Opinion
B272353
12-13-2017
CHRISTOPHER WILLIAMS, Plaintiff and Appellant, v. ALLSTATE INSURANCE COMPANY, Defendant and Respondent.
Law Offices of Kevin T. Barnes, Kevin T. Barnes and Gregg Lander; Trush Law Office and James M. Trush for Plaintiff and Appellant. Seyfarth Shaw, Andrew M. Paley, James M. Harris, Sheryl L. Skibbe and Kiran A. Seldon for Defendant and Respondent.
NOT TO BE PUBLISHED IN THE OFFICIAL REPORTS
California Rules of Court, rule 8.1115(a), prohibits courts and parties from citing or relying on opinions not certified for publication or ordered published, except as specified by rule 8.1115(b). This opinion has not been certified for publication or ordered published for purposes of rule 8.1115. (Los Angeles County Super. Ct. No. BC382577) APPEAL from a judgment of the Superior Court of Los Angeles County. John Shepard Wiley, Jr., Judge. Reversed in part and affirmed in part. Law Offices of Kevin T. Barnes, Kevin T. Barnes and Gregg Lander; Trush Law Office and James M. Trush for Plaintiff and Appellant. Seyfarth Shaw, Andrew M. Paley, James M. Harris, Sheryl L. Skibbe and Kiran A. Seldon for Defendant and Respondent.
____________________
Plaintiff Christopher Williams brought a wage and hour class action against his employer, Allstate Insurance Company, on behalf of a class of auto field adjusters. The class was initially certified, then decertified after the trial court found individual issues predominated. Williams successfully sought writ review in this court, and we directed the class to be recertified. The trial court did so, and subsequently decertified the class again - this time, on the basis that Williams had failed to present a viable trial plan for the class action. Williams again appeals. We reverse in part due to intervening authority and otherwise affirm.
FACTUAL AND PROCEDURAL BACKGROUND
1. Underlying Facts
Much of our discussion of the underlying facts and initial procedural background is drawn from our opinion in the prior writ proceeding, Williams v. Superior Court (2013) 221 Cal.App.4th 1353 (Williams I).
Allstate "employs several hundred auto field adjusters in California. Auto field adjusters travel to sites such as body shops to inspect, and analyze the value of, damaged vehicles. In 2005, Allstate changed the classification of its auto field adjusters from salaried employees to hourly employees in response to litigation challenging their misclassification as employees exempt from the protection of overtime wage laws. [Citation.]" (Williams I, supra, 221 Cal.App.4th at p. 1356.)
A. Off the Clock Work
"Auto field adjusters receive their daily work schedules of vehicle inspection appointments by logging onto Allstate's 'Work Force Management System' software loaded onto their work laptops. Although the adjusters are hourly employees entitled to overtime if they work more than eight hours a day or 40 hours a week, the Work Force Management System software is not a time recordkeeping program, nor does Allstate maintain any other timeclock system." (Williams I, supra, 221 Cal.App.4th at p. 1357.) "Rather than track the actual hours an adjuster works, the Work Force Management System instead presumes each adjuster's eight-hour workday begins when the adjuster arrives at his or her first vehicle-inspection appointment of the day." (Ibid.) "Allstate's presumption that an adjuster's workday begins with the first appointment as set by the Work Force Management System does not take into account any work the adjuster may have performed before the day's first appointment." (Ibid.)
"Belying Allstate's assumption of an eight-hour workday, [Williams] submitted declarations from numerous adjusters stating they typically worked more than eight hours a day and 40 hours a week after Allstate reclassified them from salaried to hourly employees. Among the overtime tasks those adjusters declared they performed outside their eight-hour shifts were (1) logging onto their work computers, (2) downloading their assignments, (3) making courtesy calls to auto repair shops and car owners to confirm appointments, (4) checking their voice mail, and (5) traveling to and from their first and last appointments of the day." (Williams I, supra, 221 Cal.App.4th at pp. 1357-1358.) "For an adjuster to work overtime, Allstate's official company policy required the adjuster to get his supervisor's prior approval. But the adjusters' declarations stated the adjusters hesitated to request overtime because they did not want to be perceived as 'bad' employees." (Id. at p. 1358.)
B. Inadequate Wage Statements
In addition to allegations of uncompensated overtime, Williams alleged that the wage statements to provided the adjusters did not comply with the statutory requirements of Labor Code section 226. That statute requires every employer to provide employees with accurate wage statements accompanying each payment of wages. The wage statements must include certain specified information, including "total hours worked by the employee" (Lab. Code, § 226, subd. (a)(2)), and "all applicable hourly rates in effect during the pay period and the corresponding number of hours worked at each hourly rate by the employee" (id., subd. (a)(9)).
Just as the Work Force Management System presumed an 8-hour workday, Allstate paid its adjusters on the assumption that they worked 80 hours every two weeks. Rather than issuing wage statements setting forth the hours worked and the hourly rates paid, each wage statement began with "Regular Earnings," which was the amount the employee would have earned if the employee had worked the presumed 80 hours. Thereafter, the wage statement included adjustments, either adding pay (for overtime hours) or subtracting pay (for time off). Wage statements did not include all adjustments for the two-week period they covered. Allstate paid employees on a two-week pay cycle, but could only include adjustments related to hours worked in the first week in the paycheck for that pay period. If there was overtime worked (or time taken off) in the second week, the adjustment would appear in the next period's pay and corresponding wage statement. Allstate contends this was permitted under Labor Code section 204, subdivision (b)(2), which provides that an employer is in compliance with Labor Code section 226 relating to total hours worked "if hours worked in excess of the normal work period during the current pay period are itemized as corrections on the paystub for the next regular pay period."
Examples in the record show that not only did wage statements include adjustments for the previous pay period, but a wage statement could sometimes include adjustments for multiple prior periods. For example, Williams's wage statement for the second half of November 2005 includes modifications for five pay periods, including one going back as far as July of that year.
