Opinion

In this class action for unpaid overtime, Farmers Insurance Exchange (FIE) has filed separate notices of appeal from a judgment awarding the plaintiffs’ class $90,009,208.12 for unpaid overtime compensation, plus prejudgment interest of $32,303,048 and from a series of postjudgment orders that inter alia adopted a plan of distribution and awarded common fund attorney fees and costs. We reverse the portion of the judgment for unpaid double-time hours worked, and remand the order re plan of distribution. In all other respects the judgment and postjudgment orders are affirmed.

PROCEDURAL BACKGROUND

1. Early Procedural History

The plaintiffs are former or current claims representatives working for FIE in California who brought this action on behalf of themselves and other claims representatives employed in a similar capacity. Their complaint filed October 2, 1996, and amended two months later alleges that FIE followed the practice of refusing to pay overtime compensation to claims representatives on the ground that they were employed in an administrative capacity and therefore exempt from the overtime pay regulations of the California Industrial Welfare Commission. The plaintiffs sought unpaid overtime compensation under Labor Code section 1194 for a period beginning on October 1, 1993, and continuing until trial, as well as injunctive relief to assure future compliance with state overtime regulations.

After extensive discovery, plaintiffs filed a motion for class certification supported by depositions of prospective class members and other evidence. Granting the motion in an order filed May 28, 1998, the trial court found that the class was “so numerous as to make it impracticable to join all members,” there were “questions of law and fact common to the class,” plaintiffs were representative of the class, and a class action was “superior to other available methods for the fair and efficient adjudication of the controversy.” The order defined the class as consisting of claims representatives working for FIE’s Personal Lines Division in California during the relevant period, who were assigned to handle property, automobile physical damage, and liability claims.

FIE sought review of the order certifying the class in an earlier appeal, which we dismissed as procedurally improper.

Plaintiffs effectively sought an adjudication of FIE’s liability in a motion for summary adjudication of the exempt or nonexempt status of the class members. In an order filed April 21, 1999, the trial court rejected FIE’s claim that the claims representatives were employed in an exempt administrative capacity and ruled that the class was subject to the overtime regulations of the Industrial Welfare Commission. The court found that claims adjusting was “a product or service which FIE’s operation exists to provide” and that claims representatives “devote their time to carrying out [this] product/service as opposed to its ‘administrative’ functions.”

The trial court subsequently awarded the plaintiffs interim attorney fees on a finding that they had “prevailed on liability issues.” Appealing from this award, FIE challenged the adjudication of nonexempt status as a predicate for the attorney fee award and questioned the trial court’s statutory authority to award interim attorney fees. In our decision in Bell v. Farmers Ins. Exchange (2001) 87 Cal.App.4th 805 [ 105 Cal.Rptr.2d 59 ] (hereafter Bell II), we upheld the trial court’s ruling that plaintiffs were nonexempt employees but reversed the interim award of attorney fees on separate statutory grounds.

2. Contested Class Certification

On September 22, 1999, approximately six months after the summary adjudication of nonexempt status, FIE moved for “an order granting declaratory judgment as requested by plaintiffs . . . , decertifying the class and issuing an injunction requiring individual class members to commence administrative proceedings before the Labor Commissioner for a determination of individual damage claims.” FIE argued that damages could best “be litigated on an individual basis” in an administrative proceedings pursuant to Labor Code section 98. Opposing the motion, plaintiffs relied in part on the rule that class decertification must be based on a showing of changed circumstances. The trial court denied the motion on this basis, finding “no changed circumstances rendering class certification no longer proper.”

FIE next moved for the court to issue a second notice of pendency of action, both to employees who received the original notice mailed in July 1998 and those who were hired thereafter, which would give employees an opportunity to opt out of the class. The trial court ordered the notice, with a request for exclusion from the class, to be sent only to new employees who had not received the first notice. Approximately 5 percent of the class ultimately chose to opt out of the class action.

