FILED
United States Court of Appeals PUBLISH Tenth Circuit
UNITED STATES COURT OF APPEALS May 1, 2020
Christopher M. Wolpert
FOR THE TENTH CIRCUIT Clerk of Court
_________________________________
R. ALEXANDER ACOSTA, Secretary of Labor, United States Department of Labor,
Plaintiff - Appellee, v. No. 19-4097
PARAGON CONTRACTORS CORPORATION; BRIAN JESSOP,
Defendants - Appellants, and JAMES JESSOP, Defendant, and PAR 2 CONTRACTORS, LLC,
Intervenor Defendant.
_________________________________
Appeal from the United States District Court for the District of Utah
(D.C. No. 2:06-CV-00700-TC)
_________________________________
Rick J. Sutherland of Jackson Lewis, PLLC, Salt Lake City, Utah, for Defendants - Appellants.
Erin M. Mohan, Senior Attorney (Kate S. O’Scannlain, Solicitor of Labor, Jennifer S. Brand, Associate Solicitor, Maria Van Buren, Counsel for Child Labor and Special FLSA
Projects, and Paul L. Frieden, Counsel for Appellate Litigation of the U.S. Department of Labor, Office of the Solicitor, on the brief), for Plaintiff - Appellee.
_________________________________
Before MATHESON, KELLY, and PHILLIPS, Circuit Judges.
_________________________________
KELLY, Circuit Judge.
_________________________________
Defendants-Appellants Paragon Contractors Corporation and Brian Jessop (Paragon) appeal from the district court’s order and findings of fact and conclusions of law regarding the calculation of back wages. Plaintiff-Appellee United States Secretary of Labor (Secretary) sought to compel Paragon to replenish a fund established to compensate children employed without pay in violation of both the Fair Labor Standards Act (FLSA) and an injunction. Paragon had previously been held in contempt for violating the injunction. See Acosta v. Paragon Contractors Corp., 884 F.3d 1225, 1229– 30 (10th Cir. 2018). The district court accepted the Secretary’s calculation that Paragon owed $1,012,960.90 in back wages and ordered Paragon to replenish the fund in the amount of $812,960.90 less any interest already accrued.
On appeal, Paragon contends that the district court failed to adhere to the elements of a back wage reconstruction case under Anderson v. Mt. Clemens Pottery Co., 328 U.S. 680 (1946). Specifically, Paragon contends that the district court erred in (a) concluding that the Secretary established a prima facie case, (b) imposing an improperly high burden for rebutting the inferences arising from that case and holding that Paragon failed to rebut certain inferences, and (c) declining to apply a statutory exemption. We have jurisdiction pursuant to 28 U.S.C. § 1291, and we affirm.
3
Background
Paragon president Brian Jessop is a leader in the Fundamentalist Latter-day Saints (FLDS) church. I Supp. App. 5, 20, 137–38. Starting in 2008 and ending in 2012, Paragon entered annual agreements with the Southern Utah Pecan Ranch (Ranch) to harvest pecans on its property. II Supp. App. 527–28. Paragon received 30 percent of the proceeds from the harvest. Id. at 530. To complete the harvest, Paragon employed hundreds of children who were members of the FLDS church. Id. at 551. Witnesses testified that FLDS families faced extreme pressure to participate in the harvest. Id. at 534–36. The last agreement between the Ranch and Paragon ended in 2012, but workers made sworn declarations that children continued to participate in the harvest through 2013. I Supp. App. 1, 38.
In 2007, the district court permanently enjoined Paragon from using oppressive child labor in violation of the FLSA. I Aplt. App. 36–38. The Secretary sought the injunction following an enforcement action in which he found that Paragon unlawfully employed children without pay. Id. Paragon did not oppose the injunction. Id. at 36. In 2012, a news crew filmed hundreds of children harvesting pecans at the Ranch during a school day. II Supp. App. 525–26. The Ranch entered into a settlement agreement with the Secretary in December 2012. III Aplt. App. 544. That agreement created a “lockbox” funded by the proceeds of that year’s harvest, which would be held by the Secretary until litigation involving Paragon was completed. Id.
In 2015, the Secretary moved the court to hold Paragon in contempt for violating the 2007 injunction. I Aplt. App. 39. After a three-day evidentiary hearing that included
4
testimony from parents and children that had worked on the Ranch, the district court held Paragon in contempt. II Supp. App. 548. In 2016, the district court issued a sanctions order directing that Paragon establish a fund to be used by the Secretary to administer a claims process for children who performed uncompensated labor. Id. at 271. The district court considered the initial $200,000 sanction an “adequate starting point” and further ordered that Paragon “make additional payments as necessary into the Fund to fulfill all court-approved claims.” Id. at 274, 281. Paragon then appealed both the contempt and sanction orders. This court affirmed those orders in part, but reversed the district court’s appointment of a special master to monitor Paragon’s compliance with the injunction. Acosta, 884 F.3d at 1229–30.
