2021. (ECF Doc. 36.) The parties subsequently advised the Court that further negotiations had failed to resolve the conflict, and a briefing schedule was set. On November 4, 2021, Plaintiffs filed a Motion for a Protective Order to stop Defendant from seeking individualized discovery from all fifty-three Plaintiffs, arguing that Defendant should be limited to representative discovery from no more than six randomly selected Plaintiffs. (ECF Doc. 37.) Defendant filed a brief in opposition, requesting full individualized discovery from all fifty-three Plaintiffs and arguing that individualized discovery is not unduly burdensome given the relatively small size of the group. (ECF Doc. 38, pp. 5-6.) In the alternative, Defendant argued that any representative discovery should involve at least thirty-four Plaintiffs, to ensure statistical significance. (Id., at pp. 7-9.) Plaintiffs responded that representative discovery from more than six Plaintiffs would be unduly burdensome, and would impose expenses that would significantly outweigh any benefit. (ECR Doc. 39.)
II. STANDARD OF REVIEW The Federal Rules of Civil Procedure grant courts broad discretion in determining the scope and method of discovery based upon the circumstances of each case. Federal Rule of Civil Procedure 26(b)(2) permits a court to limit discovery, particularly when such discovery may be duplicative, more readily obtained from another source, or when the burden or expense outweighs the benefits of the discovery. III. DISCUSSION A. Representative Discovery is Appropriate As a general matter, Defendant’s request for individualized discovery from the named Plaintiff and all fifty-two Opt-in Plaintiffs, including full written discovery and depositions, is
not well taken. A review of both the remedial purposes of the FLSA and the applicable caselaw supports a finding that Plaintiffs have shown good cause for the issuance of a protective order requiring the use of representative discovery in this collective action. The FLSA is “a broadly remedial and humanitarian statute … designed to correct ‘labor conditions detrimental to the maintenance of the minimum standard of living necessary for health, efficiency, and general well-being of workers…’” Dunlop v. Carriage Carpet Co., 548 F.2d 139, 143 (6th Cir. 1977) (quoting 29 U.S.C. s 202(a)); see also Keller v. Miri Microsystems LLC, 781 F.3d 799, 806 (6th Cir. 2015) (noting “Congress passed the FLSA with broad remedial intent”). “To prevail in an FLSA overtime suit, a plaintiff must prove, by a preponderance of the evidence, that he ‘performed work for which he was not properly compensated.’” Moran v. Al Basit LLC, 788 F.3d 201, 205 (6th Cir. 2015) (quoting Anderson v. Mt. Clemens Pottery Co., 328 U.S. 680, 687, 66 S.Ct. 1187, 90 L.Ed. 1515 (1946), superseded by statute on other grounds, Portal–to–Portal Act of 1947, Pub.L. No. 80–49 § 4(a), 61 Stat. 86–87 (codified at 29 U.S.C.
254(a)), as recognized in Integrity Staffing Solutions, Inc. v. Busk, 574 U.S. 27, 135 S.Ct. 513, 516–17, 190 L.Ed.2d 410 (2014)). “The remedial nature of this statute and the great public policy which it embodies, however, militate against making that burden an impossible hurdle for the employee.” Moran, 788 F.3d at 205 (quoting 328 U.S. at 687, 66 S.Ct. 1187). The purposes of the FLSA include “(1) reducing the burden on plaintiffs through the pooling of resources, and (2) efficiently resolving common issues of law and fact that arise from the same illegal conduct.” Morgan v. Fam. Dollar Stores, Inc., 551 F.3d 1233, 1264 (11th Cir. 2008) (citing Hoffmann–La Roche, Inc. v. Sperling, 493 U.S. 165, 170, 110 S.Ct. 482, 107 L.Ed.2d 480 (1989)); see also Gentrup v. Renovo Servs., No. 1:07-cv-430, 2010 WL 6766418 at *1 n. 1 (S.D. Ohio Aug. 17, 2010) (“[A] collective action under Section 216(b) allows employees
‘the advantage of lower individual costs to vindicate rights by pooling resources’ as well as benefitting the judicial system by ‘efficient resolution in one proceeding of common issues of law and fact.’”) (quoting Hoffmann–La Roche, 493 U.S. at 170); O’Brien v. Ed Donnelly Enterprises, Inc., 575 F.3d 567, 584 (6th Cir. 2009) (collective actions “serve an important remedial purpose” by ensuring that “a plaintiff who has suffered only small monetary harm can join a larger pool of similarly situated plaintiffs” and the “pool can attract effective counsel”). Courts have found representative discovery to be appropriate and in keeping with the remedial purposes of the FLSA, even where the number of opt in plaintiffs was relatively small. See, e.g., Gentrup, 2010 WL 6766418 at *7 (finding “in the spirit of collective actions, representative discovery is appropriate” and even “superior given the alleged existence of company-wide policies, the size of the class, the burdensome nature of full discovery upon 115 Plaintiffs in 20 different states, and the potentially small amount of each individual claim”); Ludlow v. Flowers Foods, Inc., No. 18-cv-1190, 2020 WL 3791638, *12-13 (S.D. Cal. July 6,