On May 24, 2001, less than a month before trial, FIE moved again to decertify the class on the ground that the recently completed depositions of the statistical sample of class members revealed that 9 percent of the class members did not claim unpaid overtime compensation. The trial court denied the motion on the ground that the inclusion of these employees in the class did not prejudice FIE and “judicial economy is served” by denying the motion.

3. Trial Management Rulings

Following the adjudication of the plaintiffs’ nonexempt status, the parties submitted to the court a series of proposals and counterproposals for management of the damages-phase trial, which led to an initial ruling entered April 28, 2000, that set the context for later pretrial rulings. In status conference statements filed in 1999, plaintiffs proposed to prove class-wide aggregate damages by statistical sampling. FIE took the “position that damages recoverable by each class member must be tried on an individual basis” and announced an intention to take an initial 52 depositions of an estimated 2,500 individuals in the class.

By February 2000, both parties were assisted by statisticians—plaintiffs retained Richard Drogin and FIE retained Roy Weinstein. Plaintiffs submitted a proposed trial management plan for a scheduled hearing on April 21, 2000, supported by a declaration of Richard Drogin, which proposed statistical sampling to achieve a one-hour-per-week margin of error with a “confidence interval” of 95 percent. He calculated that a sample of 95 employees would be necessary to achieve this level of accuracy. In response, FIE called for “individual mini-trials,” but, if they were not allowed, it proposed a sampling plan based on the same 95 percent confidence interval but designed to achieve a half-hour-per-week margin of error. Its statistician, Roy Weinstein, calculated that a sample of 1,325 employees would be required.

In reply, plaintiffs noted that the wide variation in estimated class size reflected different assumptions about the variability of the data. They proposed a two-stage plan. The parties would first conduct a pilot random sample of 50 class members to refine their assumptions as to the variability of the data, and then seek rulings on the appropriate margin of error and the size of the employee sample. Generally accepting this methodology, Weinstein filed a supplemental declaration shortly before the hearing on April 21, 2000, that characterized the plaintiffs’ proposal as “reasonable.” At the hearing itself, the parties announced an agreement on the initial step of taking depositions of a randomly chosen sample of 50 individuals.

The order following the hearing filed April 28, 2000, provided that “[tjhe damages-phase trial” would “determine classwide aggregate damages” by extrapolating from “a representative sampling of class members.” The order directed depositions of “an initial pilot sample of 50 randomly selected class members” and appointed a special master to resolve any disputes concerning these depositions. The court directed a target date for completion of these depositions and asked for the submission of a joint report of their experts. The order further directed the parties to “regularly meet-and-confer” following the depositions “in an attempt to reach agreement on the number of weekly overtime hours worked and/or the number of hours worked per week by each of the 50 deposed class members.”

Both parties filed status conference statements for a hearing on August 4, 2000. Plaintiffs proposed a margin of error of 1.5 overtime hours per week and FIE proposed a margin of error of 0.75 hours per week. In a joint declaration submitted for the hearing, the experts, Drogin and Weinstein, announced that they had “identified a range of sample sizes associated with different margins of error. A decision as to the appropriate margin of error for purposes of this litigation is left for the court’s discretion.” The margins of error ranged from one-half hour per week to two hours. Thus, a margin of error of 0.75 hours would call for a sample size of 529 and a margin of error of 1.5 hours would require a sample of 159. In either case, the results of the 50 depositions in the pilot sample could be deducted from the total number of individuals examined.

The experts’ calculations were made possible by the parties’ success in reaching agreement on work calendars reflecting the deposition testimony regarding work patterns of the 50 employees in the pilot sample. In addition, the parties “agreed upon a method [of] determining an appropriate measure for overtime hours worked” and calculated total overtime hours for each deponent. They then divided total overtime hours of all deponents by total weeks worked to arrive at “an estimate of average weekly overtime hours worked.”

At the August 4th hearing, the trial court suggested that the parties consider a one-hour weekly margin of error, which, according to the experts’ joint declaration, would require a total of 330 depositions. With the apparent agreement of the parties (see pp. 755-757, post), the court provisionally ordered 280 additional depositions to be taken, with this number to be reviewed after completion of 100 depositions.