Allstate could not dispute that, for at least part of the class period, its wage statements did not comply with the Labor Code. Prior to 2010, the wage statements did not include the hours worked or the hourly rate; the statements simply assumed 80 hours were worked in every two-week period, unless otherwise indicated. An employee could calculate the hourly rate of pay by dividing the regular earnings by 80, but neither the number of hours nor the rate actually appeared on the wage statements. Beginning in 2010, a "Rate" column was added to the wage statement, which disclosed both the regular earnings hourly rate as well as the hourly rates for "Regular Overtime" and "Premium Overtime," when overtime was worked. Williams argued the revised wage statements were still inadequate, due to the fact that employees had to look at a minimum of two wage statements to determine "all applicable hourly rates in effect during the pay period and the corresponding number of hours worked at each hourly rate by the employee." (Lab. Code, § 226, subd. (a)(9).)
2. Complaint and Class Certification
In December 2007, Williams brought suit against Allstate, on behalf of himself and all others similarly situated. He alleged nine causes of action, including: failure to pay overtime wages, failure to timely furnish accurate itemized wage statements, and unfair business practices (Bus. & Prof. Code, § 17200). The remaining six causes of action are no longer at issue.
Williams moved for class certification. The court granted certification of an "Off the Clock" class and a "Wage Statement" class. The Off the Clock class was defined as "Defendant's California-based hourly-paid Auto Field Adjusters from January 1, 2005 to June 6, 2011, to the extent that Defendant failed to pay for off-the-clock work for the following specific tasks performed prior to the first inspection of the day; logging on and off computer systems, preparing and checking voicemail messages, checking for schedule and travel changes, obtaining directions to the first inspection if there is a travel change, and making courtesy calls."
The Off the Clock class was also certified as an Unfair Business Practices class. There is no dispute that the viability of an Off the Clock class determines its viability as an Unfair Business Practices class. We therefore refer to it only as the Off the Clock class, with the understanding that our disposition applies to both claims.
The Wage Statement class was defined as "Defendant's California-based hourly-paid employees from December 19, 2006 to June 6, 2011, to whom Defendant did not furnish accurate wage statements pursuant to Labor Code [section] 226(a) and (e)."
3. The First Decertification Order
In 2011, in light of intervening U.S. Supreme Court authority (Wal-Mart Stores, Inc. v. Dukes (2011) 564 U.S. 338, 131 S.Ct. 2541 (Wal-Mart)), the trial court decertified the Off the Clock class, on the theory that Allstate must be permitted to litigate its defenses to the claims of each individual class member - including its assertion that not all adjusters worked off the clock, and that some of those who might have worked off the clock did so for de minimis periods. (Williams I, supra, 221 Cal.App.4th at p. 1359.)
4. Reversal of the First Decertification Order
Williams filed a petition for writ of mandate, which we summarily denied. He sought review by our Supreme Court. The Supreme Court granted the petition and returned the matter to us with directions to issue an order to show cause why relief should not be granted. We issued the order to show cause, received briefing, held argument, and issued the writ. (Williams I, supra, 221 Cal.App.4th at pp. 1359-1360, 1371.)
We concluded the Wal-Mart case on which the trial court had relied to grant decertification was distinguishable. (Williams I, supra, 221 Cal.App.4th at pp. 1362-1364.) We need not rehash our analysis here. Important for our purposes is that we recognized that Williams's off-the-clock claim is not based on the idea that some supervisors required uncompensated overtime while others did not - which could require an investigation into the subjective intent of each of the individual supervisors. Instead, the claim is based on Williams's assertion that "there is a companywide policy to deny overtime pay." (Id. at p. 1364.) We explained, "[h]ere, the question is whether Allstate had a practice of not paying adjusters for off-the-clock time. [Citation.] The answer to that question will apply to the entire class of adjusters." (Id. at p. 1365.) To the extent Allstate disputed "whether a companywide practice existed of adjusters working off the clock," we concluded that this went "to the merits of the class claims." (Id. at pp. 1369-1370.) We explained, "[i]t may be true that some adjusters never worked off the clock, and such adjusters were thus not injured by Allstate's practice of adjusters working off the clock. But the existence of individuality as to damages does not defeat class certification. [Citations.]" (Id. at p. 1370.)
We granted the writ petition and directed the trial court to recertify the Off the Clock class. (Williams I, supra, 221 Cal.App.4th at p. 1371.) The trial court complied.
5. Summary Adjudication is Denied With Respect to the Wage Statement Class
With the Off the Clock class recertified, the parties turned to the Wage Statement class, by cross-motions for summary adjudication.
It does not appear to be seriously disputed that, at the very least, the pre-2010 wage statements violated the requirements of Labor Code section 226. Those statements did not include the hourly rate or the hours worked. The issue, however, was whether the class would be entitled to damages for that violation.
Labor Code section 226, subdivision (e)(1) provides statutory damages for an "employee suffering injury as a result of a knowing and intentional failure by an employer to comply with subdivision (a)" of the statute. Injury was the battleground of the summary adjudication motions. Allstate argued that it was entitled to summary adjudication because no class member suffered any injury from its wage statements, as the class members could determine the missing information by simple mathematics. In opposition, Williams argued that a 2013 amendment to Labor Code section 226 clarified that injury is presumed by the provision of an inaccurate statement itself.
The trial court denied both motions. The court concluded that the evidence showed that some members of the class had been injured while others had not, making trial necessary on this point. The court also concluded that the 2013 amendment had no retroactive effect, so did not apply to this class (which did not go beyond 2011). The case then wended its way toward trial. 6. Williams's First Proposed Trial Plan and Dr. Drogin's First Declaration
In October 2015, Williams submitted his first trial plan. His plan began with the undisputed premise that the Off the Clock class had 284 members and the Wage Statement class had 2,318 members. Williams contemplated that proof for both classes would rely, in part, on testimony of "randomly selected class members."