After the conclusion of the claims period, the Secretary filed a proposed schedule of payments with the district court. II Aplt. App. 283–89. Paragon objected and the parties conducted limited discovery on the calculation process. Id. at 300–13. The parties agreed to several stipulations and the Secretary submitted an amended schedule. Id. at 313–15.
The district court held an evidentiary hearing on the amended schedule. The district court admitted into evidence a summary of claims information and redacted claims forms. I Aplt. App. 34; II Aplt. App. 335; III Aplt. App. 612 n.9; II Supp. App. 333–36. It also heard testimony from the official who prepared the proposed schedule for the Secretary. II Aplt. App. 290–91, 335. Paragon presented testimony from two witnesses who did not file claims but contended that they or their children chose to volunteer — rather than work — at the Ranch during some of the relevant years. II Aplt.
5
App. 335, 493, 497–98, 502; III Aplt. App. 563. The district court found that Paragon owed $1,012,960.90 in back wages and ordered it to replenish the fund in the amount of $812,960.90 less any interest accrued.
Discussion
We review questions of law de novo and the amount of a compensatory sanction for clear error. FTC v. Kuykendall, 371 F.3d 745, 76 (10th Cir. 2004). The methodology the district court uses to arrive at the amount, “‘such as determining the proper elements of the award or the proper scope of recovery is a question of law’ we review de novo.” Id. at 763 (quoting S. Colo. MRI, Ltd. v. Med-Alliance, Inc., 166 F.3d 1094, 1100 (10th Cir. 1999)). We review the district court’s findings of fact for clear error. Metzler v. IBP, Inc., 127 F.3d 959, 965 (10th Cir. 1997). A finding is clearly erroneous if we are left with a definite and firm conviction that a mistake has been made. Id.
A. Prima Facie Case The district court did not err by concluding that the Secretary carried his burden of establishing a prima facie case under Mt. Clemens. In that case, the Supreme Court reasoned that employers could not be allowed to benefit from unlawfully refusing or failing to keep records required under the FLSA by depriving employees of the means to establish a precise number of improperly compensated hours worked. Mt. Clemens, 328 U.S. at 687. To remedy this problem, the Court fashioned a burden-shifting framework:
[A]n employee has carried out his burden if he proves that he has in fact performed work for which he was improperly compensated and if he produces sufficient evidence to show
6
the amount and extent of that work as a matter of just and reasonable inference. The burden then shifts to the employer to come forward with evidence of the precise amount of work performed or with evidence to negative the reasonableness of the inference to be drawn from the employee’s evidence. If the employer fails to produce such evidence, the court may then award damages to the employee, even though the result be only approximate.
Id. at 687–88. Under this framework, a prima facie case may be made with representative evidence from a sample of employees. Donovan v. Simmons Petroleum Corp., 725 F.2d 83, 85–86 (10th Cir. 1983). In examining whether a prima facie case is made, courts consider that “[t]he reasonableness of the [government’s] proposed calculation depends in part on the availability of other, more reasonable alternatives to that proposal.” Acosta v. Off Duty Police Servs., Inc., 915 F.3d 1050, 1065 (6th Cir. 2019).
Mt. Clemens applies here because Paragon failed to keep any records of the work children performed at the Ranch. The Secretary met his burden to make a prima facie case and estimate the number of hours worked given the unique difficulties in getting better information. The district court ordered the Secretary to design and administer a process by which the eligible children could make sworn claims to the Secretary. The Secretary received claims from 104 eligible children,1 some of which detailed the dates
1 The Department of Labor initially received 135 claims but, “[b]ecause the FLSA’s child labor provisions permit 16- and 17-year olds to work in agriculture without restriction, see 29 U.S.C. [§] 203(1), 29 U.S.C. [§] 213(c)(1)-(2), the parties agreed that the Department would propose compensatory payments only for individuals who worked on the Ranch at ages 15 or younger.” Aplee. Br. at 12. After removing the 16- and 17-year olds, 104 claimants remained. II Aplee. App. 354, 361.
7
and times they worked for Paragon on the Ranch. III Supp. App. 633–736. The Secretary prepared a summary of these claims, calculated a proposed schedule of payments, and presented testimony at the evidentiary hearing. I Supp. App. 215. The Secretary used data obtained from the claims forms to fill in gaps where information was not available. See Aplee. Br. at 29–30. In addition, the Secretary’s prima facie case was supported by evidence from the contempt hearing, where several witnesses — including child workers — testified about work practices. See I Supp. App. 1–2, 16, 30, 40. Considering CNN’s video of SUPR, which showed hundreds of children at work, and testimony from witnesses who worked at SUPR — who testified that anywhere from 200 to 600 children worked at SUPR — we can reasonably infer that many child laborers did not submit a form.2 The Secretary’s evidence here surmounts the Mt. Clemens bar of establishing “the amount and extent of [the] work as a matter of just and reasonable inference.” 328 U.S. at 688. The government faced unique challenges in trying to obtain live testimony from impacted child laborers in this case.3 In light of these difficulties, requiring the government to do more would be tantamount to rewarding Paragon for its failure to keep records.