Later in the year, Drogin and Weinstein filed a second joint declaration stating that they had reviewed the results of an additional 99 depositions (one transcript was not available) and calculated that a one-hour-per-week margin of error could be achieved with a sample size of only 286. Relying on the declaration, the trial court ordered an additional 136 depositions.

The parties ultimately exceeded this goal and deposed a total of 295 individuals. Remarkably, they reached agreement on work calendars reflecting the testimony of all deponents. (See p. 757, post.) The two experts both calculated that the average (or mean) weekly overtime recorded in the work calendars was 9.42 hours per week and they calculated the margin of error to be 0.9 hours per week, slightly better than originally contemplated.

FIE nevertheless argued that the sample was subject to varying interpretations. These differences in interpretation between plaintiffs and FIE framed the issues at trial.

4. Trial

In a brief jury trial, the plaintiffs presented only two expert witnesses, statistician Richard Drogin, and a certified public accountant, Paul Regan. FIE relied on a single witness, statistician Roy Weinstein.

Drogin recounted that, in early 2000, he began working with FIE’s expert, Weinstein, to develop a random sample for a statistical estimate of average overtime. He related the process of developing agreed work calendars for a sample of 295 employees and calculating overtime hours under California law. The parties reached virtually identical calculations of the total overtime hours in the sample, i.e., 438,000, and the total number of weeks the deponents worked within the class of claims representatives, i.e., 46,607. By simply dividing total overtime hours by total weeks, it could be calculated that the average weekly overtime was 9.42 hours—a figure that included 0.37 hours per week of work entitled to double-time compensation. Because the sample was representative of the population, Drogin testified that it was “statistically appropriate” to extrapolate the average number of overtime hours to the entire class. The margin of error for this extrapolation was plus or minus 0.9 of an hour per week, with a 95 percent level of confidence.

Regan testified that he secured FIE payroll records for all 2,402 class members for the period beginning October 1, 1993, and ending June 26, 2001. In a 625-page printout, he calculated overtime on a person-by-person basis using the average overtime figure of 9.42 hours per week and aggregated the individual calculations into a classwide total. He separately calculated unpaid overtime compensation reflecting time-and-a-half of base salary, overtime on bonus pay, and overtime reflecting a double-time rate of pay. The totals were $89,425,942, for time-and-a-half overtime, $1,265,255 for overtime on bonus pay, and $1,210,337 for double time. When Weinstein later offered a somewhat different interpretation of FIE’s payroll records in his trial testimony, Regan presented on rebuttal a revised calculation of $88,647,787 for time-and-a-half overtime on base salary, which gave effect to FIE’s interpretation of its records.

Through its expert witness, Weinstein, FIE presented a case for determining aggregate overtime on the basis of median weekly overtime of 7.27 hours. Weinstein offered two reasons for using the median figure rather than the mean which was used by Drogin. First, he maintained that the mean may be misleading when a sample includes extreme values, or outliers, deviating widely from other data. He stressed the variation in weekly overtime reported by the 295 deponents, including 11 employees who reported working an average of 26 hours or more of overtime. Secondly, he argued that the sample results were distorted upward by “response bias.” Some 36 claims representatives did not respond to deposition notices and subpoenas, forcing the parties to depose another group of 36 randomly selected class members. He inferred that those who responded to the notices felt more positively about the plaintiffs’ claims than those who failed to respond and therefore the substitution affected “the integrity of the original sample” by introducing an upward bias.

The case was submitted to the jury on the afternoon of July 9, 2001. The next day, the jury unanimously returned a special verdict finding that unpaid time-and-a-half overtime compensation owed to the class as a whole was $88,798,871.12 and unpaid double-time overtime compensation was $1,210,337. The former figure corresponded closely to Regan’s revised calculation of overtime on base salary but excluded overtime on bonus pay; the latter figure accepted Regan’s calculation of unpaid double-time compensation.