Williams did not necessarily intend for all of the randomly selected class members to testify at trial. He envisioned a plan where they would be deposed, and Williams would "present the results of the testimony through the depositions, testimony of their expert, and/or the random sample at trial."
Williams supported this trial plan with the declaration of Dr. Richard Drogin, an expert statistician. As to what would ultimately be the critical issue of the sample size, Dr. Drogin conceded that it was impossible to determine the appropriate sample size without first learning about the "variability" within the populations being studied. He therefore recommended first using a "relatively small pilot sample" to measure variation.
"Variability" is a measure of how standardized a result is throughout a population. If there is too much variability in a class, "individual issues are more likely to swamp common ones and render the class action unmanageable." (Duran v. U.S. Bank Nat. Assn. (2014) 59 Cal.4th 1, 33.)
He further stated that the "choice of sample size depends on the desired margin of error and the amount of resources available for implementing the sample plan. It is up to the user of the statistics (i.e. trier of fact) to decide on an acceptable confidence level and margin of error." Therefore, Dr. Drogin declined to propose a specific margin of error or sample size, but provided various alternatives from which the court could choose. His chart assumed a 95 percent confidence level, and then set forth the various margins of error which would be produced by different sample sizes, based on certain assumptions regarding variation. The larger the sample, the smaller the margin of error. 7. Allstate's Opposition and Dr. Slottje's Declaration
Allstate opposed Williams's first trial plan. The bulk of its opposition consisted of the declaration of a competing expert statistician, Dr. Daniel J. Slottje, who evaluated and criticized Dr. Drogin's declaration.
Dr. Slottje believed Dr. Drogin's declaration suffered from "critical flaws such that Dr. Drogin's opinions regarding his view of the appropriate sample size and his proffered sampling methodology should not be relied upon in this matter."
Dr. Slottje explained that a 95 percent confidence level (or higher) is the standard in performing scientific statistical analysis. Moreover, the generally accepted margins of error for scientifically valid statistical analysis and applied research range between ±0.1 percent and ±5 percent.
Dr. Slottje agreed with Dr. Drogin that variation in the population was presently unknown. However, Dr. Slottje argued that "[b]asic statistics books explain the statistical procedures and provide best practice ways" to perform an analysis in that situation. "[D]etermining the sample size when variation in the population under scrutiny is unknown is straightforward. In this case, using a 95% confidence interval and ±5% margin of error, the sample size for a population of 284 (off-the-clock class) would be 164 class members and the sample size for a population of 2,376[] would be 331 class members, for similar parameters." 8. Williams's Reply and Dr. Drogin's Second Declaration
The slight change in the size of the Wage Statement class from 2318 to 2376 is unexplained, but also undisputed.
In reply, Williams submitted a second declaration of Dr. Drogin, changing the trial plan in response to Dr. Slottje's criticisms.
Dr. Drogin explained the concept of one-tailed and two-tailed confidence intervals. A two-tailed confidence interval provides a result that is bounded on both sides by the margin of error. (For instance, a 95% confidence level that X% plus or minus 5% of class members worked overtime.) A one-tailed confidence interval provides a result bounded only on one side. (For instance, a 95% confidence level that at least X% of class members worked overtime.) Dr. Drogin argued that "for liability, a one-tailed confidence interval is relevant."
To the extent Dr. Drogin recommended a one-tailed confidence interval for liability, he did so only with the assumption of a 95 percent confidence level. He stated, "in this litigation, the Court could determine a [damages] proceeding is necessary if there is a high confidence (95%) that the true value of the proportion who worked off-the-clock before the first inspection is above [ ] a certain level."
Dr. Drogin also suggested that, when using a one-tailed confidence interval for liability, a larger margin of error is acceptable on the theory that what matters is the lower bound itself, not the margin of error. For example, a 95 percent confidence level that something is true 80 percent of the time with a 9 percent one-tailed margin of error would mean a 95 percent confidence level that it is true at least 71 percent of the time. What matters for liability, according to Dr. Drogin, is the 95 percent confidence that the proposition is true at least 71 percent of time - not that the margin of error used happened to be 9 percent.
He was, however, insistent that "for the determination of aggregate classwide damages awarded to the Off the Clock [c]lass, if liability is found, a two-tailed confidence interval will be appropriate. In other words, it will be necessary to obtain an accurate estimate of the damages, in order to avoid awarding damages that are too low or too high." A one-tailed confidence interval was suggested only for liability.
With all that said, Dr. Drogin proposed sample sizes without calculating the confidence level or margin of error associated with them. He proposed a random sample of 30 class members to testify as to liability for the Off the Clock class. Damages for the Off the Clock class would likely require a larger sample, but Dr. Drogin did not suggest a number, preferring to first determine variability from the testimony of the original sample. As to the Wage Statement class, Dr. Drogin again recommended a random sample of 30 class members. 9. Allstate Submits Dr. Slottje's Second Declaration
Allstate responded to Dr. Drogin's second declaration with a second declaration of Dr. Slottje. Dr. Slottje argued that Dr. Drogin's opinions had completely changed, but his declaration still suffered from critical flaws.
Dr. Slottje stated, "There is no ambiguity about this. Dr. Drogin is now proposing trying liability in this case based on sampling only thirty people for each class." Dr. Slottje disagreed with Dr. Drogin's suggestion that a one-tailed confidence interval would be appropriate for liability, because the question to be resolved is "what the true population percentage is that has suffered the alleged violations."
Dr. Slottje then calculated the margins of error for a 30-person sample (for both classes) for a two-tailed 95 percent confidence interval. For a class of 284 members (the Off the Clock class), a sample of 30 results in a margin of error of ±16.95 percent, and for a class of 2,376 members (the Wage Statement class), a 30-person sample gives a margin of error of ±17.78 percent. Margins of error of that size "are so wide as to be useless in virtually any application." "In other words, an analysis using such a large margin of error gives us not only useless, but misleadingly useless information (particularly when, as Dr. Drogin does here, he implies that knowing such useless information to a 95% degree of confidence level somehow matters.)"