2020) (finding the volume of individual discovery requests and deposition notices propounded on class of 115 opt-in plaintiffs would impose a significant burden on plaintiffs); Adkins v. Mid- America Growers, Inc., 141 F.R.D. 466 (N.D. Ill. 1992) (holding “[i]n a class action such as this, class-wide discovery was more appropriate,” and stating individualized discovery from 81 opt-in plaintiffs “exemplifies the hazards of individual class plaintiff discovery”); Sutton v. Diversity at Work Grp. Inc., No. 1:20-CV-00682, 2021 WL 2334488, at *2-3 (S.D. Ohio June 8, 2021) (addressing scope of representative discovery for class of 20 opt-in plaintiffs where “[b]oth parties agree to use representative discovery”); Rosenbohm v. Cellco P'ship, No. 2:17-CV-731, 2019 WL 2141901, at *3 (S.D. Ohio May 16, 2019) (addressing scope of representative discovery where defendant “is not opposed to sampling or representative discovery itself”).
While other courts have found individual discovery to be appropriate for a small group of opt-in plaintiffs, they have generally based these decisions on a case-specific analysis of the evidence and a weighing of the opposing interests and burdens of the parties. See, e.g., Frisbie v. Feast American Diners, LLC, 17-cv-6270, 2020 WL 2305083, *3 (W.D.N.Y. May 7, 2020) (finding individualized discovery of 20 opt-in plaintiffs supported when plaintiffs provided no more than a “bare bones assertion” that the burdens would outweigh the benefits, and no more than “bald statements” that the defendants had superior access to information); Mejia v. Bimbo Bakeries USA Incorporated, No. CV-16-654, 2018 WL 11352489, *3 (D. Ariz. May 4, 2018) (permitting individual discovery from 45 opt-in plaintiffs in the form of seven document requests that the court specifically found sought “relevant and discoverable information [that] is not improperly cumulative or overly burdensome”); White v. 14051 Manchester, Inc., No. 4:12-cv- 469, 2013 WL 1867113, *2 (E.D. Mo. May 2, 2013) (finding it was not unduly burdensome for 50 opt-in plaintiffs to respond to specific written discovery the Court found to be “not impossible
for a normal person to understand and answer,” while also finding the counter-proposal that 10 plaintiffs respond to a simple questionnaire prepared by plaintiffs would not provide “the information necessary” for the decertification process); Daniel v. Quail Intern., No. 3:07-CV-53, 2010 WL 55941, *1 (M.D. Ga. Jan 5, 2010) (finding individual discovery of 36 opt-in plaintiffs supported when there was no evidence that the representative sample proposed by plaintiffs would be statistically significant, and plaintiffs failed to show that individualized discovery would raise sufficient efficiency concerns to justify representative discovery); Page v. Grandview Marketing, Inc., Nos. 2:09-cv-1150, 2:09-cv-1152, 2010 WL 11629668, * 6-7 (D. Nev. Oct. 6, 2010) (permitting individual written discovery for a class of 80 opt-in plaintiffs, where the proposed sample of 8-10 plaintiffs would not be statistically representative and the
relevant requests would be “reasonably limited” to relevant information and documents “not available to Defendants” as specifically outlined in the order). After reviewing the circumstances and holdings in the above cases, the undersigned finds that the situations where individualized discovery was approved are distinct from this case in the scope of discovery sought and/or the burden imposed on the plaintiffs, and are therefore not instructive here. See Ludlow, 2020 WL 3791638, at *5 (taking “little guidance” from decisions that permitted “far less discovery”); Strauch v. Computer Scis. Corp., No. 3:14 CV 956 JBA, 2015 WL 540911, at *2 (D. Conn. Feb. 10, 2015) (“[T]here is far from a ‘bright line’ test as to the number of opt-in plaintiffs which tips the balance in favor of representative sampling as opposed to individualized discovery.”). Here, Defendant is seeking full written discovery and depositions from the named Plaintiff and all fifty-two Opt-in Plaintiffs. While the specific discovery requests are not in evidence,1 Plaintiffs have provided estimates of the time it would take their representatives to
respond to fifty-three sets of written discovery requests and defend fifty-three depositions, and the resulting attorney fees. (ECF Doc. 37, p. 11.) They have also provided adjusted fee estimates in the event that written discovery requests and depositions are permitted for a representative sample of thirty-eight Plaintiffs, as proposed by Defendant. (ECF Doc. 39, p. 3.) In both situations, Plaintiffs assert that the expenditures in fees alone would “exceed[] Plaintiff’s counsel’s current damages estimate for the opt in class, even when including liquidated damages.” (ECF Doc. 39, p. 3.) This evidence suggests there would be a significant burden on Plaintiffs if the proposed scope of discovery was permitted, as the continued prosecution of this lawsuit on behalf of the Opt-in Plaintiffs might become cost prohibitive.