5. Posttrial Proceedings

Shortly after the jury verdict, plaintiffs proposed three options for a plan of distribution: (1) a distribution attributing the average weekly overtime hours to all employees and calculating individual payments solely on the basis of employment tenure and salary, (2) a distribution preceded by the mailing of claim forms requiring class members to say if they claimed “some” damages or none, and (3) a detailed claim form with an administrative proof-of-claims procedure. Plaintiffs argued that FIE had no standing to intervene in the plan of distribution. For its part, FIE called for the third option of an administrative proof-of-claims procedure.

Without directly ruling on FDE’s right to object, the trial court stated at a hearing on August 1, 2001, that it would allow FIE to comment on the plan of distribution as a matter of courtesy and directed the parties to meet and confer to resolve issues presented by the third option. In two later hearings, the plaintiffs developed a revised version of the third option reflecting FIE’s written objections and arguments.

On September 20, 2001, the court entered an order adopting plaintiffs’ revised plan of distribution, subject to certain further modifications. The plan provided for the distribution of a damages fund consisting of the award for unpaid overtime ($90,009,208.12), plus prejudgment interest, less costs and fees payable to class counsel as a percentage of the common fund and “service payments” to the five named plaintiffs compensating them for their efforts in bringing the action. Any statutory attorney fees and costs awarded under Labor Code section 1194 would be offset against attorney fees and costs payable from the common fund. The court entrusted administration of the plan of distribution to a claims administrator and retained jurisdiction to oversee its administration.

The plan provided that payments to the 295 employees in the random sample, as well as 48 additional employees deposed during the damages phase of the litigation, would be based on their deposition testimony of overtime hours worked. Other payments would be made only to those employees submitting proof-of-claim forms with statements of average weekly overtime hours worked. The payments would be based on these statements of overtime worked, as well as the wages and tenure of the employees as class members. The information regarding employees’ wages and tenure would be drawn from the FIE database and preprinted on the claim forms mailed to class members, who would be given an opportunity to challenge it. Any unresolved disputes would be referred to a special master.

In the event claims should exceed the net damages fund, the claims administrator would distribute to each class member a percentage of the fund equivalent to that class member’s claim share as a proportion of all class members’ claim shares. If claims should be less than the net damages fund, the judgment gave FIE a right to recover undistributed prejudgment interest but deferred a decision on any unpaid residue of the damages fund itself to a later proceeding under Code of Civil Procedure section 384.

A judgment filed on September 24, 2001, adopted by reference the plan of distribution. Two subsequent postjudgment orders addressed important aspects of the plan of distribution. First, on November 28, 2001, the court awarded to plaintiffs’ counsel attorney fees in the amount of 25 percent of the total damages fund recovered for the class, consisting of back pay and interest, plus expenses in the sum of $941,884.73. Secondly, on April 23, 2002, the court appointed a claims administrator and approved a claim form, with an accompanying background statement and claim form instruction.

The background statement accompanying the claim form explained that the employees were entitled to money under the terms of a judgment awarding damages for unpaid overtime compensation to claims representatives working for FIE in California. The amount of the damage fund was determined by deposing a representative sample of 295 employees regarding their overtime hours. The statement informed employees that the deposed employees worked an average of 9.42 overtime hours per week and set forth in a chart the precise numbers of employees reporting at intervals between zero to 38 hours of overtime, i.e., 39 employees worked between 4.01 and 6 hours of overtime, 35 between 6.01 and 8 hours, etc.

The instructions gave precise directions on how to calculate overtime hours worked. The estimate of weekly overtime hours worked was to be based on the employee’s “best recollection, but it need not be precise if [the employee did] not have precise records or recollections.” By signing the claim form, the employee was “declaring under penalty of perjury” that the information provided was correct to the best of his or her knowledge.

