Maldonado v. Epsilon Plastics, Inc., 232 Cal. Rptr. 3d 461, 22 Cal. App. 5th 1308 (2018)

April 18, 2018 · Court of Appeal, Fifth District, California · B278022; B281129
232 Cal. Rptr. 3d 461, 22 Cal. App. 5th 1308

Olvin MALDONADO et al., Plaintiffs and Respondents,
v.
EPSILON PLASTICS, INC., Defendant and Appellant.

B278022
B281129

Court of Appeal, Second District, Division 8, California.

Filed April 18, 2018

Stuart Kane, Robert J. Kane, Peter L. Wucetich, Newport Beach, and Shane P. Criqui, for Defendant and Appellant.

Abrolat Law, Nancy L. Abrolat and Shahane A. Martirosyan, Redondo Beach, for Plaintiffs and Respondents.

RUBIN, J.

*1312Plaintiff employees were successful in a wage and hour class action against defendant and appellant Epsilon Plastics. Specifically, at four different times, Epsilon employees worked on a 12-hour/day schedule, under which they were paid for 10 hours at the regular rate of pay and 2 hours of overtime. This Alternative Workweek Schedule (AWS) would have been permissible if it had been adopted in accordance with the rules set forth in the applicable wage order. However, the trial court concluded, after a bench trial, that the AWS had not been properly adopted. The court further concluded that Epsilon's failure to pay overtime for the ninth and tenth hours of work, in reliance on the improperly adopted AWS, was not in good faith. As a result of the improperly adopted AWS, plaintiffs obtained judgment for unpaid overtime, interest, waiting time penalties ( Lab. Code, § 203 ), inaccurate wage statement penalties ( Lab. Code, § 226 ), and attorney's fees.1

Epsilon appeals, arguing: (1) the evidence does not support the trial court's conclusion that the AWS was improperly adopted in one of the four periods; (2) the evidence does not support the full award of damages for unpaid overtime; (3) the evidence does not support the trial court's conclusion of lack of good faith for two of the four periods, undermining the award of waiting time penalties; (4) the evidence does not support the award of waiting time penalties for certain former employees; (5) the wage statement penalties must be reversed because plaintiffs suffered no injury; (6) the attorney's fee award was untimely sought; and (7) the attorney's fee award incorporated a multiplier that was not supported by the evidence. We agree with Epsilon only in two respects: the evidence does not support the full award of damages for unpaid overtime; and the wage statement penalties must be reversed. We therefore affirm in part, reverse in part, and remand for recalculation of damages, and reconsideration of the attorney's fee award.

FACTUAL AND PROCEDURAL BACKGROUND

Because of the complexity of the issues and the variety of mathematical calculations for different time periods, we spend considerable time reciting the facts and procedural history through and including the court's final statement of decision.

1. The Plant and the Plaintiffs

Epsilon manufactures plastic bags. The manufacture requires the operation of one or more lines of machines which are designed to operate 24 hours per *1313day. Whenever the machines are shut down, it takes up to six hours to restart them. This process creates a lot of wasted plastic, and excessive wear and tear on the machines. For this reason, Epsilon strongly preferred *464to run its plant 24 hours a day, seven days a week.

Plaintiffs are production employees who operate the machines. Plaintiffs are largely Spanish-speaking, and many are uneducated.

2. The Two Schedules Used by Epsilon

As a general rule, overtime pay is required for each hour in excess of 8 hours in one day, or 40 hours in one week. (§ 510, subd. (a).) Epsilon could have run the plant full time with four shifts of employees working 8-hour shifts, with minimal overtime.2 But Epsilon did not run the plant in that fashion.

A. The 10/2 AWS

Instead, Epsilon's employees each worked 12-hour shifts-four shifts in one week and three in the next.3 If no AWS had been adopted, Epsilon could operate its plant in this manner, but would be required to pay its employees overtime for each hour in excess of 8 hours each day. In other words, the employees would be paid regular time for the first 8 hours of each 12-hour shift, and overtime for the last 4 hours. However, Epsilon used an AWS, under which the employees were paid regular time for the first 10 hours, and overtime for the last 2 (the 10/2 AWS). In weeks that an employee worked 48 hours (4 shifts), the employee would receive 40 hours of regular pay and 8 hours of overtime; in weeks that the employee worked 36 hours (3 shifts), the employee would receive 30 hours of regular pay and 6 hours of overtime.

Under the 10/2 AWS, Epsilon also agreed to give its employees a half-hour paid meal break. As a result, the employees were paid for the full 12 hours of each shift, even though they only worked 11.5 hours. A dispute regarding overtime pay for these meal breaks would ultimately become the main damages issue at trial.

B. The Ten Day/Eight-Hour Schedule

At times, Epsilon did not have enough orders to justify operating the plant 24 hours a day, 7 days a week. However, it was keenly aware of the *1314problems caused by continually starting and stopping its machines. Therefore, when it could not operate the plant full time, it adopted a schedule of 24 hours a day for 10 days straight, then closed down for 4 days, before restarting for another 10 days, and so forth. Epsilon put its employees into three 8-hour shifts, and had each shift work for 8 hours, for 10 days straight. Epsilon paid no overtime, because it structured its workweek such that employees were working the last five days of one week and the first five days of the next, thereby never exceeding 40 hours in one week.

It can come as no surprise that Epsilon's employees vastly preferred the 10/2 AWS to the Ten Day/Eight-Hour schedule. On the 10/2 AWS, they worked fewer days and received greater pay (both for additional hours worked and for overtime). In contrast, the Ten Day/Eight-Hour schedule required them to commute to work more often, and work ten days in a row, for less money. However, although Epsilon's employees testified that they preferred the 10/2 AWS to the Ten Day/Eight-Hour schedule, several of them also testified that, had they been offered the option of a 12-hour schedule with 4 hours *465of overtime, they would have preferred that. It does not appear that were ever given this option, nor were they given the option of running the plant full time with 8-hour schedules and minimal overtime.

3. The Governing Authority for Adoption of an AWS

Before we address the circumstances in which Epsilon adopted the AWS each of the four times it did so, we provide an overview of the legal requirements for an AWS.

Overtime compensation is required to be paid for any work in excess of eight hours in one workday unless an exception applies. One such exception is an "alternative workweek schedule adopted pursuant to Section 511." (§ 510, subd. (a)(1).) Section 511 in turn, provides that an employee may adopt an AWS only if it receives approval in a secret ballot election by at least two-thirds of the affected employees. Specific requirements for the adoption of an AWS are then set forth in the applicable wage orders. For the manufacturing industry, we are concerned with Industrial Welfare Commission wage order 1-2001. ( Cal. Code Regs., tit. 8, § 11010.) This wage order permits an AWS which has up to 10 hours of work per day at regular pay with up to 2 additional hours to be paid at the overtime rate. ( Cal. Code Regs., tit. 8, § 11010, subd. 3(B)(1) ; see Mitchell v. Yoplait (2004) 122 Cal.App.4th Supp. 8, 12, 19 Cal.Rptr.3d 267.)