Nonetheless, even using a one-tailed confidence interval, Dr. Slottje stated that the margins of error with a 30-person sample size in these two populations are over 12 percent "which are still far too high to be useful."
10. The Trial Court Rejected Williams's Trial Plan
The court rejected Williams's trial plan on four grounds. Assuming without deciding that a statistical approach is a legitimate basis on which to establish liability (as opposed to damages), the court found four problems with Dr. Drogin's plan:
The first problem was its failure to address variation. Dr. Drogin agreed that one must know variation to determine an appropriate sample size, but Dr. Drogin did not perform a pilot study or even explain what one would look like. By his second declaration, Dr. Drogin had dismissed the idea of a pilot study for liability, seeking to establish liability with a 30-person sample, and use that as a pilot study for damages only.
The second problem was sample size. While Dr. Slottje had identified a formula which resulted in an Off the Clock sample size of 164 and a Wage Statement sample size of 331, Dr. Drogin did no such thing, identifying a sample size of 30 with no logical or mathematical basis.
The third problem was sample bias. Dr. Drogin had suggested using the same 30 witnesses for both classes, which made the sample not a random sample of the Wage Statement class.
The fourth problem was what the court referred to as the liability rule. "What statistical liability rule is the statistical analysis to inform? How is the fact finder to infer liability to the class from a mere sample?" The court asked, "Would a showing that 1% of the class performed unpaid work mean Allstate is liable to the class? 51%? Williams provides no workable answer to this question because he does not address the issue."
11. Allstate's Second Motion to Decertify the Class
Sixty days later, in the absence of any new trial plan having been submitted by plaintiffs, Allstate moved to decertify the class as unmanageable. 12. Williams's New Trial Plan and Dr. Kriegler's Declaration
In addition to arguing the absence of a viable trial plan, Allstate argued that the evidence simply failed to show that all Off the Clock class members worked off the clock. To the contrary, Allstate argued that company policy was that no adjusters should work off the clock, and if they did, they were to report it and get paid for it. We do not address this argument as it goes to the merits of the class claim, not certifiability.
Williams opposed the motion to decertify by proposing a new trial plan, and sought a court trial on liability and damages.
Under the new trial plan, the liability phase for the Off the Clock class would not rely on statistical sampling at all. Williams intended to establish a companywide practice of off-the-clock work with Allstate's computer records, payroll records, policies and practices. This was intended to alleviate the trial court's concerns regarding a statistical approach to liability; liability would instead be established traditionally, without statistical analysis.
As for damages for the Off the Clock class, Williams sought to rely on statistical sampling. "Plaintiff proposes a sample of the Off-the-Clock class will testify at deposition and/or trial. The sample will be selected by the expert and will be statistically valid based on an acceptable confidence interval and margin of error [citation]." "Plaintiff will call the sample to testify at deposition and/or trial[] and will elicit their direct testimony via a predetermined set of questions." Williams proposed a sample of between 76 and 142 class members, which he believed to be manageable. The expert testimony that supported this range of numbers will be discussed momentarily.
At this point, Williams inserted a footnote explaining that class members are located throughout California. Relying on Code of Civil Procedure section 2025.620, subdivision (c)(1), which provides that the deposition of a person residing more than 150 miles from trial may be used for any purpose, Williams stated, "a portion of the sample who reside over 150 miles from the Courthouse will testify at trial via their depositions." He added "The Court also has discretion to allow other class members to testify at trial via their depositions if it is 'in the interest of justice.' [Citation.]"
As for the Wage Statement class, liability would be based on statistical sampling along the lines proposed by Dr. Slottje in opposition to the first trial plan. Statutory damages could then be calculated; if, for example, the trier of fact concluded 70 percent of the class suffered injury, it should award 70 percent of what total statutory damages for the class would be.
As to the key issue of the number of witnesses necessary to testify for each class, Williams submitted the declaration of a new expert statistician, Dr. Brian Kriegler. Dr. Kriegler's declaration was, in summary, an attempt to justify reducing the sample sizes from the numbers Dr. Slottje had proposed, by reducing accuracy in the interests of manageability.
The first way Dr. Kriegler sought to reduce sample size was by potentially reducing the confidence level. Dr. Kriegler agreed that 95 percent is a "commonly used confidence level," but he argued that "other levels can be sufficient in certain circumstances." While he did not outright recommend a new confidence level, he proposed allowing the court to choose between 95 percent and 84 percent.
The second way in which Dr. Kriegler lowered the sample size was by proposing the use of one-tailed confidence intervals. As we have discussed, two-tailed confidence intervals provide results with margins of error on both sides. One-tailed intervals provide only the lower boundary of the margin of error. Dr. Kriegler advocated "one-tailed confidence intervals so that the Court can be sufficiently confident that true population values are at least above a certain value." While Dr. Drogin had previously argued for one-tailed confidence intervals for liability, but not damages, Dr. Kriegler wanted to use them for both determinations.
The third way in which Dr. Kriegler reduced sample size was by using a 10 percent margin of error, rather than 5 percent. He assumed a 10 percent margin of error based on a prior California case. (Bell v. Farmers Insurance Exchange (2004) 115 Cal.App.4th 715, 723 (Bell) [in an overtime class, the trial court requested the parties to continue sampling until they could reach a result of average hours of weekly overtime plus or minus one hour; the result was 9.42 hours plus or minus 0.9 hours - a two-tailed 9.6% margin of error].)
Using these three methods, Dr. Kriegler's recommended sample sizes for the Off the Clock damages class ranged from a sample of 76 for 84 percent confidence (in a one-tailed 10% margin of error) to a sample of 142 for 95 percent confidence (in the same one-tailed 10% margin of error).