DISCUSSION

A. Law of the Case

In its first assignment of error, FIE urges that we reconsider our decision in the prior appeal in light of “[r]ecent developments in federal law.” As FIE recognizes, the reconsideration of a prior appeal is ordinarily precluded by the law-of-the-case doctrine. “Under this doctrine, ‘the decision of an appellate court, stating a rule of law necessary to the decision of the case, conclusively establishes that rule and makes it determinative of the rights of the same parties in any subsequent retrial or appeal in the same case.’ [Citation.]” (Nally v. Grace Community Church (1988) 47 Cal.3d 278, 301-302 [ 253 Cal.Rptr. 97 , 763 P.2d 948 ]; see also Adams v. Pacific Bell Directory (2003) 111 Cal.App.4th 93, 97-100 [ 3 Cal.Rptr. 3d 365 ].) The law of the case, however, is a procedural rule that is subject to certain exceptions. (George Arakelian Farms, Inc. v. Agricultural Labor Relations Bd. (1989) 49 Cal.3d 1279, 1291 [ 265 Cal.Rptr. 162 , 783 P.2d 749 ].)

“The principal ground for making an exception to the doctrine of law of the case is an intervening or contemporaneous change in the law.” (Clemente v. State of California (1985) 40 Cal.3d 202, 212 [ 219 Cal.Rptr. 445 , 707 P.2d 818 ].) The exception is limited to changes in “the controlling rules of law.” (Salazar v. Eastin (1995) 9 Cal.4th 836, 859 [ 39 Cal.Rptr.2d 21 , 890 P.2d 43 ], italics added; DiGenova v. State Board of Education (1962) 57 Cal.2d 167, 179-180 [ 18 Cal.Rptr. 369 , 367 P.2d 865 ].) As explained in Gore v. Bingaman (1942) 20 Cal.2d 118, 123 [ 124 P.2d 17 ], “the rule should never be made the instrument of injustice. Thus, where the controlling rules of law have been altered or clarified in the interval between the first and second appeal and adherence to the previous decision would result in defeating a just cause, it has been held that the court will not hesitate to reconsider its prior determination.” (See also Davidson v. Superior Court (1999) 70 Cal.App.4th 514, 530 [ 82 Cal.Rptr.2d 739 ]; Building Industry Assn. v. City of Oceanside (1994) 27 Cal.App.4th 744, 759-760 [ 33 Cal.Rptr.2d 137 ].)

FIE predicates its claim of an intervening change in the controlling rules of law on an opinion letter of the Administrator of the Department of Labor’s Wage and Hour Division, two appellate court decisions and one district court decision arising from the Ninth Circuit (Bothell v. Phase Metrics, Inc. (9th Cir. 2002) 299 F.3d 1120 ; Webster v. Public School Employees of Washington (9th Cir. 2001) 247 F.3d 910 ; Palacio v. Progressive Ins. Co. (C.D.Cal. 2002) 244 F.Supp.2d 1040 ), and two other federal district court decisions (Jastremski v. Safeco Ins. Companies (N.D.Ohio 2003) 243 F.Supp.2d 743 ; Schaefer v. Indiana Michigan Power Co. (W.D.Mich. 2002) 197 F.Supp.2d 935 ). To assess the significance of these authorities, we must first review our prior decision.

In Bell II, we upheld the trial court’s summary adjudication that claims representatives working for FIE’s Personal Lines Division in California were not exempt from the overtime pay requirements of the Industrial Welfare Commission’s wage order No. 4, codified in California Code of Regulations, title 8, section 11040, subdivision 1(A). Our analysis proceeded in four steps. First, we construed the exemption as requiring an employer to prove both that an employee is employed in an “administrative capacity” and that the employee is “ ‘engaged in work which is primarily intellectual, managerial, or creative, and which requires exercise of discretion and independent judgment.’ ” (Bell II, supra, 87 Cal.App.4th 805, 810-812 .) Secondly, we concluded that federal authorities construing parallel provisions of the Fair Labor Standards Act of 1938 ( 29 U.S.C. § 201 et seq. (ELSA)) were relevant to construing the exemption provision of wage order No. 4. (Bell II, supra, at pp. 812-819.) Thirdly, we found that a core concept appearing in the federal interpretative regulations and decisional law—the administrative/production worker dichotomy—provided a useful approach to determining whether an employee worked in an “administrative capacity” as required by the exemption provision of wage order No. 4. (Bell II, supra, at pp. 819-823.) Fourthly, we found that the evidence submitted on the motion for summary adjudication compelled the conclusion that the plaintiff employees were production workers within the meaning of the administrative/production worker dichotomy and, on that ground, we held that they were not employed in an administrative capacity within the meaning of the exemption of title 8, section 11040, subdivision 1(A). (Bell II, supra, at pp. 823-829.)