The wage order provides detailed requirements for the adoption of such an AWS:

- it shall begin with a proposal "in the form of a written agreement proposed by the employer" ( Cal. Code Regs., tit. 8, § 11010, subd. 3(C)(1) );
*1315- in "order to be valid, the proposed alternative workweek schedule must be adopted in a secret ballot election, before the performance of work, by at least a two-thirds (2/3) vote of the affected employees in the work unit. The election shall be held during regular working hours at the employees' work site" (id . at subd. 3(C)(2) );
- prior to the vote, the employer "shall have made a disclosure in writing to the affected employees, including the effects of the proposed arrangement on the employees' wages, hours, and benefits. Such a disclosure shall include meeting(s), duly noticed, held at least 14 days prior to voting, for the specific purpose of discussing the effects of the alternative workweek schedule" (id . at subd. 3(C)(3) );
- the results of the election shall be reported by the employer to the Division of Labor Statistics and Research within 30 days (id . at subd. 3(C)(6) );
- employees affected by the AWS "may not be required to work those new hours for at least 30 days after the announcement of the final results of the election" (id . at subd. 3(C)(7) ); and
- the employer "shall not intimidate or coerce employees to vote either in support of or in opposition to a proposed alternative workweek" (id . at subd. 3(C)(8) ).

4. The Four Challenged Periods in Which Epsilon Used the 10/2 AWS

The present lawsuit encompasses four periods in which Epsilon used the 10/2 AWS. The trial court found that Epsilon's adoption of the 10/2 AWS was faulty, and not in good faith, each time. As Epsilon does not challenge these findings with respect to the later two times, we focus the bulk of our discussion on the first two periods.

A. The First Period: April 26, 2007-October 13, 2008

Although the first challenged period began in April 2007, Epsilon did not actually *466adopt the 10/2 AWS at that time. Instead, April 2007 was simply as far back as the statute of limitations allowed plaintiffs to reach. The 10/2 AWS in effect in 2007 had been in operation when Epsilon acquired the plant in 2002. Epsilon's predecessor was Apple Plastics; Epsilon purchased Apple's *1316assets when the latter filed for bankruptcy. The issue then, turned not on whether Epsilon had properly adopted the 10/2 AWS, but whether Apple had.4

The evidence at trial regarding Apple's adoption of the 10/2 AWS was minimal. There was certainly evidence that the 10/2 AWS was in effect at Apple. Several workers remembered it, and one testified to having voted at Apple to keep the schedule, but there was no testimony as to a vote at Apple initially adopting the 10/2 AWS. Indeed, Epsilon's controller, Jim Gifford-who had worked in the same position for Apple since 1993-testified that although he recalls Apple employees voting on the AWS, he was unable to identify a time when the employees voted on it before they actually worked on that schedule.

Instead, Epsilon relied on two documents found by Gifford. The first is a May 1, 1995 memo, found in an Apple file, directed to "Production Employees." It states, "We have been on the 12 hour workday schedule. Would like to know if you would like to continue working 12 hour shifts. We would like to have your opinion on this matter. Please read following information regarding the 12 hour work day schedule. Indicate on the bottom half of this memo if you are in favor or not in favor of the 12 hour work schedule." There follows a description of the 10/2 AWS, and a tear-off at the bottom of the memo on which the employee can vote.5 It states, "Special Note: The 12 Hour Work Day Schedule has been implemented since May 1993 and needs to be reviewed on an annual basis and evaluated to ensure that the program is working properly."

The second is a December 15, 1999 letter, on Apple letterhead, addressed to the Industrial Welfare Commission, requesting "an exemption to continue with the current alternative work schedule that is currently in place." After explaining the 10/2 AWS, it states, "We have had this schedule since 1993 and have surveyed employees on May 1995 and May 1996 as to whether they would like to continue with the 12 hours shift as described above. More than 2/3 of the employees voted to continue with the 12 hours work schedule." The letter closes with, "Again, I would like to respectfully request the exemption to continue with our alternate work schedule. Respectfully, I await your reply on this matter." There was no evidence whether this letter was actually sent to, or received by, the Commission.

*1317The trial court would ultimately conclude that there was no evidence the 10/2 AWS was properly adopted for this initial period. Specifically, there was no evidence of a written disclosure, a meeting, voting, a 30-day waiting period, or a report to the state within 30 days.

At one point during this period, in January 2008, Epsilon conducted a revote to *467confirm its employees' agreement to the 10/2 AWS. Exhibits indicate that it was a secret ballot, preceded by a written memo, circulated to employees in Spanish, explaining the 10/2 AWS. An employee who recalls the vote does not remember any meeting prior to the vote. Additionally, the exhibits demonstrate that a supervisor took part in the vote (he voted yes), even though he was salaried and therefore not part of the work group subject to the 10/2 AWS.

The 10/2 AWS continued until October 2008. At that time, the recession reduced the plant's orders, and Epsilon was unable to keep operating 24 hours per day, 7 days a week. Instead, the plant switched to the Ten-Day/Eight-Hour schedule described earlier. The return to this schedule does not implicate AWS regulations.

B. The Second Period: October 11, 2009-November 23, 2009

In October 2009, orders picked up, and the plant was able to return to the full 24/7 schedule. Controller Gifford delegated to Human Resources administrator Marisol Mendoza the task of conducting the vote to return to the 10/2 AWS. Mendoza was given an order that "we had to move back to the 12-hour shift." It was her understanding that Epsilon had no choice; it had to move to the 12-hour shift in order to fulfill its orders.6

Mendoza was untrained on the AWS adoption procedure. She conducted her own research, by reviewing how a previous Human Resources administrator had handled it, and searching the Department of Labor's website for direction. She is certain that she had access to counsel if she needed it, but chose not to contact the lawyers, explaining, "I just went by what the-I think it was the Department of Labor asked for. It was pretty simple. Just follow those guidelines, and you know everything was done in good faith."

Mendoza prepared a memorandum, dated October 6, 2009, stating, "Due to our customer demands our plant will be moving from an (8) eight hour shift to a (12) twelve hour shift. This will be ongoing for approximately (1) one month or more should we get more orders in. To get an [sic ] employee input we are conducting this election. Below please indicate if you agree with the *131812) twelve hour shift schedule or if you disagree with it." The memo explained the terms of the 10/2 AWS, although it did not specifically state that, without adoption of the 10/2 AWS, overtime pay on a 12-hour shift would begin after the first 8 hours.

After preparing the memorandum, Mendoza had meetings with each shift of employees on October 6, 2009. During the meetings, she explained that Epsilon "will be going to the 12-hour schedule" and that employees would be paid 10 hours at regular time and 2 hours overtime. The employees voted on the 10/2 AWS at the meetings.

The employees' vote was in favor of the 10/2 AWS. Several employees testified that they were told that, although they were asked to vote, they were told to vote yes "because it was going to happen anyway." Indeed, one employee testified that he was told by the plant manager that if he did not agree with the AWS, he could leave at any time, as there were a lot of people outside who wanted work. Additionally, the salaried supervisor who should not have voted in October 2008 voted again in October 2009.

The 10/2 AWS went into effect on October 12, 2009, six days after the vote. On October 16, 2009, Mendoza wrote a letter *468to the Division of Labor Statistics and Research, in order to comply with the AWS requirements and inform the Division of what Epsilon "was trying to do." The letter explained that the employees voted unanimously on October 6, 2009 to adopt the 10/2 AWS, and that the schedule went into effect on October 12, 2009.

This adoption of the 10/2 AWS was short-lived. The plant returned to the Ten-Day/Eight-Hour schedule in late November or early December 2009. Again, this schedule does not implicate AWS rules.

C. The Third Period: October 18, 2010-October 31, 2010

The plant briefly returned to the 10/2 AWS for two weeks in October 2010 with no vote whatsoever. Mendoza conceded that she did nothing to implement the 10/2 AWS during this period. She testified, however, that she thought Epsilon was "okay in doing it" because the employees preferred the 10/2 AWS to the Ten-Day/Eight-Hour schedule.