As to the Wage Statement class, Dr. Kriegler agreed in principle with Dr. Slottje's formula for the sample size necessary to establish liability. He had no objection to a 95 percent confidence interval or a 5 percent margin of error. However, he preferred a one-tailed confidence interval, which reduced the necessary sample size from 331 to 243.
13. Hearing and Ruling
After Allstate filed a reply, the matter proceeded to hearing. The court provided the parties with a copy of its written tentative decision, which was to decertify the class for failure to provide a manageable trial plan. The court accepted Dr. Slottje's calculation of the necessary sample sizes: 164 for the Off the Clock class and 331 for the Wage Statement class. A trial of over 500 witnesses would take five or six months, said the trial court, and would be utterly unmanageable.
Allstate's reply included the declaration of yet another statistician expert. The trial court ultimately did not rely on that declaration. We therefore do not discuss it.
The court was not impressed by Dr. Kriegler's attempts to reduce the sample size by reducing confidence levels, and noted that an 84 percent confidence level is "unprecedented in the statistics literature, so far as Kriegler explains or this court can discover." The court also was baffled by Dr. Kriegler's reliance on a one-tailed test for Off the Clock damages. This was particularly true given that Dr. Drogin, Williams's own prior expert, agreed that a two-tailed test was appropriate for damages. The court found Dr. Kriegler's reasons for recommending a one-tailed test to be "not cogent." The court noted that Dr. Kriegler implied that "no one is interested in how high damages might be in this case. That is incorrect. Both parties are interested in that question." The court concluded that Dr. Kriegler departed from "professional analytical conventions without cogent justification" and "strain[ed] to reach results favorable to his client." On that basis, the court rejected Dr. Kriegler's opinion.
The trial court did not specifically find fault in Dr. Kriegler's use of a 10 percent margin of error rather than a 5 percent margin of error.
Williams attempted to argue against the court's tentative. He argued that 84 percent is a sufficient confidence level as it is far in excess of a preponderance of the evidence. In any event, he noted that if the court wanted to use a 95 percent confidence level, Dr. Kriegler had provided the necessary sample size for that as well.
The court concluded that Williams had repeatedly failed to propose a reliable plan for trial, and granted decertification. It adopted its tentative.
14. Notice of Appeal
Williams filed a timely notice of appeal. A decertification order is appealable. (Lubin v. The Wackenhut Corp. (2016) 5 Cal.App.5th 926, 934 (Lubin).)
15. Change in Authority For The Wage Statement Class
While this appeal was pending, the Court of Appeal issued its opinion in Lubin, supra, 5 Cal.App.5th at page 934. Lubin addresses whether the 2013 amendment to Labor Code section 226, which presumed injury from a defective wage statement, should be given retroactive effect. The court concluded that the statute was a clarification of existing law, which therefore applied to the Lubin dispute. (Lubin, at p. 959.) Williams argues that Lubin changes the legal landscape on which the parties had litigated the trial plan for the Wage Statement class. Since injury can now be presumed, it can be presumed classwide, and there is no need for representative damage testimony with respect to that class at all.
We first consider whether the trial court erred in ruling on the decertification motion before it. We conclude that it did not. However, we then consider whether the recent authority of Lubin mandates reversal of the court's order as it relates to the Wage Statement class. We conclude that it does.
DISCUSSION
1. Standard of Review
"We review a decertification order for an abuse of discretion. [Citations.] Decertification requires new law or newly discovered evidence showing changed circumstances. [Citation.] A motion for decertification is not an opportunity for a disgruntled class defendant to seek a do-over of its previously unsuccessful opposition to certification. 'Modifications of an original class ruling, including decertifications, typically occur in response to a significant change in circumstances, and "[i]n the absence of materially changed or clarified circumstances . . . courts should not condone a series of rearguments on the class issues." [Citation.]' [Citation.] '[A] class should be decertified "only where it is clear there exist changed circumstances making continued class action treatment improper." ' [Citation.]" (Williams I, supra, 221 Cal.App.4th at pp. 1360-1361.)
"In an exception to a customary rule of appellate practice, we review the court's rationale for its order. (The customary rule is to review the result of the court's order, not its rationale.) [Citations.] We thus review only the reasons the court stated for its order, and we reverse if those reasons do not support the order. [Citations.]" (Williams I, supra, 221 Cal.App.4th at p. 1361.) "We will affirm an order denying class certification if any of the trial court's stated reasons was valid and sufficient to justify the order, and it is supported by substantial evidence. [Citation.] We will reverse an order denying class certification if the trial court used improper criteria or made erroneous legal assumptions, even if substantial evidence supported the order. [Citations.]" (Knapp v. AT&T Wireless Services, Inc. (2011) 195 Cal.App.4th 932, 939.)
2. Manageability
Often class certification appeals are concerned with the issue of whether common questions of law or fact predominate over individual questions. (E.g., Sav-On Drug Stores, Inc. v. Superior Court (2004) 34 Cal.4th 319.) This is not one of them. A related issue, however, is whether the action is manageable as a class action. Our Supreme Court has held that "[i]ndividual issues do not render class certification inappropriate so long as such issues may effectively be managed. [Citations.]" (Id. at p. 334.) Even after the class has been certified, "if unanticipated or unmanageable individual issues do arise, the trial court retains the option of decertification. [Citations.]" (Id. at p. 335.) That is precisely what occurred here. Although the class was initially certified for reasons of common issues predominating, it was decertified because the plaintiff class was unable to come forward with what the trial court believed to be a manageable trial plan.
3. Use of Statistics
The parties marshalled significant statistical evidence to present to the trial court in support of their respective positions on manageability. There is no per se ban on using statistics or representative evidence in a class action. As the United States Supreme Court recently explained, "[a] representative or statistical sample, like all evidence, is a means to establish or defend against liability. Its permissibility turns not on the form a proceeding takes—be it a class or individual action—but on the degree to which the evidence is reliable in proving or disproving the elements of the relevant cause of action. [Citations.]" (Tyson Foods, Inc. v. Bouaphakeo (2016) ___ U.S. ___, ___ [136 S.Ct. 1036, 1046].)