In this appeal, FIE primarily directs its argument at the third and fourth parts of the opinion. The analysis in the third section of the opinion evolved from our prior conclusions that the Industrial Welfare Commission (IWC) intended the term “administrative capacity” to have an independent meaning and that the parallel provisions of federal law could provide guidance in determining that intended meaning. We noted that federal interpretative regulations relating to the term “administrative capacity” could be grouped into those relating to “the role of administrative employees in a business enterprise, the actual duties of the employees, and the employees’ level of remuneration.” (Bell II, supra, 87 Cal.App.4th 805, 819 .) Of these three approaches, it was the body of federal interpretative law dealing with the employee role in the business enterprise that served to give the term “administrative capacity” a meaning independent of other language in title 8, section 11040, subdivision 1(A). A central concept in this body of interpretative law is the administrative/production worker dichotomy.

The administrative/production worker dichotomy draws on 29 Code of Federal Regulations (hereafter C.F.R.) part 541.2 (2003), which defines the statutory phrase “administrative capacity” as referring to an employee whose primary duty consists of “work directly related to management policies or general business operations of his employer . . . .” (See also § 541.214(a).) As elucidated in part 541.205(a), this language calls for a distinction between “those types of activities relating to the administrative operations of a business as distinguished from ‘production’ or, in a retail or service establishment, ‘sales’ work.”

This dichotomy suggested by 29 C.F.R. part 541.205(a) has become a generally accepted framework of analysis in the decisional law construing the statutory term “administrative capacity” in the FLSA. The cases distinguish between “ ‘those employees whose primary duty is administering the business affairs of the enterprise from those whose primary duty is producing the commodity or commodities, whether goods or services, that the enterprise exists to produce and market.’ ” (Bell II, supra, 87 Cal.App.4th 805, 821 , citing Dalheim v. KDFW-TV (5th Cir. 1990) 918 F.2d 1220, 1230 , fn. omitted.) Thus, in Reich v. Chicago Title Ins. Co. (D.Kan. 1994) 853 F.Supp. 1325, 1330 , the court “examined the duties of the plaintiff escrow closers ‘to determine whether they carry out Chicago Title’s day-to-day operations [as production workers] ... or whether they administer the business affairs . . . [of the company as administrative workers].’ ” (See Bell II, supra, at pp. 822-823.)

The fourth section of our opinion in Bell II reviewed the evidence concerning the work of claims representatives in the California branch claims offices of FIE’s Personal Lines Division. We concluded; “the undisputed evidence places the work of the claims representatives squarely on the production side of the administrative/production worker dichotomy. The undisputed evidence establishes that claims adjusting is the sole mission of the 70 branch claims offices where plaintiffs worked. The claims representatives are fully engaged in performing the day-to-day activities of that important component of the business.” (Bell II, supra, 87 Cal.App.4th 805, 826 .)

Turning to the propriety of the summary adjudication, we noted that “the administrative/production worker dichotomy is a somewhat gross distinction that may not be dispositive in many cases.” (Bell II, supra, 87 Cal.App.4th 805, 826 .) While federal interpretative regulations served to elucidate and qualify the distinction, there are no comparable interpretative regulations governing IWC work orders. Accordingly, we stressed “that California courts must use great caution in granting summary judgment or summary adjudication on the basis of such a broad distinction as the administrative/production worker dichotomy.” (Id, at p. 827.) Nevertheless, we found that the evidence in the record before us was sufficiently clear to justify the trial court’s summary adjudication. Our opinion in Bell II was based on the restricted record before us and cannot be read out of that context.