On the other hand, Defendant has offered no specific evidence or argument to support a finding that the scope of discovery sought here is necessary to avoid similar costs or burdens on Defendant. Defendant argues that individualized discovery is appropriate because each job category has different duties, and therefore requires a separate analysis for the “off-the-clock” claims. (ECF Doc. 38, pp. 5-6.) But the evidence offered in support of this argument – a set of three job descriptions – focuses exclusively on the duties performed while the employees were
1 The specific terms of the discovery requests are not before the Court, as the present dispute is limited to the determination whether representative discovery is appropriate and, if so, what scope of representative discovery is appropriate. It is noted that the procedural posture of the case does not lend itself to a detailed inquiry into specific discovery requests, as the parties currently face a discovery deadline in a mere two weeks absent a favorable ruling from the District Judge on their pending Joint Motion to Vacate the Scheduling Order. (ECF Doc. 34.) Even if the discovery period is ultimately enlarged, it is evident that a prompt and efficient discovery process will be expected. on the clock. (ECF Docs. 38-1, 38-2, 38-3.) The current evidence does not support an finding that the clock in, clock out, or payroll procedures would necessarily or materially differ among the Opt-in Plaintiffs, who all worked at a single facility. Given the important remedial purposes of the FLSA, as discussed above, the undersigned
concludes that representative discovery is appropriate in this case. B. Statistically Significant Representative Discovery is Preferred With representative discovery permitted, the next issue to be addressed is the appropriate sample size for that discovery. Plaintiffs correctly argue that the usefulness of representative discovery turns on “the degree to which the evidence is reliable in proving or disproving the elements of the relevant cause of action.” (ECF Doc. 37, p. 6 (quoting Tyson Foods, Inc. v. Bouaphakeo, 136 S. Ct. 1036 (2016)).) Plaintiffs are also correct that a representative sample of a given population may “allow[] the use of data from a sample … to ‘draw conclusions about the entire population” (id. (quoting Rosenbohm v. Cellco Partnership, No. 2:17-CV-731, 2019 WL 2141901 (S.D. Ohio May 16, 2019))), and that “[l]imiting discovery to a statistically significant
representative sampling… will both reasonably minimize the otherwise extraordinary burden imposed on the plaintiffs and their counsel and yet afford the defendant a reasonable opportunity to explore, discover and establish an evidentiary basis for its defenses.” (EFC Doc. 37, p. 7 (quoting Smith v. Lowe’s Home Centers, Inc., 236 F.R.D. 354, 357-58 (S.D. Ohio 2006))). In arguing for these standards, however, Plaintiffs do not offer specific support for their assertion that limiting discovery to six randomly selected Plaintiffs in this case would produce statistically significant results. They simply note that this number would make up 11.4% of the class, and cite to other cases where courts permitted representative discovery using similar or smaller percentages of plaintiffs. (ECF Doc., pp. 7-8.) This type of “comparative percentage analysis” has been disapproved by some courts because it does not specifically correlate to statistically significant findings. See Sutton v. Diversity at Work Grp. Inc., No. 1:20-CV-00682, 2021 WL 2334488, at *4 (S.D. Ohio June 8, 2021) (finding “comparative percentage analysis” is “of little value”); Daniel, 2010 WL 55941, at *1 (“Plaintiffs ask the Court to limit individualized