2 In fact, given that the reported claims appear to substantially underrepresent the number of child laborers employed, the government’s calculation may well be far less than the back wages owed. 3 Paragon compounded the difficulty by refusing to provide the government with the names of its employees at the Ranch, instructing its employees to lie about its use of child labor, and refusing to respond to government subpoenas, among other things. I Aplt. App. 270.
8
B. Rebutting the Prima Facie Case Paragon argues that the district court misread Mt. Clemens and set the bar too high for rebutting a prima facie case. The district court stated that an absence or inadequacy of records “shifts to the employer the burden of specifically and expressly rebutting the reasonable inferences drawn from the employee’s evidence as to the amount of time spent working.” III Aplt. App. 623. Paragon claims that this formulation of the burden is heavier than what the Supreme Court actually imposed when it required “the employer to come forward . . . with evidence to negative the reasonableness of the inference to be drawn from the employee’s evidence.” Mt. Clemens, 328 U.S. at 687–88.
We disagree. This court has held that “failure to comply with [the] obligation to keep records shifts to the employer the burden of specifically and expressly rebutting the reasonable inferences drawn from the employee’s evidence as to the amount of time spent working.” Bledsoe v. Wirtz, 384 F.2d 767, 771 (10th Cir. 1967) (emphasis added). The district court merely repeated this language. Paragon urges that the decision in Bledsoe was “misplaced” and “has not been followed or cited with approval by any subsequent Tenth Circuit case.” Aplt. Br. at 35. Paragon overlooks Donovan v. United Video, Inc., 725 F.2d 577, 584 (10th Cir. 1984), in which the court cited Bledsoe for its articulation of the Mt. Clemens test. Moreover, Bledsoe is still good law in this circuit and one panel of this court may not overrule another absent en banc consideration or an intervening Supreme Court decision. United States v. White, 782 F.3d 1118, 1126–27 (10th Cir. 2015). Neither of those conditions are present here.
9
Paragon contends that it carried its burden to negate the inferences arising from the Secretary’s evidence. It advances several theories under this heading, but most involve disagreements with how the Secretary filled in evidentiary gaps where information on claims forms was inadequate. The Supreme Court has said that “[t]he employer cannot be heard to complain that the damages lack the exactness and precision of measurement that would be possible had he kept records in accordance with the [FLSA].” Mt. Clemens, 328 U.S. at 688. The Secretary explained his methods of calculation in detail. See Aplee. Br. at 14–15. Claims forms with complete information on days and times worked were used to compute an average (here, the mode), which then filled in the gaps on forms where the information was incomplete or not disclosed. Id. The parties then agreed to several stipulations regarding the work hours, which necessitated adjustments. II Aplt. App. 313–14. Additional adjustments were made after the evidentiary hearing, including reducing hours to account for meal breaks. III Supp. App. 739.
Paragon objects to several features of the Secretary’s gap-filling analysis. It argues that the Secretary (1) used an unrepresentative sample to calculate averages, (2) used a mode instead of a mean, (3) included hours worked in 2013 despite Paragon’s claim that it did not participate in that year’s harvest, (4) failed to exclude non- compensable travel time, and (5) failed to exclude hours worked by 12 and 13-year olds outside school time that Paragon claims are exempt under the FLSA.
The first two arguments are unconvincing under the Mt. Clemens burden-shifting framework. Where a claimant provided concrete information on the claims form, the
10
Secretary used that to calculate unpaid wages. Aplee. Br. at 14. But where the information was missing, the Secretary calculated averages using the mode value from representative evidence and used those figures.4 Id. Paragon has shown merely that it could arrive at estimates of hours worked more favorable to itself were it in the Secretary’s position. That is both unsurprising and insufficient to negate the just and reasonable inferences arising from the Secretary’s analysis. Paragon has not shown that the Secretary’s methods were so deficient that inferences arising from its analysis cannot be trusted.5 The dissent concludes that DOL failed to carry its burden to establish a prima facie case under Mt. Clemens with respect to claimants who did not report the days or weeks they worked. It contends that the Secretary failed to provide “a reasonable explanation based on sound statistical analysis” for his estimate of days and weeks worked. We disagree. The dissent’s analysis improperly shifts the burden back to the Secretary. Under the Mt. Clemens framework, the Secretary’s burden is to show “the amount and
4 Confusion about exactly what constitutes an “average” — and what does not — arose in briefing and at oral argument. The “average” is “[a] familiar but elusive concept. Generally, an ‘average’ value purports to represent or to summarize the relevant features of a set of values; and in this sense the term would include the median and the mode. In a more limited sense an average compounds all the values of the set, e.g., in the case of the arithmetic or geometric means. In ordinary usage ‘the average’ is often understood to refer to the arithmetic mean.” Average, Oxford Dictionary of Statistical Terms (6th ed. 2003) (emphasis added). 5 At oral argument, Paragon appeared to suggest that the Secretary should have included claims forms that reported no data on hours or weeks worked in his calculation of averages, presumably as zeroes. Oral Argument at 7:28–42. We are not convinced that it would have been reasonable for the Secretary to construe a blank field on a form created to facilitate reporting of uncompensated work as reporting zero days or weeks worked.