The record showed that the claims representatives processed “a large number of small claims” and, according to FIE’s own characterization, their duties were restricted to “ ‘the routine and unimportant.’ ” (Bell II, supra, 87 Cal.App.4th 805, 827-828 .) We concluded that these work activities “place[dj plaintiffs in the sphere of rank and file production workers” in this part of the FIE business organization. (Bell II, supra, at p. 828.) Our conclusion obviated the need to inquire whether the employees’ duties came within the criteria of subdivision 1(A)(1) of wage order No. 4, “that is, whether the plaintiffs are ‘engaged in work which is primarily intellectual, managerial, or creative, and which requires exercise of discretion and independent judgment’ . . . .” (Bell II, supra, at p. 828.)

We find nothing in Webster v. Public School Employees of Washington, supra, 247 F.3d 910 , and Bothell v. Phase Metrics, Inc., supra, 299 F.3d 1120 , that is inconsistent with our decision. The decisions both pursued the administrative/production worker dichotomy as a generally accepted framework of analysis, but the factual situations in the two cases did not easily lend themselves to this customary analysis.

In Webster , the plaintiff was a field representative for a public employees’ union. Seeking to put himself on the production side of the dichotomy, he argued that he assisted in performing the primary service goal of the unions—the negotiation of collective bargaining units. The court observed that the purpose of the dichotomy is to clarify the meaning of “ ‘work directly related to the management policies or general business operations’ ” as found in 29 C.F.R. part 541.2. (Webster v. Public School Employees of Washington, supra, 247 F.3d 910, 916 .) The dichotomy would defeat the purpose of the administrative exemption if it were applied to work units carrying out management functions. Following Bratt v. County of Los Angeles (9th Cir. 1990) 912 F.2d 1066 , the court found that the plaintiff was an exempt administrative employee because his primary duty involved advising the bargaining units how to conduct their business. (Webster v. Public School Employees of Washington, supra, at pp. 916-917.)

In Bothell , the plaintiff was a field service engineer for one of his employer’s largest customers. He claimed to be a repairman; his employer characterized him as an account executive. Applying the administrative/production worker dichotomy, the district court granted summary judgment on the ground that the job was “ancillary” to his employer’s main activities and therefore did not fall on the production side of the dichotomy. The appellate court observed that the dichotomy “is useful only to the extent that it helps clarify the phrase ‘work directly related to the management policies or general business operations.’ [Citation.]” (Bothell v. Phase Metrics, Inc., supra, 299 F.3d 1120, 1126.) But “[ojnly when work falls ‘squarely on the “production” side of the line,’ has the administration/production dichotomy been determinative.” (Id. at p. 1127, fn. omitted.) Remanding the case to the district court for trial, the court held that the plaintiff’s status called for a “fact-specific inquiry” into the plaintiff’s work activities. (Id. at p. 1128.)

Webster and Bothell underscore our observations in Bell II that “the administrative/production worker dichotomy is a somewhat gross distinction that may not be dispositive in many cases.” (Bell II, supra, 87 Cal.App.4th 805, 826 .) But the decisions do not draw into question our use of the dichotomy under the specific facts of the present case. The work at issue here falls “ ‘squarely on the “production” side of the line’ ” (Bothell v. Phase Metrics, Inc., supra, 299 F.3d 1120, 1127); it places the plaintiffs in a broad category of rank and file workers carrying out the service that an important component of the FIE business organization exists to provide.

In Palacio v. Progressive Ins. Co., supra, 244 F.Supp.2d 1040 , the federal district court reached a result different from Bell II , but pursued the same mode of analysis and distinguished the facts of the case from those in Bell II . The plaintiff, a claims adjuster for an automobile insurance company, brought actions for unpaid overtime compensation under the FLSA and wage order No. 4 of the California Industrial Welfare Commission. The court granted summary judgment for the employer under the ELSA, but denied the motion for summary judgment under wage order No. 4, relying on the authority of Bell II .

In deciding the ELSA claim for the employer, the court analyzed the case in terms of the administrative/production worker dichotomy but found that plaintiff’s job came within the employer’s administrative operations as defined in 29 C.F.R. part 541.205(b) (“The administrative operations of the business include the work performed by . . . employees engaged in ‘servicing’ a business as, for example, advising the management, . . . negotiating, representing the company . . .”).