The seminal California case on the use of statistical or representative evidence is Duran v. U.S. Bank Nat. Assn. (2014) 59 Cal.4th 1 (Duran). In Duran, the trial court concluded a plaintiff class of 260 employees had been misclassified as exempt from the overtime laws based on the representative testimony of only 21 class members, despite the fact that neither party had advocated for such a small sample. (Id. at pp. 12, 16.) Based on the sample's testimony, the court found that all of the sample witnesses were misclassified; plaintiffs' expert concluded that the finding of misclassification could be extrapolated to the class with a 13 percent margin of error, at a 95 percent confidence level. (Id. at pp. 21-22 & fn. 15.) The trial court agreed, then further extrapolated the average amount of overtime reported by the sample (11.86 hours) to the entire class - even though the plaintiffs' own expert testified this had a margin of error of plus or minus 43.3 percent (at a 95% confidence level). (Id. at pp. 22, 24.)
The Supreme Court in Duran reversed the court's judgment for the plaintiff class for two reasons: first, the court had failed to permit the defendant to litigate its individual affirmative defenses; and second, the court's implementation of statistical sampling was itself flawed. (Duran, supra, 59 Cal.4th at p. 25.) We are here concerned with the second point only.
Williams disclaimed any intention to rely on sampling to establish liability to the Off the Clock class, intending to proceed by traditional proof (e.g., Allstate's own records) only. Thus, Allstate's ability to establish its affirmative defenses to Off the Clock liability is not at issue.
If a plaintiff seeks to use representative testimony to infer facts about liability and damages, there must be a sufficiently representative sample for the inferences to be justified. (Duran, supra, 59 Cal.4th at p. 38.) "Several considerations determine whether a sample is sufficiently representative to fairly support inferences about the underlying population." (Ibid.)
One such consideration is sample size. (Duran, supra, 59 Cal.4th at p. 42.) "It is impossible to determine an appropriate sample size without first learning about the variability in the population." (Ibid.) "One way to assess population variability is through the use of surveys." (Ibid.) "With input from the parties' experts, the court must determine that a chosen sample size is statistically appropriate and capable of producing valid results within a reasonable margin of error." (Ibid.)
A second consideration is that the margin of error must not be intolerably large. (Duran, supra, 59 Cal.4th at p. 46.) "Statisticians typically calculate margin of error using a 95 percent confidence interval, which is the interval of values above and below the estimate within which one can be 95 percent certain of capturing the 'true' result." (Ibid.) In Duran, our Supreme Court had no difficulty in concluding that, even though the confidence level was 95 percent, the 43.3 percent margin of error in calculating damages was too large to reliably support the trial court's award. (Id. at p. 46.) The court did not express an opinion on whether the 13% margin of error for liability was too high, on the basis that there were so many errors in the choice of the testifying witnesses that the sample was not truly random, and the real margin of error could have been considerably higher. (Id. at pp. 48-49.)
The Court of Appeal in Bell, supra, 115 Cal.App.4th at page 756 found a 32 percent margin of error to be insufficient to survive a due process challenge under the circumstances, although it was careful to state that margin of error alone does not "afford a bright-line constitutional distinction."
In short, "[t]rial courts have broad discretion in many areas. But they cannot exercise that discretion in ways contrary to the internal rules of a scientific specialty, such as statistics, and then rely on that specialty's established reliability as if the rules had been followed." (Duran, supra, 59 Cal.4th at p. 45.)
4. The Court Did Not Abuse Its Discretion in Decertifying the Two Classes For Unmanageability
In this case, Williams had two - three, if we count Dr. Drogin's two declarations separately - opportunities to submit a trial plan for this case, and he failed to satisfy the court that a class trial would be manageable.
The trial court's concerns were with sample size. Specifically, the court accepted Dr. Slottje's testimony that appropriate sample sizes, with a 95 percent confidence level and a ±5 percent margin of error, would have been 164 for the Off the Clock class and 331 for the Wage Statement class. The court concluded that a trial with nearly 500 witnesses would be unmanageable. The court found Dr. Kriegler's plan, which reduced sample size by relying on an 84 percent confidence level and a one-tailed, 10 percent margin of error, was unsupported.
The court did not specifically find fault in Dr. Kriegler's use of a 10 percent margin of error rather than a 5 percent margin of error. Its concern was with his use of a one-tailed margin of error and an 84 percent confidence level.
Reviewing this determination for abuse of discretion, we find none. The court reasonably insisted on a 95 percent confidence level. Dr. Slottje testified that a 95 percent confidence level is the standard in performing scientific statistical analysis. While plaintiffs' initial expert, Dr. Drogin, did not actually propose a confidence level, his chart of proposed sample sizes and margins of error assumed a 95 percent confidence level. In Dr. Drogin's second declaration, he specifically assumed a 95 percent confidence level for damages for the Off the Clock class. We do not hold that a 95 percent confidence level is necessary for all uses of statistical analysis in trial plans; we simply hold here that the trial court did not abuse its discretion in requiring a trial plan which rose to the usual standard in performing scientific statistical analysis.
On appeal, Williams sought judicial notice of the Federal Judicial Center's Reference Manual on Scientific Evidence (3d ed. 2011), pages 245 to 246, which, according to Williams, "discusses confidence intervals ranging from 68% to 99.7%." Williams is technically correct; confidence intervals as low as 68 percent are discussed in the manual. The manual is, at that point, explaining how margins of error relate to confidence levels. When sampling is performed, the average and the "standard error" (also called "standard deviation") are calculated. When the value of the standard deviation is used as the margin of error, a 68 percent confidence interval results; when the margin of error is double the standard deviation, the confidence interval is 95 percent; when the margin of error is triple the standard deviation, the confidence interval is 99.7 percent. (Ibid.) At no point does the manual suggest that a confidence interval as low as 68 percent may legitimately be used. If anything, it suggests the opposite: "The 95% confidence level is the most popular, but some authors use 99%, and 90% is seen on occasion." (Id. at p. 245.) Williams has furnished no authority in which an appellate court has applied a 68 percent confidence interval.