D. The Fourth Period: May 22, 2011 to March 31, 2013

Mendoza did conduct a vote related to Epsilon's final adoption of the 10/2 AWS in May 2011. However, as her May 31, 2011 letter to the Division *1319indicates, the 10/2 AWS took effect on May 23, 2011, but the employees did not vote on it until May 31, 2011.7

When asked at trial whether a vote after the AWS was implemented complied with the rules for an AWS, Mendoza testified, "At that time, since this was already something implemented in the past, um, some things may have changed, um, because the process was, you know, similar, the same, and I could have mentioned it, like I said last time on my safety meetings, which are conducted monthly. And because it's so-it was, like, kind of, like, a popular schedule and a lot of people wanted it, we kind of went that way."

5. The Complaint

On April 26, 2011, plaintiffs Olvin Maldonado and Manuel Cobian Hernandez, individually and on behalf of the class of hourly employees of Epsilon, brought suit against Epsilon for violation of wage and hour laws and violation of the unfair competition law ( Bus. & Prof. Code, § 17200 ).8 The complaint alleged numerous wage and hour violations, not all of which plaintiffs pursued at trial. The unfair competition law cause of action was based on the same alleged wage and hour violations.

6. Class Certification

In December 2012, plaintiffs moved to certify a class with respect to a more limited set of claims: (1) failure to pay overtime for hours 9 and 10 under the 10/2 AWS each time it was implemented; (2) failure to pay overtime for work more than 40 hours in a week (and failure to give one day of rest in seven) under the Ten-Day/Eight-Hour schedule; (3) denial of meal and rest breaks; (4) failure to provide accurate pay stubs, as the pay stubs did not reflect the overtime rate for all overtime hours actually worked; and (5) waiting time penalties for those former employees who were not paid all earned wages.

*469On September 30, 2013, the trial court granted the motion in part and denied it in part. The court granted certification with respect to the claims for overtime, inaccurate wage statements, waiting penalties, and related unlawful business practices. The court denied certification as to the claims for failure to provide meal and rest breaks. The court explained that certification of these *1320latter claims was denied because those claims required individualized fact-specific inquiries, such that individual issues predominated over common ones.9

7. Summary Adjudication of Ten-Day/Eight-Hour Schedule Claims

Epsilon next moved for, and obtained, summary adjudication of plaintiffs' claim that they were owed overtime for the Ten-Day/Eight-Hour Schedule. The court concluded that the schedule was legal, and that no additional overtime was owed. Plaintiffs do not challenge this ruling on appeal.

8. Court Trial-Phase One

The matter proceeded with a bifurcated trial, with the parties trying the equitable claim under the unfair competition law to the trial court first. The sole issue at this phase of the trial was whether Epsilon owed plaintiffs overtime for hours 9 and 10 when the 10/2 AWS was in effect or if, to the contrary, the payment of overtime for those hours was excused by the proper adoption of an AWS.

The court heard testimony from Mendoza and others at Epsilon regarding their adoption of the 10/2 AWS for the four relevant periods. Employees also testified as to their recollection of the procedures followed for adopting the 10/2 AWS.

Following written briefing, the court issued its statement of decision in favor of plaintiffs. The court concluded that, in all four periods, there was no validly implemented AWS. The court was specifically troubled by the fact that, throughout the entirety of the period, Epsilon "failed to inform the Employees that by agreeing to the [10/2 AWS], they were waiving overtime pay for hours nine and ten of the 12-hour shift that they would otherwise be entitled to." The court also noted specific problems with the adoption in each of the four periods. In the first period, there was no written disclosure, no meeting, no voting, no 30-day waiting period, and no report to the state within 30 days. In the second period, the meeting was the same day as the vote, not 14 days before; and the AWS was implemented 6 days, not 30 days later. For the third period, there was no vote at all or any other attempt to comply with the procedures. For the fourth period, the AWS had been implemented before the vote.

As the 10/2 AWS was never properly adopted, the court concluded the failure to pay overtime for hours 9 and 10 was not excused, and Epsilon *1321violated both the Labor Code and the unfair competition law. The case would therefore proceed to the second phase, where the parties would litigate the amount of unpaid overtime, as well as liability and damages for the waiting time and inaccurate wage statement claims.

9. Court Trial-Phase Two Testimony

The phase two trial was broadly concerned with three issues: (1) the calculation of damages owed for the failure to pay overtime for hours 9 and 10 under the 10/2 AWS; (2) whether Epsilon had a good faith *470defense to the failure to pay overtime, which would excuse it from waiting time penalties; and (3) whether Epsilon was also liable, and in what amount, for inaccurate wage statements.10 We discuss the evidence on each issue.

A. Damages for Uncompensated Overtime

As we have explained, under the 10/2 AWS, employees were paid regular time for 10 hours and overtime for 2 hours; this time included 1/2 hour for a meal break even though, as a general rule, an off-duty meal break need not be paid. ( Cal. Code Regs., tit. 8, § 11010, subd. 11(C).) Plaintiffs and Epsilon each presented expert testimony as to the amount of unpaid overtime owed to the class; the difference between them was simply how to handle the paid meal break which had been part of the 10/2 AWS.

Specifically, Epsilon took the position that, since its employees did not work during their lunch breaks, they actually did not work full 12-hour shifts, but only worked 11.5 hours. As they had already received 2 hours of overtime for each shift, their damages should be calculated as 1.5 hours of unpaid overtime per shift only. Notably, Epsilon did not seek to offset the half-hour it had paid for lunch against the damages owed. It simply had its expert, Robert Plante, make calculations on the basis that there should be no overtime calculated on the paid half-hour meal break. Under Epsilon's calculation, once damages were awarded, each employee's compensation for a 12-hour shift would be: 8 hours of regular pay, plus 3.5 hours of overtime, plus a .5 hour regular pay bonus for lunch.

Plaintiffs had hired two experts. Their first expert, Jim Skorheim, performed his calculations, but was then fired by plaintiffs, who sued him for fraud and extortion in connection with his billing. Their second expert, forensic accountant Henry John Kahrs, actually testified at trial. When Kahrs realized the only difference between Skorheim's initial calculations and *1322Plante's calculations was whether overtime was due for the half-hour meal break, Kahrs went into the punch clock data to determine the actual hours the employees had worked, and tried to figure out if they had, in fact, been paid for a 30-minute break.

Kahrs originally pulled "30 or 31 random punch cards and calculated what the time in, the time out, the total number of hours they worked." In that initial sample, which was actually 32 individual time cards, he concluded that only two employees were paid for the meal break, and the remainder were not. In addition, he concluded that, of the employees who were not paid for their meal break, 21 were underpaid by at least a minute, while nine others were overpaid (but not by the full 30 minutes which would constitute payment for the meal break). He did a second random sample of 19 and reached a similar conclusion. Kahrs extrapolated from this data, offsetting the overpayments against the underpayments, and concluding that, on average, each class member was underpaid by about 60 cents per shift. Plaintiffs did not seek to recover this amount; however, they relied on this testimony to bolster their conclusion that the employees were not paid for their meal breaks.