11
extent of [the employee’s] work as a matter of just and reasonable inference.” The Secretary did that here by collecting claims forms and filling in the gaps on those that did not report specific information by using the mode value. Though the dissent questions the rigor of the statistical analysis, we must remember that Mt. Clemens seeks to prevent an employer from benefitting from its failure to keep records. See 328 U.S. at 688 (“The employer cannot be heard to complain that the damages lack the exactness and precision of measurement that would be possible had he kept records.”).6 The dissent makes an argument that Paragon never clearly raised and of course we cannot cross the line into advocacy. See Cordova v. Aragon, 569 F.3d 1183, 1191 (10th Cir. 2009) (“It is not our role to sift through the record to find evidence not cited by the parties to support arguments they have not made.”); Craven v. Univ. of Colo. Hosp. Auth., 260 F.3d 1218, 1226 (10th Cir. 2001) (“We will not manufacture arguments for an appellant, and a bare assertion does not preserve a claim.”) (quoting Entm’t Research Grp., Inc. v. Genesis Creative Grp., Inc., 122 F.3d 1211, 1217 (9th Cir. 1997)). The dissent’s analysis relies upon outside sources that Paragon did not discuss or even cite.7
6 We note, also, that the Secretary will not face the unique difficulty gathering evidence it had here in most cases. As shown by United Video and Simmons Petroleum, the government often includes deposition testimony as part of its prima facie case. Such testimony is favored because it provides the opportunity for parties to cross-examine witnesses and expose weaknesses in the government’s case. See Marshall v. Truman Arnold Distrib. Co., 640 F.2d 906, 911 (8th Cir. 1981) (concluding the defense rebutted the prima facie case with cross examination). 7 For example, the dissent cites three different statistical treatises. Paragon cited none in its briefing before this court.
12
While the dissent is correct that the law does not prohibit us from reaching outside the briefs, it has provided no persuasive reason for doing so given what Paragon actually argued and the Mt. Clemens framework.
The dissent claims that it is enough that Paragon generally “challenged DOL’s methodology as to the incomplete-reporting claimants,” and its calculation of the “universal average.” It also observes that the district court discussed and rejected Paragon’s appeal for the use of a weighted average. But the question is whether Paragon ever made an argument against the Secretary’s use of the mode, not whether it objected to his calculations generally. None of the dissent’s citations show Paragon claiming the mode was not an appropriate measure of central tendency for the dataset, which is the heart of the dissent’s argument. Indeed, the dissent observes that Paragon never even used the word “mode” in the district court. The district court would have needed a touch of clairvoyance to divine the argument the dissent now attributes to Paragon. The word “mode” was also missing from Paragon’s opening brief before this court. It appeared for the first time on a single page of Paragon’s reply brief. See Aplt. Reply Br. at 12. We do not think that a bare allegation that the Secretary’s analysis was “conclusory, unfounded and misguided” justifies this court supplying Paragon with a statistical argument it never made on its own behalf, either here or at the district court. Aplt. Br. at 33.
The dissent may be correct that, statistically speaking, the mode was not the best measure of central tendency to describe this particular set of data. Perhaps Paragon could have presented testimony from statistical or subject matter experts to argue that the mode was not an appropriate measure of central tendency to infer values for the missing data in
13
this set, or that the arithmetic mean was a more appropriate one.8 It did not. The district court was well within its discretion to decide this case the way it did based on the evidence and arguments advanced.
Paragon also failed to negate the inference that it was responsible for work performed on the Ranch in 2013, after the end of the last agreement with Paragon in 2012. Forty-nine claimants stated that they worked on the Ranch in 2013. III Aplt. App. 616. Paragon argues here, as it did in the district court, that it immediately halted its relationship with the Ranch at the expiration of the agreement in 2012. The district court rejected this argument as “speculative,” and found that Paragon had put forward “insufficient evidence” to support it. Id. at 617. This finding was not clearly erroneous — Paragon has offered nothing more than the agreement and its own assertion about its actions post-expiration.
Paragon argues that the district court should have reduced the total of hours worked for travel time. It relies on evidence that children sometimes gathered at a prayer meeting before traveling to the Ranch. Aplt. Br. at 22. Evidence of sporadic practice by some of the workers was not enough to rebut the Secretary’s contrary evidence.
8 Nor is it clear that such an argument could change the outcome on remand. The median responses for reported data are the same as the mode responses — 6 days per week and 2.5 to 3 months or more per year worked. See III Supp. App. 632. To compel the government to use the arithmetic mean, Paragon would presumably need to establish that it is the only average value appropriate for drawing inferences about missing data from reported data in this case.