We reach the same conclusion with respect to the trial court's determination that a two-tailed margin of error would be necessary, and that Dr. Kriegler's attempt to justify a one-tailed margin of error was not supported. Again, as the trial court found, Dr. Kriegler's declaration is contradicted by both defense expert Dr. Slottje and plaintiffs' first expert Dr. Drogin. Dr. Slottje stated that a two-tailed approach was the standard. Although Dr. Drogin had suggested a one-tailed confidence interval was appropriate for determinations of liability, he agreed that a two-tailed approach to damages was necessary, "in order to avoid awarding damages that are too low or too high." Yet, Dr. Kriegler advocated for a one-tailed approach to damages, on the apparent basis that an award of damages that are "too low" is an acceptable trade-off when the alternative is decertification.
Combined, these problems identified by the trial court in Dr. Kriegler's plan (low confidence level and one-tailed confidence interval) undermine the viability of his plan. Even when plaintiffs' first expert Dr. Drogin had advocated for a one-tailed approach to the liability determination he did so only when it was based on a high, 95 percent confidence level. Dr. Kriegler not only would use a one-tailed approach to damages, he would do so at a much lower confidence level. Dr. Drogin's testimony itself was sufficient evidence that Dr. Kriegler's trial plan was not statistically sound, and the trial court did not err in rejecting it.
In sum, the trial court's conclusion that Williams did not present a viable triable plan which complied with Duran is not an abuse of discretion. The decertification motion was therefore properly granted.
On appeal, Williams approaches the question differently, devoting over 10 pages of his opening brief to the argument that the trial court abused its discretion in excluding Dr. Krigler's testimony. But the trial court's ruling was not an evidentiary one; the court was not excluding an expert's testimony at trial. Instead, the court affirmatively considered that testimony, addressed it at length in its ruling, and determined that it did not support a legitimate trial plan.
The confusion arises because the trial court cited to a case regarding the exclusion of expert testimony. The trial court stated, "In sum, Kriegler departs from professional analytical conventions without cogent justification. Kriegler strains to reach results favorable to his client. Under Sargon Enterprises, Inc. v. University of Southern California (2012) 55 Cal.4th 747, 772, the court entirely rejects Kriegler's clearly invalid and unreliable opinion." While it is true that the cited case related to the exclusion of evidence, it is apparent from the text itself that the trial court did not exclude Dr. Kriegler's opinion, but "reject[ed]" it as "invalid and unreliable."
Finally, Williams argues that even if the court accepted Dr. Slottje's plan for statistical sampling, which required more than 500 witnesses, the trial would not be unmanageable. This is because "most of the testimony would be obtained through depositions - not trial testimony. The results of the depositions could be presented through expert testimony at trial." This argument is not supported by the record. Williams's trial plan explained that Williams "will call the sample to testify at deposition and/or trial" and that "a portion of the sample who reside over 150 miles from the Courthouse will testify at trial via their depositions." Williams cannot now defeat the court's ruling by proposing trial by expert.
Williams's initial trial plan had proposed that results of depositions would be presented at trial. Its language was undermined by the reply declaration of Williams's then-expert, Dr. Drogin, who rather adamantly stated, "class members will testify at trial, under oath, and would be subject to cross-examination and other possible impeaching evidence. Moreover, the interpretation of this witness testimony will ultimately be done by the court, not an expert."
5. The Lubin Opinion and the Wage Statement Class
As we have concluded the court did not abuse its discretion in decertifying the class for unmanageability, we should affirm. However, while the appeal was pending, the Court of Appeal issued its opinion in Lubin, supra, 5 Cal.App.5th 926, thereby changing the framework under which the Wage Statement class action must be tried.
As we have observed, it is conceded that at least some of Allstate's wage statements did not include all of the information required by Labor Code section 226. Subdivision (e) of that statute provides statutory damages for an "employee suffering injury as a result of a knowing and intentional failure by an employer to comply" with the necessary disclosures. A dispute arose in the case law as to what it meant for an employee to "suffer[] injury" within the meaning of the statute. At one end were cases such as Kisliuk v. ADT Security Services, Inc. (C.D. Cal. 2008) 263 F.R.D. 544, 548-549, which held that the injury requirement may be satisfied merely by an employee's receipt of an inadequate wage statement without actual injury. At the other end were cases like Price v. Starbucks Corp. (2011) 192 Cal.App.4th 1136, 1142-1143 (Price), which held that mere receipt of an inadequate wage statement did not establish injury, nor do employees satisfy the injury requirement if they only had to do simple mathematics to determine the missing information and confirm that they were properly compensated. Somewhere in between the two extremes were cases like Jaimez v. Daiohs USA, Inc. (2010) 181 Cal.App.4th 1286, 1306-1307, which concluded there was a "fairly minimal standard" to establish injury and the difficulty and expense incurred in reconstructing pay records in a wage and hour lawsuit itself might satisfy it.