Epsilon pointed out three general problems with Kahrs's random sampling. First, plaintiffs questioned whether a sample as small as Kahrs's could have any statistical *471significance. Kahrs's sample consisted of no more than 51 line items of data, out of a total of 56,000 time sheet line items. This is less than one-tenth of 1 percent of the data. Kahrs testified that this tiny sample was nonetheless statistically significant, stating, "If you go through and you calculate a confidence interval, even if you ran a million pieces in your sample, if you took 35 random items, your confidence level of being correct would be roughly above 95 percent. That's why randomness is so important. [¶] The object isn't to go through and test hundreds and hundreds of thousands. The object is to take a small sample and take a look at it and see if it's representative." Kahrs was not himself a statistician, but testified that there was a statistician on his staff. Kahrs testified that he told someone in his office to randomly choose the records to constitute the sample with the statistician, saying, "Do it however is the most random and gives us a 95-percent confidence interval."

Second, it appears that several of the samples Kahrs selected were actually from 8-hour shifts, for which Epsilon had never agreed to pay for the meal break. Of his initial sample of 32, 10 of entries were from 8-hour shifts, and thus should be excluded from his data. Our review of his second sample of 19 shows another 10 entries were from 8-hour shifts. Kahrs conceded this would "drag down the average."

*1323Third, some of the samples he selected were from employees who were not members of the class. Only production line employees were part of the class; maintenance employees were not. Yet, of the 51 line items in his random sample, 26 came from maintenance employees. Faced with the suggestion that the majority of his data did not come from class members, Kahrs nonetheless believed that this would not call into question the accuracy of his analysis. He testified that the randomness of his sample established its accuracy. Eventually, he conceded that "In hindsight, I may choose a different sample and take a look at this," but he did not believe it was problematic to include non-class member data in his sample.

There appears to be another problem with Kahrs's analysis of his line items of data. By calculating what he believed to be a number of minutes over- or under-paid for an entire day, Kahrs went beyond the issue of whether the employees were paid for their meal breaks and instead offset paid meal break time against unpaid time-clock rounding time. This was problematic, as there was no class certified on the issue of improper time-clock rounding, and the only issue on which Kahrs should have been testifying was whether the meal breaks had actually been taken.

To counter Kahrs's conclusion that employees on the 10/2 AWS were not paid for their meal break, Epsilon presented various time sheet line items establishing that at least some of the employees were paid for some meal breaks. Kahrs responded that Epsilon's data was cherry-picked, not random, and stood by his conclusion that although some of the employees were, in fact, paid for meal breaks, "these people on the whole were not paid for their lunch." Kahrs also testified that he had his staff just randomly pick pages in the time card data, and the results confirmed that the employees were not paid for meal breaks.

At one point, Kahrs testified that if the class is not awarded the extra half-hour of overtime in dispute, the employees will be underpaid, but if the half-hour is awarded, they will be overpaid. This led defense counsel to ask, "Wouldn't the happy medium be maybe to split it in half, your damages number and Mr. Plante's number, and meeting in the middle on this issue?"

*472Kahrs responded, "Well, I suggested that at least 15 times during the course of this to plaintiffs' counsel, who tried to get them to stipulate to that so that I wouldn't have to come here. Yeah. That would have been brilliant."

B. Good Faith Defense

Section 203 provides that employees must be paid their wages upon discharge or quitting. If the wages are not paid in accordance with law, the *1324employee's wages "shall continue as a penalty" for up to 30 days. However, a "good faith dispute" will preclude the imposition of these waiting time penalties.

Once the court found overtime was owed under the 10/2 AWS, Epsilon would also be liable for waiting time penalties to its former employees, unless it could establish there was a good faith dispute. Epsilon therefore introduced testimony at the second phase of the trial on this point.11

Specifically, human resources administrator Mendoza testified that she had intended to comply with all AWS implementation requirements. She said, "I tried to do the best that I can based on the research; so yes. I was trying to comply with the law." She continued, "Everything was done in good faith, and I was trying to follow and comply with the law." Plant manager Tamayo Covarrubias testified he believed all iterations of the 10/2 AWS were valid, regardless of the failures to comply with the provisions set forth in the wage order, because all the employees preferred the 10/2 AWS. Controller Gifford testified that when Epsilon took over from Apple, it believed the 10/2 AWS had been properly implemented, although he conceded that, on behalf of Epsilon, he did nothing to confirm that fact beyond looking in the file and seeing the two documents on which he had relied in phase one of the trial.

C. Wage Statements

Plaintiffs also sought damages for inaccurate wage statements. The evidence introduced at trial on this point, however, was virtually nonexistent. A handful of wage statements for one employee were introduced into evidence, reflecting that he was paid under the 10/2 AWS. There was no testimony by any of the class members as to damages arising from the wage statements.

10. Final Statement of Decision

After additional briefing, the court issued a statement of decision, supporting an award of over $900,000 in favor of the plaintiff class. The court's statement of decision explained its ruling on the issues in controversy as follows:

A. Amount of Overtime Damages

Plaintiffs were entitled to a total of 4 hours of overtime for each 12-hour shift. As they had already received overtime for two of those hours, and *1325regular time for the other two, damages would be calculated as the overtime premium (difference between regular wage and overtime wage) for two hours. Notably, there would be no reduction for the meal period. The court reached this conclusion on the basis that most punches for meal breaks were for less than 30 minutes, and sometimes the employees did not punch out at all for a break. Moreover, the meal breaks were *473not duty free, as the employees were required to continue to monitor their machines. The court was persuaded by Kahrs's analysis that if 30 minutes of overtime premium were to be deducted, the employees would be underpaid, due to both time-clock rounding and the fact that employees rarely received a full 30-minute break. The court completely adopted Kahrs's analysis, specifically noting that Kahrs "utilized random samplings that provided a 95% confidence interval," and that "even if one were to remove the 8-hour shifts and the non-class members, still only one employee received a lunch of at least 30 minutes according to the punch details. Further, when confronted with this, Kahrs confirmed that it did not change the conclusion."

While the court declined to deduct 30 minutes from the overtime award, it noted that this was "not to be confused with Labor Code meal break violations which were not at issue during this trial for the Class, only for the individual Plaintiffs." Nonetheless, the court went on to conclude that Epsilon "failed to show that it uniformly gave the Employees a paid 30-minute, duty free lunch for each 12-hour shift worked to entitle it to reduce the overtime damages." Here, the court itemized evidence supporting this conclusion, including that there was no schedule to guarantee breaks; managers often summoned employees back to their machines when they were on break; employees could not leave the premises for breaks during the night shift; and there were not enough employees to provide coverage for the breaks.

B. Lack of Good Faith

Epsilon did not establish the good faith defense to the imposition of waiting time penalties. Epsilon failed to substantially comply with the AWS requirements. In addition to its obvious failures-such as voting after the AWS was put into effect on the fourth period, and not voting at all on the third period-the court specifically found other evidence of lack of good faith in the fact that Epsilon never actually informed the employees that by voting for the 10/2 AWS, they were giving up two hours of overtime to which they were entitled. Moreover, Epsilon offered the employees a paid meal break to entice them to vote for the 10/2 AWS. The court also found bad faith in Epsilon telling the employees how to vote, and allowing employees who were not subject to the 10/2 AWS to vote for its adoption.

*1326C. Wage Statement Penalties

Plaintiffs were entitled to wage statement penalties. The court's discussion that the wage statements caused the plaintiffs injuries was limited to the following: "The evidence presented at trial revealed that the paystubs that Epsilon provided to the Employees inaccurately listed overtime hours worked by the Employees at regular-time hourly rates, rather than the enhanced overtime rates as required by law, thereby causing injury, i.e., the Employees were paid regular rate instead of the enhanced overtime rate. Although Epsilon actually paid the Employees the regular rate of pay for hours 9 and 10, as shown on the paystub, this was inaccurate because the law required Epsilon pay overtime premium for hours 9 and 10."