14
C. Statutory Exemption under § 213(c)(1)(B)
Finally, Paragon argues that the district court should have excluded hours worked by 12 and 13-year olds outside of school. See 29 U.S.C. § 213(c)(1)(B) (providing an exemption for 12 and 13-year olds that are either (1) employed on the same farm as parents or working with parental consent, and (2) working outside school hours). To qualify for this exemption, employers must obtain written consent from a worker’s parents if they are not employed on the same farm. 29 C.F.R. § 570.2(b). Statutory exemptions from the FLSA are affirmative defenses and the employer bears the burden of establishing entitlement. Leerman v. Frontier Fire Prot., Inc., 685 F.3d 1151, 1156 (10th Cir. 2012).
Paragon adduced no evidence that it obtained written consent from the parents of workers allegedly entitled to the § 213(c)(1)(B) exemption. Paragon attempts to avoid the force of the written consent requirement by mischaracterizing the regulation as an “Informative Guideline” to which we owe no judicial deference. Aplt. Br. at 21. This is simply wrong. The regulation was promulgated through notice-and-comment rulemaking and must be analyzed under the traditional regulatory deference framework. See Employment of Minors Below Certain Ages, 39 Fed. Reg. 36,940–43 (Oct. 15, 1974) (proposed rule); Employment of Minors Under Certain Ages, 41 Fed. Reg. 26,834–36 (June 29, 1976) (final rule). Paragon has not properly challenged the Secretary’s authority to promulgate 29 C.F.R. § 570.2(b) or explained why the regulation does not apply in this case.
15
Paragon argues that the evidence shows that parents were employed on the Ranch at the same time as eligible child workers. Aplt. Br. at 19–20. We are not persuaded. The claims forms Paragon highlights also reveal that the claimants worked on the Ranch when they were between the ages of four and seven, too young to qualify for an exemption even if their parents were employed. III Aplt. App. 555–56. Paragon also directs us to testimony from the father of children who worked at the Ranch, in which he claims that “eight year olds could come as long as they were properly supervised with older children or with their parents.” Id. at 559. Again, this evidence tends to show that the children employed were too young to qualify for an exemption. Finally, Paragon points to photos of the Ranch showing adults working alongside children. But the record does not identify the adults in the photos, their relationship to the children, or the ages of the children. Paragon failed to establish its entitlement to a statutory exemption and the district court did not err by declining to deduct any hours on this basis.
AFFIRMED.
19-4097, Scalia v. Paragon Contractors Corp., et al. MATHESON, Circuit Judge, concurring and dissenting:
I concur with one exception. Although I join in affirming the district court’s back-wages award to the claimants who reported on their claim forms the days and weeks they worked, I would remand for further consideration of the appropriate compensatory award to the claimants who did not report the days or weeks they worked. The Department of Labor (“DOL”) did not provide “sufficient evidence to show the amount and extent of [their] work as a matter of just and reasonable inference.” Anderson v. Mt. Clemens Pottery Co., 328 U.S. 680, 687 (1946), superseded by statute on other grounds.
I. BACKGROUND
In 2007, the district court enjoined Defendants Paragon Contractors Corporation and Brian Jessop (“Paragon”) from using oppressive child labor in violation of the Fair Labor Standards Act. In 2015, the court held them in contempt because they employed minors on a pecan ranch from 2008 to 2013 in violation of the injunction. It ordered them to compensate the minors for the back wages they were owed. Because Paragon did not keep records of the minors’ work, the court ordered DOL to conduct a one-year claims process to determine the amount of back wages. After DOL completed that process, the court conducted an evidentiary hearing. Based on the evidence presented at the hearing, the court adopted DOL’s calculation of $1,012,960.90 in back wages.
DOL’s claims process invited any minor who had worked at the ranch to submit a form. The form asked for the information needed to calculate each individual’s compensation award, including the number of days worked per week and the number of
2
weeks worked per season. DOL received 104 eligible forms, but some of them were incomplete in two key respects.
First, 47 forms did not report the days worked per week. Supp. App. at 632; see App. at 369. Of the 57 claims that did, 37 reported 6 days;
8 reported 5 days;
5 reported 4.5 days;
6 reported 4 days; and
1 reported 3 days.
Supp. App. at 632.
Second, 21 forms did not report the weeks worked per year. Id.; see App. at 369.
Of the 83 claims that did, 43 reported “2.5 to 3 months or more;”
8 reported 1-2 months;
18 reported 3 weeks;
9 reported 1-2 weeks; and 5 reported “up to 1 week.”
Supp. App. at 632.