discovery in this action to approximately 15% of the class. However, they point to nothing in the record to demonstrate that this number is statistically significant.”). While courts need not apply “a uniform rule of statistical reasoning” for representative discovery, they do recognize “the value of ‘statistically significant’ samples in representative discovery among FLSA Plaintiffs.” Sutton, 2021 WL 2334488, at *3 (citing Smith, 236 F.R.D. at 358); see also Reich v. S. New England Telecommunications Corp., 121 F.3d 58, 67 (2d Cir. 1997) (“there is no bright line formulation [of a representative sample] that mandates reversal when the sample is below a percentage threshold”). Nevertheless, the Supreme Court has recognized that “[t]he fairness and utility of statistical methods ... will depend on facts and circumstances particular to those cases.” Tyson Foods, Inc. v. Bouaphakeo, 577 U.S. 442, 460,
136 S.Ct. 1036, 194 L.Ed.2d 124, (2016); see also Secretary of Labor v. DeSisto, 929 F.2d 789, 793 (1st Cir. 1991) (“the adequacy of the representative testimony necessarily will be determined in light of the nature of the work involved, the working conditions and relationships, and the detail and credibility of the testimony”). Defendant argues that a recent decision from the Southern District of Ohio is instructive here, to the extent that representative discovery is permitted. (ECF Doc. 38, pp. 7-8.) The Court in Rosenbohm used broadly available statistical methods to calculate the representative sample class using a 95% confidence interval and 10% margin of error. See Rosenbohm, 2019 WL 2141901 at *1-2. Using the number supplied by this calculation, the Court determined that a randomly selected sample of such plaintiffs would provide a “statistically sound” means to ensure that “the sample is representative of th[e] larger population.” Id. at *2. Another Court in the Southern District of Ohio recently applied the same calculation to arrive at a representative sample in an FLSA case. See Sutton, 2021 WL 2334488, *3-5.
Applying the same confidence interval and margin of error in this case would call for a relatively large sample of thirty-four representative Plaintiffs, roughly 65% of the total class. (ECF Doc. 38, p. 9.) This is because a larger percentage of the group is generally required for a sample to be statistically significant when the total class size is small. See Rosen v. Reckitt & Colman, Inc., No. 91 CIV. 1675 (LMM), 1994 WL 652534, at *3 (S.D.N.Y. Nov. 17, 1994) (offering examples to demonstrate the problem of achieving statistical “stability” in sampling small sets); citing Moultrie v. Martin, 690 F.2d 1078, 1083 & nn. 7–10 (4th Cir. 1982) (“[I]t is axiomatic in statistical analysis that the precision and dependability of statistics is directly related to the size of the sample being evaluated.”); United States v. La Chance, 788 F.2d 856, 866-67 (2d Cir. 1985) (explaining the dependence of statistical “fluctuations” upon sample size), cert.
denied, 479 U.S. 883 (1986); see also Rosenbohm, 2019 WL 2141901 at *2 (explaining principles of statistical sampling). Plaintiffs respond that the representative sample in Rosenbaum was only about 2.4% of the total class of Plaintiffs, but do not ultimately comment on the question of statistical significance for smaller classes such as this one. (ECF Doc. 39, p. 2.) Seeing no basis to align Plaintiffs’ suggested percentage method with statistical significance, the undersigned concludes that the principles underlying representative discovery weigh in favor of using a statistically significant sample for at least some portion of the discovery process. It is further noted that random selection of the representative class is recognized to be a most statistically sound means for approaching representative discovery. See, e.g., Sutton, 2021 WL 2334488, at *2 (“While random sampling is not a hard-and-fast rule in class-based discovery, courts have stood behind the proposition that a randomly selected sample helps reduce the likelihood of bias”) (citing Hostetler v. Johnson Controls, Inc., No. 3:15-CV-226 JD, 2016 WL 3662263 at *12. (N.D. Ind.