11. Judgment

On July 26, 2016, the court entered judgment in favor of plaintiffs, consisting of $249,866.60 in unpaid overtime, $391,933.50 in waiting time penalties, $161,250 in wage statement penalties, and $132,247.30 in prejudgment interest on the unpaid overtime, for a total of $935,297.40.

*47412. Notice of Judgment

On August 10, 2016, plaintiffs served Epsilon a file-stamped copy of the judgment, giving notice that it had been entered.

13. Attorney's Fees

On October 11, 2016, plaintiffs sought their attorney's fees under section 218.5. Plaintiffs sought fees in a lodestar amount, enhanced by a multiplier of 1.7.

Epsilon opposed the motion on the basis that it was untimely, because the clerk allegedly served a file-endorsed copy of the judgment when it was entered on July 26, 2016. Epsilon also challenged the lodestar in some respects; and the multiplier.

Plaintiffs responded that the clerk's mailing was not sufficient to start the clock running for a motion for attorney's fees, and that the motion was timely when measured from plaintiffs' August 10, 2016 notice of judgment.

The court agreed with plaintiffs, finding the motion to be timely. It awarded fees in the amount of $888,811.50, based on a somewhat reduced lodestar, and a multiplier of 1.5.

*132714. Appeal

Epsilon filed timely notices of appeal from both the judgment and the award of attorney's fees. We have consolidated the two appeals for argument and resolution in a single opinion.

DISCUSSION

The several arguments on appeal are affected by various burdens of proof and our standard of review. We discuss each under the applicable argument.

1. Challenges to the Overtime Damages Based on Improper Adoption of the AWS

Plaintiffs were awarded damages for unpaid overtime due to the improper adoption of the 10/2 AWS in four separate periods. Defendant challenges both the finding of liability with respect to one of the four periods and the calculation of damages.

A. Substantial Evidence Supports the Finding of Liability

With respect to the first period the 10/2 AWS was in effect-the period based on Apple's earlier adoption of its 10/2 AWS-Epsilon argues that the AWS was properly adopted as a matter of law, based on undisputed evidence. Epsilon argues, "The AWS that was in place during this time period was first implemented in the early 1990s by Apple, and no evidence was presented that Apple failed to follow necessary procedures to implement this AWS." However, it was not plaintiffs' burden to establish that Apple failed to follow the procedures; it was Epsilon's burden to establish that it did.

We begin with the premise that section 510, subdivision (a) provides that any work in excess of 8 hours in one day shall be compensated at the overtime rate of time and a half. This does not apply, however, to an employee working pursuant to a properly adopted AWS. (§ 510, subd. (a)(1).)

"[T]he assertion of an exemption from the overtime laws is considered to be an affirmative defense, and therefore the employer bears the burden of proving the employee's exemption. [Citations.]" ( Ramirez v. Yosemite Water Co. (1999) 20 Cal.4th 785, 794-795, 85 Cal.Rptr.2d 844, 978 P.2d 2.)

*475Thus, Epsilon had the burden to establish the 10/2 AWS was properly adopted. Specifically, as to the first period, Epsilon had the burden to establish that Apple had properly adopted its AWS. Epsilon's evidence on this point was limited to documentary evidence indicating that the 10/2 AWS was *1328adopted in 1993 and confirmed by votes in 1995 and 1999. But while the two documents indicate the 10/2 AWS was implemented in 1993, neither document states that there was a pre-adoption vote, by secret ballot or otherwise. Controller Gifford, the only Epsilon witness who testified to working at Apple in 1993, conceded that he could not testify to a vote happening prior to the Apple employees working on the 10/2 AWS.

This does not establish as a matter of law that the 10/2 AWS was properly adopted. Instead, there was an absence of evidence that it was. Subsequent attempts (by Apple and Epsilon) to reaffirm the employees' commitment to the 10/2 AWS after it had been adopted do not meet the requirements of a pre-adoption vote, preceded by written notice and meetings.

Epsilon failed to meet its burden, and the trial court did not err in finding there was insufficient evidence that the 10/2 AWS was properly adopted. Our analysis would be the same whether we employed a de novo review based on uncontested facts (as suggested by Epsilon) or substantial evidence to the extent there were conflicting facts or conflicting inferences to be drawn from undisputed facts (as argued by plaintiffs).

B. Overtime Damages Were Miscalculated

As the 10/2 AWS was not properly adopted (in any of the four periods at issue), the trial court awarded overtime to the plaintiff class. The sole dispute at the second trial, regarding the damage calculation, was whether to award plaintiffs 1.5 hours of overtime premium for each shift or 2 hours of overtime premium for each shift. Epsilon conceded that the plaintiffs were entitled to 1.5 hours. The issue was simply whether they were entitled to the additional half-hour for the time which Epsilon allocated to the paid meal break.

Here, it is plaintiffs who had the burden of proof, but who wrongly argue the burden was on Epsilon. Plaintiffs were seeking an award of damages; they had the burden of proof. ( Evid. Code, § 500.) As a general rule, employees have the burden of proving that they performed work for which they were not compensated. ( Hernandez v. Mendoza (1988) 199 Cal.App.3d 721, 727, 245 Cal.Rptr. 36.) To be sure, employees need not prove this time with perfect exactitude when the employer's time records are inaccurate or inadequate. ( Ibid. ) But that is not this case, and plaintiffs do not argue that it is. Indeed, plaintiffs' evidence on this point was based on Kahrs's testimony, which was, in turn, based on Epsilon's detailed punch card records.

Instead, plaintiffs take the position that Epsilon bore the burden of proof because it was seeking a "reduction" in overtime damages for the meal break.

*1329Specifically, they argue that "in order for Epsilon to get this lunch reduction, Epsilon needed to prove that: 1) it gave the Employees with the 'opportunity' to take a 30-minute duty-free lunch; and 2) the lunch lasted a full 30 minutes; and 3) the lunch occurred before the fifth hour of work; and 4) the Employees were relieved of all duties, including leaving the facility at will; and 5) [Epsilon] did not discourage the Employees from taking the 30-minute lunch."

Plaintiffs here rely on section 512, the Labor Code provision governing meal breaks. But plaintiffs overlook the fact that *476this was not a cause of action under the Labor Code for the failure to provide statutorily-mandated meal breaks. Plaintiffs had alleged such a cause of action and the court had denied class certification on it. Instead, this was merely an element of damages for unpaid overtime. In short, the question was whether, under the 10/2 AWS, the employees worked 11.5 or 12 hours. The burden of proof that this additional half-hour was worked was squarely on plaintiffs.12

The misallocation of the burden of proof on this element drove plaintiffs, their expert Kahrs, and the trial court, to focus on the wrong issue. Plaintiffs believed they had defeated Epsilon's defense merely by establishing that not every employee took the full 30-minute meal break at every shift. But plaintiffs instead needed to establish that every class member worked for the entirety of the break during every shift, so that 30 minutes (or some specified lesser amount) could be added onto the overtime calculation. Perhaps this could have been established by a statistical analysis of the time card punch data. (Compare Duran v. U.S. Bank National Assn. (2014) 59 Cal.4th 1, 40, 172 Cal.Rptr.3d 371, 325 P.3d 916 [discussing the use of statistical sampling to prove damages in overtime cases] ( Duran ).) But because plaintiffs believed the burden of proof was on Epsilon, plaintiffs made no effort to affirmatively establish the amount of time that plaintiffs worked over and above the 11.5 hours defendant conceded (as the trial court had already found an invalid AWS).