DOL’s only witness at the evidentiary hearing, Wage and Hour Division District Director Kevin Hunt, described his approach to the claims lacking day and week information. To calculate the proposed wages owed for those claims, he used the most
3
frequently reported days and weeks worked—the mode1—from the claim forms that included this information. Because 37 claims reported a 6-day work week and 43 claims reported numbers between “2.5 to 3 months or more” of work per season, id., Mr. Hunt attributed a 6-day work week and 13-week work season to the incomplete claims, App. at 369-70, 428. He called each of these numbers the “universal average.” Id. at 428.2 He therefore used the “universal average” (mode) of the reported numbers, rather than the mean3 or median,4 to estimate the amounts that the incomplete-reporting claimants worked. Aplee. Br. at 35.5 Mr. Hunt testified “this [was] the first time in [DOL] Wage Hour’s history that [it]
had a claim form process like this.” App. at 356. Even though the claims process lasted for more than one year, he did not follow up with the claimants who did not submit day
1 “The mode is the number that occurs most frequently in a set of data.” 1 Paul C.
Giannelli & Edward J. Imwinkelried, Scientific Evidence § 15.02[a] at 824 (5th ed. 2012) (“Giannelli & Imwinkelried”).
2 “In ordinary language, the arithmetic mean, the median, and the mode seem to be referred to interchangeably as ‘the average.’ In statistical parlance, however, the average is the arithmetic mean.” David H. Kaye. & David A. Freedman, in Michael J. Saks, et al., Reference Manual on Scientific Evidence 238 n.70 (3d ed. 2011).
3 “The mean is the arithmetic average. It is a computational figure.” Giannelli & Imwinkelried, § 15.02[a] at 824.
4 “The median is a positional figure. When the elements are arranged in order of numerical size, the median is the figure in the middle. Half the observations will lie above the median and half below.” Giannelli & Imwinkelried, § 15.02[a] at 824.
5 The mean, the median, and the mode are the three ways of measuring central tendency. Frederick J. Gravetter & Larry B. Wallnau, Statistics for the Behavioral Sciences 73 (9th ed. 2013). “The goal of central tendency is to find the single score that is most typical or most representative of the entire group.” Id.
and week information because he had “sufficient information” from the claimants who did. Id. at 370-71.6 He used the reported data’s mode because the mean “would have driven [his calculation] down significantly.” Id. at 479. The mean of the days worked per week was 5.46 days. The mean of the weeks worked per season was 7.7 weeks.
After Mr. Hunt testified about using the “universal average” (mode), the district court questioned whether DOL “ha[d] . . . met its [Mt. Clemens] burden” on the incomplete claims. Id. at 474. The court considered “disregarding all those [claims] . . . and telling the [DOL] to go back and start at zero with them or . . . to start from zero with all.” Id. Although DOL did not provide the court with any additional information or explanation concerning the incomplete claims, the court decided DOL had carried its burden as to all 104 claims without explaining why its concerns were alleviated.
In the findings of fact section of its memorandum and order, the district court said that Mr. Hunt’s “approach [wa]s consistent with Wage Hour’s internal policies and guidance, and case law . . . in back wage reconstruction cases” and that it “was reasonably calculated to identify a pattern and practice of hours worked at [the ranch] based on information reported by the claimants.” Id. at 614.7 It also stated “[h]is methodology was reliable and consistent with the evidence in this case.” Id.
6 Mr. Hunt did not attribute his lack of follow up to any interference from Paragon.
See App. at 370-71.
7 When the district court said Mr. Hunt’s analysis was “consistent with Wage Hour’s internal policies and guidance,” App. at 614, it was referring generally to his use of representative data from a sample of employees—not specifically to his use of the data’s mode. The court cited Mr. Hunt’s testimony to this effect. See id. (citing transcript as provided at App. at 356-57, 371, 477).
In the conclusions of law section of its memorandum and order, the district court said that DOL had met its burden under Mt. Clemens “to establish losses sustained by the minors as a result of Defendants’ contumacy by just and reasonable inference with evidence that is neither speculative nor arbitrary,” and that the award was “based on concrete information provided by 104 claimants.” Id. at 619.8 The court also pointed out that “where the court imposes a fine upon the contemnor for compensatory purposes, the amount must be based upon the actual losses sustained as a result of the contempt, and cannot be speculative or arbitrary.” Id. (citation omitted). The only reference to the incomplete forms was a statement that it was “reasonable” for Mr. Hunt “to utilize averages where necessary based on a pattern of violations.” Id. at 622.
II. ANALYSIS
I agree with the majority that DOL faced “unique difficulties” in obtaining information from potential claimants, that the reported claims likely “underrepresent the number of child laborers employed,” that “we can reasonably infer that many child laborers did not submit a [claim] form,” that “the government’s calculation may well be far less than the back wages owed,” and that we should not “reward[] Paragon for its failure to keep records,” especially when Paragon obstructed DOL’s ability to obtain information from potential claimants. Maj. Op. at 6-7. I also agree with the amount of the award to the claimants who filled out the days and weeks worked on their claim forms and agree that claimants who did not provide that information should receive back wages.
8 But as noted above, many of the 104 claim forms lacked information about the days or weeks worked.
My concern is limited to DOL’s failure to explain how it attributed the amount of time worked to the claim forms that lacked this information.