July 11, 2016)). For the reasons stated, the undersigned finds that Defendant may direct written discovery requests to a class of thirty-four randomly selected Plaintiffs, in addition to the named Plaintiff, in the manner and to the extent set forth more specifically below. C. Limited Discovery is Justified Under the FLSA and the Federal Rules Having found that some amount of discovery is appropriate for the relatively large representative class of thirty-four Plaintiffs, it remains to be determined what amount of discovery is consistent with the litigation needs of the parties and the remedial purposes of the FLSA. As noted above, Plaintiffs have provided fee estimates suggesting that full discovery relating to the representative class of thirty-four Plaintiffs could make the entire case cost
prohibitive. (ECF Doc. 37, p. 11; ECF Doc. 39, p. 3.) Defendant contends that broad discovery is appropriate, but has provided little evidence or argument to support this contention. 1. Limited Written Discovery is Permissible With respect to written discovery requests, the undersigned finds that the remedial purposes of the FLSA support an appropriately narrow and efficient discovery process to ensure that the Plaintiffs have access to the protections of that statute. This is consistent with the decisions of other courts, where the scope of permitted discovery has been narrowly tailored to fit the applicable circumstances. See, e.g., Gentrup, 2010 WL 6766418 at *7 (allowing written discovery to be served on 9 named plaintiffs and 22 additional opt-in plaintiffs, with defendants to depose 16 plaintiffs out of a group of 115); Bonds v. GMS Mine Repair & Maintenance, Inc., No. 13–cv–1217, 2014 WL 6682475, at *2 (W.D. Pa. Nov. 25, 2014) (permitting 5 written interrogatories to be served on 157 opt-in plaintiffs, permitting Defendant to then select 20 (12%) of those plaintiffs to respond to an additional 10 interrogatories and 8 requests for
production, and be subjected to depositions); O'Toole v. Sears Roebuck & Co., No. 11 C 4611, 2014 WL 1388660, at *2 (N.D. Ill. April 10, 2014) (limiting written discovery to 7 interrogatories and 17 requests for production, to be served on 33% of final total of opt-in plaintiffs, numberinh 700 at the time of the decision, and permitting defendant to depose up to 10% of the opt-in plaintiffs for no more than four hours each); Scott v. Bimbo Bakeries, Inc., No. 10–3154, 2012 WL 6151734, at *4–6 (E.D. Pa. Dec. 11, 2012) (permitting service of interrogatories on no more than 10% of the opt-in plaintiffs, numbering 650 at the time, and permitting defendant to depose 15-20 opt-in plaintiffs for no more than five hours each); In this case, where the class of Plaintiffs covers a variety of job descriptions but a single employer and site location, the undersigned concludes that relatively narrow written discovery
responses from a statistically significant sample of thirty-four Plaintiffs will allow Defendant sufficient opportunity to gather any needed evidence regarding the challenged practices and any similarities or differences among the Plaintiffs. Accordingly, the undersigned directs that: (1) Plaintiff Treven Johnson shall respond to pending written discovery requests consistent with Federal Rules of Civil Procedure 33(a), 34, and 36(a); and (2) A randomly selected group of thirty-four Opt-in Plaintiffs shall respond to limited written discovery requests that do not exceed ten (10) interrogatories and five (5) requests for production of documents. The parties are directed to meet and confer to agree upon the process for the random selection of Plaintiffs subject to this representative discovery. 2. Defendant May Conduct Ten Depositions With respect to depositions, the analysis must be different. While the Federal Rules of Civil Procedure generally permit written discovery to be propounded on all parties to an action, the Rules relating to depositions are distinct. Pursuant to Federal Rule of Civil Procedure
30(a)(2)(A)(i), a party must seek the Court’s leave to take more than ten depositions. The burden of persuading the court that these additional depositions are necessary is on the party requesting the additional depositions. See Moore v. Abbott Labs, No. 2:05-cv-1065, 2009 WL 73876, at *1 (S.D. Ohio Jan. 8, 2009) (“Because this limit is intended to curb abusive discovery practices, it stands to reason that a party wishing to conduct more than 10 depositions has the burden of persuading the court that these additional depositions are necessary”). This burden cannot be met through general assertions. Id. (citing Schiller America, Inc. v. Welch Allyn, Inc., No. 06- 21439-CIV, 2007 WL 2702247, *1 (S.D. Fla. Sept. 14, 2007)). In this case, Defendant has not made a particularized showing necessary to support a finding that more than ten depositions are necessary to its defense of the present matter. Given
the collective nature of the action and the ability to obtain information regarding most factual distinctions between the Plaintiffs by way of written discovery, the record before the Court does not support a finding that Defendant has need of such extensive deposition testimony. The remedial purposes of the FLSA are also considered in making this determination, as well as the representations of Plaintiffs regarding the impact of excessive litigation costs on their ability to seek the remedies made available by the FLSA. Accordingly, and for the reasons stated above, Plaintiffs’ Motion for Protective Order is granted with respect to the excessive notices of deposition. Defendant may conduct depositions of no more than ten (10) Plaintiffs, as permitted under the Federal Rules of Civil Procedure. Because this more limited number of ten depositions is not statistically significant under the standards discussed above, the undersigned finds that random sampling is not necessary and Defendant may determine the criteria to be used in selecting the Plaintiffs to be deposed. To the extent that Defendant feels its ability to defend this action is reliant on obtaining testimony from
individuals who performed each of the twelve identified job categories, it is noted that several Opt-in Plaintiffs have been identified as having performed more than one of those jobs. Thus, to the extent that Defendant believes discovery regarding the various job categories is essential to its defense, it is evident that ten depositions will yet suffice to meet that need.
Dated: November 17, 2021
s/ Amanda M. Knapp AMANDA M. KNAPP UNITED STATES MAGISTRATE JUDGE