This failure of proof is sufficient to undermine that portion of the trial court's award granting plaintiffs the additional half hour of overtime damages. However, we believe it necessary to briefly address three failings in the theory pursued by plaintiffs and supported by Kahrs's testimony.

First, Kahrs, as we have discussed, sought to determine whether the employees were paid for their meal breaks. He did so not by simply checking *1330whether the time cards reflected 30-minute meal breaks were taken. Instead, he added up the total amount of minutes the time cards reflected were worked. Finding that employees clocked in earlier and clocked out later than the times for which they were paid (due to time-clock rounding), he offset the "unpaid" minutes due to time-clock rounding against the "paid" minutes for meal breaks, and concluded that meal breaks were not, on the whole, compensated. But time-clock rounding itself may be permissible or impermissible; there is authority governing when a employer may use time-clock rounding. (See Silva v. See's Candy Shops, Inc. (2016) 7 Cal.App.5th 235, 239, 249, 212 Cal.Rptr.3d 514.) In this case, even if plaintiffs' complaint could be broadly interpreted to encompass a claim for improper time-clock rounding, plaintiffs did not seek class certification on it, and, most importantly, never tried the issue of whether Epsilon was liable for it. Kahrs's overtime calculations simply assumed that plaintiffs were entitled to be paid for every minute they were on the clock-effectively imposing liability on Epsilon for time-clock rounding in the course of a damages analysis for an improperly adopted AWS. This was improper.

Second, Kahrs's attempt to establish any conclusion at all with respect to the class, based on a random sample of 51 pieces of data, over half of which did not even pertain *477to class members, is the sort of "profoundly flawed" mock statistical analysis by a non-statistician rejected by the California Supreme Court in Duran . ( Duran, supra, 59 Cal.4th 1 at p. 13, 172 Cal.Rptr.3d 371, 325 P.3d 916.) As the Supreme Court explained: "Sampling is a methodology based on inferential statistics and probability theory. 'The essence of the science of inferential statistics is that one may confidently draw inferences about the whole from a representative sample of the whole.' [Citation.] Whether such inferences are supportable, however, depends on how representative the sample is. '[I]nferences from the part to the whole are justified only when the sample is representative.' [Citation.] Several considerations determine whether a sample is sufficiently representative to fairly support inferences about the underlying population." ( Id . at p. 38, 172 Cal.Rptr.3d 371, 325 P.3d 916.) It is enough to say Kahrs made no determination of variability, and simply stated that his random sample of less than 1 in 1000 was sufficient because it was random. This is inadequate as a matter of law.

Third, we reject Kahrs's attempt to cloak his testimony in the statistical respectability of a 95 percent "confidence interval" as not only unsupported, but missing key data. The 95 percent "confidence interval," as used by statisticians, is the "interval of values above and below the estimate within which one can be 95 percent certain of capturing the 'true' result." ( Duran, supra, 59 Cal.4th at p. 46, 172 Cal.Rptr.3d 371, 325 P.3d 916.) The interval is expressed in terms of the margin of error; and if the margin of error is too large, the result can be rejected as a matter of law. ( Ibid. ) Here, Kahrs testified to a 95 percent "confidence interval" in the conclusion that the employees were not, on the whole, paid *1331for meal breaks. This is not an interval at all, and not a statement in which, statistically speaking, one can have a 95 percent "confidence interval." No testimony was elicited on the margin of error.

Having rejected Kahrs's inferences, we are left with the question of whether plaintiffs established entitlement to any overtime, over and above the 1.5 hours per shift conceded by Epsilon. Kahrs's conclusions may be lacking, but his data demonstrates that some employees did not, in fact, get paid for every meal break. This anecdotal evidence, even if truly random, is not sufficient to justify an award of overtime for any particular amount of minutes. This is especially true given that Epsilon presented competing anecdotal evidence that some employees were paid for some meal breaks.

However, on appeal, Epsilon argues not for a reversal of the additional half hour of overtime, but for the court to split the difference and award the midpoint between 1.5 hours per shift and 2 hours per shift. As Epsilon concedes that this is an appropriate amount of damages, we will direct the judgment be reduced to that amount. On remand, the trial court shall make the calculation. Because the amount of overtime damages will change following recalculation, the court will also be required to recalculate the waiting time penalties and prejudgment interest.

2. Challenges to the Award of Waiting Time Penalties

Epsilon has two challenges to the award of waiting time penalties. First, it argues that it has established its good faith defense as a matter of law with respect to the first two periods in which it used the 10/2 AWS. Second, it argues that those waiting time penalties imposed for the period after March 31, 2013, were not supported by the evidence.

*478A. The Evidence Supports the Finding of Lack of Good Faith

"If an employer willfully fails to pay, without abatement or reduction, ... any wages of an employee who is discharged or who quits, the wages of the employee shall continue as a penalty from the due date thereof at the same rate until paid or until an action therefor is commenced; but the wages shall not continue for more than 30 days." ( § 203, subd. (a).)

"A willful failure to pay wages within the meaning of Labor Code Section 203 occurs when an employer intentionally fails to pay wages to an employee when those wages are due. However, a good faith dispute that any wages are due will preclude imposition of waiting time penalties under Section 203. [¶] (a) Good Faith Dispute. A 'good faith dispute' that any wages are due occurs when an employer presents a defense, based in law or fact which, if *1332successful, would preclude any recover[y] on the part of the employee. The fact that a defense is ultimately unsuccessful will not preclude a finding that a good faith dispute did exist. Defenses presented which, under all the circumstances, are unsupported by any evidence, are unreasonable, or are presented in bad faith, will preclude a finding of a 'good faith dispute.' " ( Cal. Code Regs., tit. 8, § 13520.)

This regulation "imposes an objective standard." ( FEI Enterprises, Inc. v. Yoon (2011) 194 Cal.App.4th 790, 802, 124 Cal.Rptr.3d 64.) "The appearance of the language 'or presented in bad faith' in the list of circumstances precluding a finding of a good faith dispute does not render the test a subjective one, but indicates that subjective bad faith may be of evidentiary value in the objective bad faith analysis." ( Id . at p. 802, fn. 9, 124 Cal.Rptr.3d 64.)

Although the trial in this case proceeded as though Epsilon had the burden to prove a good faith dispute, the law on this point is not entirely clear. Certainly, an argument could be made that since section 203 provides that waiting time penalties are awarded only for a "willful" failure to pay wages, lack of good faith is an element of willfulness on which plaintiffs had the burden of proof. (See, e.g., Armenta v. Osmose, Inc. (2005) 135 Cal.App.4th 314, 325-326, 37 Cal.Rptr.3d 460.) We need not resolve the burden of proof issue, however, as Epsilon's only argument on appeal is that it had established good faith as a matter of law, a point that we reject. The issue of good faith is, in actuality, an issue of fact reviewed for substantial evidence. ( Road Sprinkler Fitters Local Union No. 669 v. G & G Fire Sprinklers, Inc. (2002) 102 Cal.App.4th 765, 781, 125 Cal.Rptr.2d 804.)