To satisfy its prima facie Mt. Clemens burden, DOL only had to “produce[]
sufficient evidence to show the amount and extent of th[e] [claimants’] work as a matter of just and reasonable inference.” Mt. Clemens, 328 U.S. at 687. As the majority notes, we review de novo Mr. Hunt’s methodology. FTC v. Kuykendall, 371 F.3d 745, 763 (10th Cir. 2004). DOL failed to meet its modest burden as to the claimants who did not report the days and weeks they worked.
Paragon challenged Mr. Hunt’s use of the most frequently reported days and weeks worked—i.e., the mode—to estimate the days and weeks worked by the incomplete-reporting claimants.9 Mr. Hunt failed to explain why it was reasonable to use
9 The majority incorrectly contends “Paragon never clearly raised” this argument.
Maj. Op. at 11. Although “[a] court should [not] raise sua sponte an argument not advanced by a party either before the district court or on appeal,” United States v. Abdenbi, 361 F.3d 1282, 1290 (10th Cir. 2004), Paragon made more than “a bare assertion” when challenging DOL’s methodology as to the incomplete-reporting claimants, both in district court and on appeal. See Craven v. Univ. of Colo. Hosp. Auth., 260 F.3d 1218, 1226 (10th Cir. 2001) (quotations omitted). Paragon repeatedly attacked DOL’s use of the most frequently reported days and weeks worked—Mr. Hunt’s “universal average”—which, by definition, is the mode.
In the district court, Paragon challenged whether DOL met its initial Mt. Clemens burden because Mr. Hunt’s “universal average” (mode) is not a “mathematical average,” App. at 430, but rather “the highest number [he] could justify,” id. at 395-96. In its order, the district court recognized that Paragon was objecting to “Mr. Hunt’s methodology” by arguing “that a weighted average [(mean)], rather than the universal average [(mode)] applied by Mr. Hunt, is the appropriate methodology” to calculate the days and weeks worked by the incomplete-reporting claimants. Id. at 614.
On appeal, Paragon references “universal average” (mode) at least 14 times. See Aplt. Br. at 10, 11, 12, 13 n.6, 14, 15, 16; Aplt. Reply Br. at 10, 11, 12, 13. It argues “Mr. Hunt’s conclusory, unfounded and misguided” analysis of the claim forms was inadequate to meet DOL’s prima facie Mt. Clemens burden, Aplt. Br. at 33, in part
7
the mode. His statement that the “universal average” (mode) would produce a higher value, see App. at 479, is mathematically true but does not provide a rationale for using it.
Paragon’s challenge to Mr. Hunt’s methodology is consistent with the recognized limitations of using the mode to draw inferences from data. Statisticians “rarely use[]” the mode to describe data because it “minimizes the number of [outliers], . . . no matter what their size.” David L. Faigman, et al., Modern Scientific Evidence: The Law and Science of Expert Testimony § 5:30 n.2 (2019-2020 ed.); see 1 Paul C. Giannelli & Edward J. Imwinkelried, Scientific Evidence § 15.02[a] at 824 (5th ed. 2012) (“To be blunt, the mode is a poor measure of central tendency” and “may be unrepresentative of the data as a whole.”); Herbert F. Weisberg, Central Tendency and Variability 27 (1992) (“The mode and median . . . do not take full advantage of the numeric information inherent in the data.”). “Generally the most effective way of summarizing the center of metric data is . . . the mean.” Weisberg, at 27; see Giannelli & Imwinkelried, § 15.02[a] at 825 (“For many purposes, the mean is the best and most important measure of central tendency . . . because it is the most representative.”).10
because his “‘universal averages’ are not in reality authentic averages” and are not based on any “clear consistent patterns” in the claims, id. at 10-12; see also id. at 12 (contending that both a “straight mathematical average” and a “weighted average” are “considerably less than the thirteen week assumption used by Mr. Hunt”).
10 The majority cautions against reliance “upon outside sources that Paragon did not discuss or even cite.” Maj. Op. at 11. But it offers no authority for its suggestion that a panel cannot supply additional source support for Paragon’s challenge to the methodology used for the incomplete-reporting claimants. “Courts frequently decide cases on lines of reasoning that can’t be found in the briefs,” Hampton v. Wyant, 296 F.3d
The drawbacks of using the mode to describe data can be seen here. Of the 83 claim forms that reported the number of weeks worked per season, 32 of them reported three weeks or less—ten weeks fewer than Mr. Hunt’s 13-week extrapolation. See Supp. App. at 632. Nor is it even clear that 13 weeks was the mode of the reported data for weeks worked per season. Although 43 of the claim forms reported “2.5 to 3 months or more,” id., that information does not tell us whether 13 weeks was the most frequently reported number—the definition of mode. For example, 2.5 months is only 11 weeks.