In 2013, in light of this confusion, the Legislature amended subdivision (e) of Labor Code section 226, adding language explaining that an "employee is deemed to suffer injury for purposes of this subdivision if the employer fails to provide accurate and complete information as required" by subdivision (a) of that statute, and "the employee cannot promptly and easily determine from the wage statement alone one or more" of several of the required items, including total hours worked, all applicable hourly rates in effect, and the corresponding number of hours worked at each rate. (Lab. Code, § 226, subd. (e)(2)(B).) The amended statute further provides that " 'promptly and easily determine' means a reasonable person would be able to readily ascertain the information without reference to other documents or information." (Lab. Code, § 226, subd. (e)(2)(C).) This provision renders the inquiry an objective one; the amendment "clarifies that injury arises from defects in the wage statement, rather than from a showing that an individual experienced harm as a result of the defect." (Lubin, supra, 5 Cal.App.5th at p. 959; see also Garnett v. ADT, LLC (E.D. Cal. 2015) 139 F.Supp.3d 1121, 1133 ["Whether an employee suffered injury is based solely on the information provided on the wage statement."].) Because the inquiry is objective, relying on whether a "reasonable person would be able to readily ascertain the information" (Lab. Code, § 226, subd. (e)(2)(C)), and not on whether any particular plaintiff could have, the inquiry now appears to be one easily amenable to common proof in a class action. (Lubin, at p. 959.)
Allstate argues against this conclusion, stating that the injury inquiry is still a subjective one even after the 2013 amendment, by relying on Milligan v. American Airlines, Inc. (9th Cir. 2014) 577 Fed.Appx. 718, 719, an unpublished 9th Circuit case. Unpublished federal authority has no precedential value, but may be cited as persuasive. (Walker v. Apple, Inc. (2016) 4 Cal.App.5th 1098, 1108, fn. 3.) Milligan is not persuasive on this point, in that, although it relies on the prior California opinion in Price for the subjective injury requirement, it does not address the 2013 amendment. Milligan was argued and submitted in 2009; the opinion was not issued until 2014 because of the automatic stay due to the defendant's bankruptcy. (Milligan, at p. 719, fn. 1.) Nothing in the opinion suggests the parties brought the amended statute to the court's attention when the stay was lifted.
Allstate relies on three additional federal cases - two of which are unpublished - which, according to Allstate, "cite Price with approval" after the date of the 2013 amendment. (See Apodaca v. Costco Wholesale Corp. (9th Cir. 2017) 675 Fed.Appx. 663, 665; Byrd v. Masonite Corp. (C.D. Cal. 2016) 215 F.Supp.3d 859, 868; Silva v. AvalonBay Communities, Inc. (C.D. Cal. 2015) 2015 WL 11422302, *9-10.) But each of the cases assumes Price remains the governing standard, with no consideration of the change in the standard wrought by the 2013 amendment. We are therefore unpersuaded by them.
Lubin considered the legislative history of the 2013 amendment, and concluded that it was not a new law, but a mere clarification of existing law. (Lubin, supra, 5 Cal.App.5th at p. 959.) As such, it was to be given retroactive effect. (Ibid.) In Lubin, the trial court had declined to certify a wage statement class on the basis that injury would have to be established on an individual basis. (Id. at pp. 958-959.) In light of the 2013 amendment, and its retroactive application, Division Four of our court reversed, concluding that the issue of injury was now amenable to class treatment. (Id. at p. 959.)
A similar result applies here. The trial court concluded, on summary adjudication, that the 2013 amendment did not apply and, instead, the subjective standard for individual injury found in Price did. This meant that whether any individual class member was injured was a matter for trial. When Williams submitted his trial plan, he sought to prove injury to the class by representative testimony. The trial court ultimately decertified the class, because it would require the testimony of nearly 500 witnesses - 331 of whom would constitute the necessary sample to establish injury to the Wage Statement class. But Price is no longer applicable, the 2013 amendment governs, and injury is subject to common proof based only on the wage statements themselves. The court's order that the Wage Statement class is unmanageable is no longer supported by the law, and therefore must be reversed.
Allstate disagrees and suggests that Lubin was incorrect in its retroactivity analysis because the legislative history of the 2013 amendment shows no intent that the statute be applied retroactively. Allstate's reasoning is faulty; it is confusing two different tests regarding retroactive application of a statute. In determining whether to apply a statute retroactively, a court must first determine whether the statute is a new law or merely a clarification. "A statute that merely clarifies, rather than changes, existing law is properly applied to transactions predating its enactment." (Carter v. California Dept. of Veterans Affairs (2006) 38 Cal.4th 914, 922.) In that case, "retroactivity" is something of a misnomer. An amendment clarifying a prior statute applies to pending cases because the amendment was, as it turns out, the law from the time of the original statute. (Id. at p. 923.) If, however, the statute sought to be applied is not a clarification but a new law, the court then considers whether the Legislature intended the law to apply retroactively - this is generally established only if the statute contains express language of retroactivity, or other sources clearly establish this was the legislative intent. (In re Marriage of McClellan (2005) 130 Cal.App.4th 247, 255.) Lubin held that the 2013 amendment was a clarification of existing law, not a new law to be applied retroactively. Therefore, there was no need for the Legislature to state an intent for retroactive application. We agree with Lubin's reasoning. The 2013 amendment applies, and the Wage Statement class is therefore not unmanageable.
The same flaw infects Allstate's argument that applying the 2013 amendment in this case would violate its due process rights. The due process inquiry applies only when the courts find a change in law that was intended to apply retroactively; when the statute in question was a mere clarification of existing law, there is no due process inquiry because there was no change in the law. (Carter v. California Dept. of Veterans Affairs, supra, 38 Cal.4th at p. 923.)
6. Conclusion
The trial court did not abuse its discretion in granting the motion for decertification on unmanageability. However, new law has undermined the trial court's conclusion with respect to the Wage Statement class only. We therefore reverse the court's order with respect to the Wage Statement class. Williams argues that because the Wage Statement class is now manageable without representative evidence, the Off the Clock class is also now manageable. We decline to decide this issue in the first instance on appeal. We leave the issue to the trial court.
DISPOSITION
The decertification order is affirmed with respect to the Off the Clock class and reversed with respect to the Wage Statement class, and the matter is remanded. The parties shall bear their own costs on appeal.
RUBIN, J. WE CONCUR:
BIGELOW, P. J.
FLIER, J.