Courts have found good faith disputes in a failure to pay wages when the legal duty to pay the wages was unclear at the time of the failure to pay. (E.g. Amaral v. Cintas Corp. No. 2 (2008) 163 Cal.App.4th 1157, 1201-1202, 78 Cal.Rptr.3d 572.) However, the existence of a bona fide legal dispute will not necessarily result in a finding of good faith when other evidence indicates the defendant knew it was not paying all wages due. ( Armenta v. Osmose, Inc., supra, 135 Cal.App.4th at pp. 325-326, 37 Cal.Rptr.3d 460.)

Epsilon challenges the trial court's findings that its failure to pay full overtime during the first two periods it used the 10/2 AWS was not the result of a good faith dispute. Epsilon argues that the issue may be resolved in its favor as a matter of law. But, there is legally sufficient evidence supporting the trial court's conclusion in favor of plaintiffs.

1. Sufficient Evidence of Lack of Good Faith in the First Period

Epsilon's argument that it had a good faith dispute that overtime was due *479under the first 10/2 AWS is based on Epsilon's evidence that it had a *1333subjective good faith belief that it had inherited a properly-adopted AWS from Apple. But, a mere subjective good faith belief that wages were not due is insufficient; the test is whether there was an objectively reasonable, even if unsuccessful, defense to the payment of wages. Here, there was no objectively reasonable factual basis for Epsilon's defense-it offered, for example, no evidence that Apple represented to it that the 10/2 AWS was properly adopted, and that it relied on that representation. Instead, the evidence showed that Epsilon made no inquiry whatsoever when it took over the plant, and simply assumed the 10/2 AWS had been properly adopted. This is sufficient to defeat Epsilon's claim of good faith.

2. Sufficient Evidence of Lack of Good Faith in the Second Period

Epsilon's argument that it had a good faith dispute that overtime was due under the second 10/2 AWS is based on its evidence that Mendoza attempted, in subjective good faith, to comply with the legal requirements for the adoption of an AWS, and that, although there were some errors in the process, Epsilon substantially complied. In short, Epsilon argues that "substantial compliance" was its defense with respect to the second period, and that although the defense did not win the day, it was sufficient to establish a good faith dispute as a matter of law.

There was a flaw in the adoption of the 10/2 AWS in the second period which could, conceivably, fall under the framework of "substantial compliance," were it the only error: the AWS was implemented 6 days after the vote, rather than 30. Harder to fit under that rubric is the fact that the employees voted on the AWS on the same day as written notice of the AWS and the meetings discussing it, rather than 14 days after. Even if this, too, were the sort of deficiency that could objectively constitute substantial compliance, there were other circumstances surrounding the adoption of the 10/2 AWS which defeat Epsilon's claim. Specifically, there was evidence that the employees were not given an opportunity to freely exercise their vote. Mendoza had been told that the plant "had no choice; it had to move to the 12-hour shift." The document she drafted proposing the 10/2 AWS said that the "plant will be moving from an (8) eight hour shift to a (12) 12-hour shift." The document asked the employees for their "input" on whether they agreed or disagreed with the schedule, but it did not indicate that an employee vote would actually govern whether the 10/2 AWS would be adopted. The employees were never told what the alternative would be if the plant needed to run 24/7 but the 10/2 AWS was not adopted. Several employees testified that they were told to vote yes, "because it was going to happen anyway." The trial court believed this evidence. Irrespective of which party had the burden of proof, this evidence defeats *1334Epsilon's claim that it proved good faith as a matter of law, and it constituted substantial evidence that plaintiffs had proved a lack of good faith under the statute.

B. Substantial Evidence Supports the Challenged Waiting Time Damages After March 31, 2013

As mentioned, plaintiffs initially hired Skorheim as their expert. They subsequently fired him and hired Kahrs. At trial, Kahrs testified to a small amount of waiting time penalties, incurred after March 31, 2013, based solely on Skorheim's calculations; Kahrs had not independently *480verified Skorheim's math.13 Epsilon's expert, Plante, also did not review any of Skorheim's calculations on this particular line item.

The trial court awarded the damages. On appeal, Epsilon argues this aspect of the award is unsupported by the evidence, because Skorheim did not testify to the calculation himself, and his credibility came into question when plaintiffs sued and replaced him.

The award is supported by Kahrs's testimony. While it is clear that plaintiffs disagreed with Skorheim in several respects (largely related to billing), neither party questioned his ability to calculate waiting time penalties from established data. Moreover, Plante, who had been retained by Epsilon to review Skorheim's calculations and point out disagreements with them, did not challenge Skorheim on this point. The evidence was admitted with no objections, and, although sparse, was legally sufficient

3. The Award of Wage Statement Penalties is Unsupported

Plaintiffs were awarded penalties for inaccurate wage statements. Specifically, they established that the wage statements were inaccurate because, whenever the plant was on the 10/2 AWS, the wage statements did not properly indicate the ninth and tenth hours were overtime. Plaintiffs did not plead, or argue, that the wage statements were inaccurate in any other particular.14

But inaccurate wage statements alone do not justify penalties; the plaintiffs must establish injury flowing from the inaccuracy. Here, the trial *1335court concluded the plaintiffs had suffered injury because they were not paid all of the overtime they were due.

On appeal, Epsilon argues that this is not sufficient to support the award, in that the failure to pay overtime flowed from the improperly adopted AWS, for which we affirm compensation, not from the inaccurate wage statements. Therefore, according to Epsilon, there is no injury as a matter of law. In response, plaintiffs pursue a new theory, not raised before the trial court: that injury is presumed under the statute. Because the issue presents primarily an issue of law, we exercise our discretion to consider it. ( In re J.C. (2017) 13 Cal.App.5th 1201, 1206, 221 Cal.Rptr.3d 579.)

Section 226, subdivision (a) itemizes nine categories of information which must be included in a wage statement. As relevant to this case, it provides: "(a) An employer, semimonthly or at the time of each payment of wages, shall furnish to his or her employee, either as a detachable part of the check, draft, or voucher paying the employee's wages, or separately if wages are paid by personal check or cash, an accurate itemized statement in writing showing (1) gross wages earned, (2) total hours worked by the employee ..., (5) net wages earned, ... and (9) all applicable hourly rates in effect during the pay period and the corresponding number of hours worked at each hourly rate by the employee ...."

*481Wage statement penalties are awarded only to employees who suffer injury "as a result of a knowing and intentional failure by an employer to comply with subdivision (a)." ( § 226, subd. (e)(1).) The statute provides, however, that an employee is deemed to suffer injury if the employer "fails to provide accurate and complete information as required by any one or more of items (1) to (9) inclusive, of subdivision (a) and the employee cannot promptly and easily determine from the wage statement alone one or more of the following: (i) The amount of gross wages or net wages paid to the employee during the pay period or any of the other information required to be provided on the itemized wage statement pursuant to items (2) to (4) inclusive, (6) and (9) of subdivision (a)." ( § 226, subd. (e)(2)(B) & (e)(2)(B)(i).) It further explains that "[f]or purposes of this paragraph, 'promptly and easily determine' means a reasonable person would be able to readily ascertain the information without reference to other documents or information." ( § 226, subd. (e)(2)(C).) In short, while the statute requires nine categories of information to be included in a wage statement, injury is only presumed if one of five specific categories is omitted, and, even then, only if a reasonable person would be unable to readily ascertain the missing information without reference to other documents or information.