Under Mt. Clemens, it was the district court’s “duty,” as the trier of fact, “to draw whatever reasonable inferences c[ould] be drawn from [DOL’s] evidence” when determining whether DOL met its prima facie burden. Mt. Clemens, 328 U.S. at 693. But neither the majority nor the district court explains why it was reasonable to draw inferences based on the reported data’s mode. They rely on Donovan v. Simmons Petroleum Corporation, 725 F.2d 83, 85-86 (10th Cir. 1983), in which we held an employee may satisfy the prima facie Mt. Clemens burden with evidence from a sample of employees. But neither Simmons nor any of the other cases cited by the district court support the use of a mode to draw inferences from an employee sample.
560, 565 (7th Cir. 2002), and conduct independent research of arguments presented, see, e.g., Cent. Kan. Credit Union v. Mut. Guar. Corp., 102 F.3d 1097, 1108 (10th Cir. 1996) (“Although the question is close, our own research reveals more support for the proposition.”); Taylor v. Denver & Rio Grande W. R.R. Co., 438 F.2d 351, 353 (10th Cir. 1971) (The “appellee has cited no contrary authority, and we have found through our own research that most courts that have dealt with the question accord with [the appellant’s position].”).
As the panel agrees, Paragon should not benefit from its failure to keep records of the minors who worked on the ranch.11 To account for an employer’s lack of recordkeeping, the Supreme Court in Mt. Clemens adopted a forgiving standard that eases the employee’s burden to prove uncompensated work. DOL, on behalf of the minors, faced a difficult task to meet this modest burden. But we should not weaken the standard further by forgiving inadequate analysis of the data DOL was able to obtain.
DOL met its prima facie burden as to the minors who reported the amounts of time they worked. Through reasonable statistical inference, DOL may have been able to meet its burden as to the minors who did not supply that information. But it did not.12 After hearing Mr. Hunt’s description of how he used the “universal average” (mode), the district court expressed concerns about this shortcoming, concerns that remain valid.
11 The majority argues that the “unique difficulties” Paragon faced in obtaining information from the minors employed at the ranch support finding DOL met its Mt. Clemens burden. Maj. Op. at 6-7. The evidence showing Paragon coached the minors to “keep . . . away from [DOL] investigators,” App. at 625, is troubling, and may excuse DOL of the usual evidence-gathering expectations. But the concern addressed here is DOL’s inadequate analysis and explanation of the incomplete forms it had in hand.
The district court tasked DOL with calculating a compensatory award to reimburse children who “submit a claim . . . for compensatory payments based on the number of hours worked.” Id. at 281 [Sanctions Order] (emphasis added). DOL cannot support its calculation for the incomplete claim forms based on back wages potentially due to minors who did not submit claims at all. See O’Connor v. Midwest Pipe Fabrications, Inc., 972 F.2d 1204, 1211 (10th Cir. 1992) (noting a compensatory sanction “must be based upon the complainant’s actual losses sustained” (quotations omitted)). The incomplete- reporting claimants are not entitled to the “highest number” of hours worked as a compensatory award merely because others did not come forward. See App. at 396.
12 The majority contends this analysis “shifts the burden back to [DOL].” Maj.
Op. at 10. But once DOL had the 104 claim forms in hand and found some of them to be incomplete, it needed to provide a reasonable explanation for how it filled in the gaps to meet its modest Mt. Clemens burden.
Although Mr. Hunt’s wish to use representative data to infer the days and weeks worked by incomplete-reporting claimants could potentially satisfy Mt. Clemens, see, e.g., Donovan v. Williams Oil Co., 717 F.2d 503, 505-06 (10th Cir. 1983),13 he did not show why his method of inference was reasonable here. As a matter of its initial Mt. Clemens burden, and given the mode’s recognized limitations, DOL needed to provide a reasonable explanation based on sound statistical analysis to attribute the maximum number of days worked per week and weeks worked per season to the incomplete- reporting claimants.14 The district court erred in failing to hold DOL to its modest Mt. Clemens burden.
DOL may be able to support its recommended award to the incomplete-reporting claimants, but it should be required to do so with a better showing on remand to the district court. See Marshall v. Truman Arnold Distrib. Co., 640 F.2d 906, 911 (8th Cir. 1981) (remanding for “a further evidentiary hearing on the amount and extent of the
13 In Donovan v. Williams Oil Company, we said that 19 employees’ testimony, “coupled with . . . documentary evidence” such as the employer’s sales reports, was sufficient to support a damage award to 15 employees who did not testify. 717 F.2d at 506.
14 Expert testimony from both sides may have “help[ed] the trier of fact to understand the evidence or to determine a fact in issue.” Fed. R. Evid. 702(a).
work performed” under Mt. Clemens).15
15 The majority suggests the outcome might be the same on remand. Maj. Op. at 13 n.8. It might. But DOL failed to support its attribution of work performed to the incomplete-reporting claimants. The majority also suggests the median might produce the same values as the mode—“6 days per week and 2.5 to 3 months or more.” Id. But DOL did not argue the median should be used. Moreover, “2.5 to 3 months or more” is a range, not a value, and it is not clear from any record evidence that the median (or the mode) would be 13 weeks—the value Mr. Hunt applied for the weeks worked.