Here, plaintiffs argue the missing information from which injury can be presumed is that of category (a)(9)-"all applicable hourly rates in effect *1336during the pay period and the corresponding number of hours worked at each hourly rate by the employee." The applicable hourly rates were included on the pay stub. Plaintiffs argue that the corresponding number of hours worked at each hourly rate was not because, in hindsight, the employees legally worked 8 hours at the regular time rate and 4 hours at the overtime rate (for each 12-hour shift), but the paystub indicated they worked 10 hours at the regular rate and 2 hours at the overtime rate.

Epsilon takes the commonsense position that the pay stubs were accurate in that they correctly reflected the hours worked and the pay received. Epsilon argues that if plaintiffs' argument were followed to its logical conclusion, the only way it could have avoided wage statement penalties while operating under the 10/2 AWS it believed was legitimate would have been to issue a wage statement which bore no similarity to the pay the employees were actually receiving. As it is illogical to think this is what the Legislature intended, plaintiffs' counter argument boils down to the proposition that any failure to pay overtime at the appropriate rate also generates a wage statement injury justifying the imposition of wage statement penalties-an apparent unintentional double recovery.15

We believe Epsilon has the better argument. We look at the statutory language. Subdivision (a) uses both the term "earned" and the term "worked." That is, categories (a)(1) and (a)(5) require the employer to provide information regarding the "gross wages earned" and "net wages earned" respectively; but these two categories are excluded from subdivision (e)(2)(B)(i)'s list of those categories whose omission gives rise to a presumption of injury. In contrast, categories (a)(2) and (a)(9) refer to the "total hours worked,"

*482and "number of hours worked at each hourly rate." These categories are included in subdivision (e)(2)(B)(i)-if they are excluded from the wage statement, injury may be presumed. There is a clearly a significance to the Legislature's decision that injury is not presumed when a wage statement fails to include wages "earned" but is presumed when the wage statement fails to include hours "worked at" a particular rate. The difference, we believe, is to account for precisely this situation-where at the time the work was performed, the work was done and paid for at a particular rate , but it was subsequently determined that the employee had actually earned the right to additional compensation. Wage statements should include the hours worked at each rate and the wages earned . In a perfect world, the first numbers will calculate out to the second. But when there is a *1337wage and hour violation, the hours worked will differ from what was truly earned. But only the absence of the hours worked will give rise to an inference of injury; the absence of accurate wages earned will be remedied by the violated wage and hour law itself, as is the case here.

This interpretation is supported by legislative intent. The purpose of section 226 is to "document the paid wages to ensure the employee is fully informed regarding the calculation of those wages." ( Soto v. Motel 6 Operating, L.P. (2016) 4 Cal.App.5th 385, 392, 208 Cal.Rptr.3d 618.) " ' "The purpose of requiring greater wage stub information is to insure that employees are adequately informed of compensation received and are not shortchanged by their employers" ' (quoting Assem. Com. on Labor and Employment, Analysis of Sen. Bill No. 1255 (2011-2012 Reg. Sess., italics added).)" ( Ibid. )

Here, Epsilon's plant was operating under the 10/2 AWS; Epsilon paid its employees pursuant to the 10/2 AWS; and its wage statements accurately reflected the pay under the 10/2 AWS. That the 10/2 AWS ultimately turned out to be invalid mandates that the employees receive unpaid overtime, interest, and attorney's fees. (§ 1194, subd. (a).) It does not mandate that they also receive penalties for the wage statements which accurately reflected their compensation under the rates at which they had worked at the time.

4. Arguments Related to Attorney's Fee Award

Plaintiffs were awarded nearly $900,000 in attorney's fees. On appeal, Epsilon contends the attorney's fee award must be reversed because it was untimely sought. In the alternative, Epsilon argues that the court abused its discretion in using a multiplier of 1.5 in calculating the award. We do not reach the latter argument. As we conclude the plaintiffs' recovery must be reduced in two respects (part of the overtime award, which impacts the waiting time penalties; and the wage statement penalties), we remand to enable the court to exercise its discretion to reconsider the amount of the fee award, should it so choose. However, we briefly address Epsilon's argument that no fees could properly be awarded because the motion seeking them was untimely filed.

A notice of motion for attorney's fees "must be served and filed within the time for filing a notice of appeal under rules 8.104 and 8.108." ( Cal. Rules of Court, rule 3.1702(b)(1).)

California Rules of Court, rule 8.104(a)(1) provides that, unless otherwise extended, the time for filing a notice of appeal is the earliest of: "(A) 60 days after the superior court clerk serves on the party filing the notice of appeal a *1338document entitled 'Notice of Entry' of judgment or a filed-endorsed copy of the judgment, showing *483the date either was served; (B) 60 days after the party filing the notice of appeal serves or is served by a party with a document entitled 'Notice of Entry' of judgment or a filed-endorsed copy of the judgment, accompanied by proof of service; or (C) 180 days after entry of judgment."

In this case, the signed file-stamped judgment is dated July 26, 2016. On August 10, 2016, plaintiffs served "Notice of Judgment" on Epsilon, attaching a file-stamped copy of the judgment. Sixty days from August 10, 2016 is October 9, 2016. That day is a Sunday. October 10, 2016 was Columbus Day, a state holiday. ( Gov. Code, § 6700, subd. (a)(12).) When "the last day for the performance of any act that is required by these rules to be performed within a specific period of time falls on a Saturday, Sunday, or other legal holiday, the period is extended to and includes the next day that is not a holiday." ( Cal. Rules of Court, rule 1.10(b).) The next court day was October 11, 2016. Plaintiffs' motion for attorney's fees was filed on that day. As long as the 60-day period is measured from August 10, 2016, plaintiffs' motion was timely.16

Epsilon argues, however, that the motion was untimely because it should be measured from the date of the clerk's certificate of mailing. According to a declaration of counsel, on July 26, 2016, the court sent a single mailing consisting of: (1) the court's one-page order indicating it signed the judgment; (2) the judgment itself; and (3) the clerk's certificate of mailing. This only starts the clock running if the clerk serves "a document entitled 'Notice of Entry' of judgment or a filed-endorsed copy of the judgment, showing the date either was served." ( Cal. Rules of Court, rule 8.104(a)(1)(A).) The clerk did not serve a document entitled "Notice of Entry"; the clerk's document was, at most, a certificate of mailing. Moreover, the document did not indicate the date it was served. Nor did the clerk serve a file-endorsed copy of the judgment showing the date it was served. The file-endorsed copy of the judgment which Epsilon's counsel represented he received in the mailing contains no proof of service beyond the original proof of service from when plaintiffs served it as a proposed judgment. We also observe that the trial court, in its attorney's fee order, found that the judgment was not a part of the mailing, stating that the exhibit attached to counsel's declaration, "does not comport with the document on file, which consists of only two pages and does not include a copy of the judgment." In sum, the July 26, 2016 mailing did not meet the requirements of the court rule sufficient to start the 60-day clock running. The attorney's fees motion was therefore timely.

*1339DISPOSITION

The case is remanded to the trial court with directions to: (1) reduce the award of overtime damages to provide the overtime premium of 1 hour 45 minutes per shift, rather than 2 hours per shift; (2) recalculate the waiting time penalties and interest accordingly; and (3) eliminate the award for wage statement penalties. The trial court's order awarding attorney's fees is vacated to permit the trial court to reconsider attorney's fees following remand. In all other respects the judgment is affirmed. The parties are to bear their own costs on appeal.

WE CONCUR:

BIGELOW, P.J.

GRIMES, J.