Turning to the state claim, the court noted that, in applying the administrative/production worker methodology, Bell II reached an opposite conclusion “upon considering somewhat similar circumstances.” (Palacio v. Progressive Ins. Co., supra, 244 F.Supp.2d 1040, 1050 .) But the court did not question the authority of Bell II under state law and observed that it presented distinguishable facts. In particular, “claims handling [in Bell II] constituted the insurance company’s business” (id. at p. 1050) and the employer restricted the claims representatives to “routine and unimportant” duties. (Ibid.) In the court’s view, the similarity and differences between the cases revealed a triable issue of fact under state law.

FIE relies on another federal district court decision, Schaefer v. Indiana Michigan Power Co., supra, 197 F.Supp.2d 935 , which has no factual similarity to the present case. The plaintiff was an environmental specialist at a nuclear power plant and was engaged in supervisory and policy-formulation work, analogous to the kind of work that Bell II expressly notes that plaintiffs did not perform. (Bell II, supra, 87 Cal.App.4th 805, 823, 828 .) The Schaefer holding that the plaintiff was an exempt employee is entirely consistent with Bell II .

Lastly, FIE cites Jastremski v. Safeco Ins. Companies, supra, 243 F.Supp.2d 743 , a federal district court decision that is in fact irreconcilable with Bell II . The decision is notable chiefly because it relies on the opinion letter dated November 19, 2002, by the Administrator of the Wage and Hour Division of the Department of Labor (hereafter DOE Opinion Letter) in denying the claim of an insurance adjuster for unpaid overtime under the ELSA.

The DOL Opinion Letter is based on a request by an association of insurance companies for an opinion confirming “that insurance claims adjusters are exempt from the overtime requirements of the Fair Labor Standards Act” and is premised on a description of the claims adjusters’ work provided by the insurance companies. The letter describes the adjusters as being “primarily responsible for resolving a claim.” They “weigh the factual information” and “evaluate whether there is coverage for the claim under the policy.” They have full authority to settle claims within “their established authority,” which “ranges from a minimum of $3,000 to more than $50,000.” On claims above their personal limits, they make recommendations to supervisors, “which are generally approved,” pursue negotiations, and “advise the company on whether to settle the claim or to pursue litigation.”

The DOL Opinion Letter concludes that the claims adjusters’ duties are “administrative in nature” on the basis of an analysis of 29 C.F.R. part 541.205. It characterizes the adjusters as being “ ‘engaged in “servicing” a business’ ” within the meaning of part 541.205(b) and states that the Wage and Hour Division “has long recognized that claims adjusters typically perform work that is administrative in nature.” The opinion then proceeds to find that the adjusters’ work is of “substantial importance to the company” within the terms of part 541.205(a). While the opinion notes that part 541.205(a) distinguishes the “administrative operations of a business” from production work, it does not mention the body of federal law applying the administrative/production worker dichotomy derived from that provision and, in fact, it cites only one judicial decision—Palacio v. Progressive Ins. Co., supra, 244 F.Supp.2d 1040 .

The deference to be given to DOL opinion letters is governed by Christensen v. Harris County (2000) 529 U.S. 576 [ 146 L.Ed.2d 621 , 120 S.Ct. 1655 ], which follows an earlier decision, Skidmore v. Swift & Co. (1944) 323 U.S. 134 [ 89 L.Ed. 124 , 65 S.Ct. 161 ]. Distinguishing opinion letters from “formal adjudication or notice-and-comment rulemaking,” the decision explains that interpretations contained in the letters “lack the force of law,” but are “ ‘entitled to respect’ ... to the extent that those interpretations have the ‘power to persuade’. . . .” (Christensen, supra, at p. 587 , citation omitted.) In a more complete statement of this point, the Skidmore decision states the opinions “do constitute a body of experience and informed judgment to which courts and litigants may properly resort for guidance. The weight of such a judgment in a particular case will depend upon the thoroughness evident in its consideration, the validity of its reasoning, its consistency with earlier and later pronouncements, and all those factors which give it power to persuade, if lacking power to control.” (Skidmore, supra, at p. 140 .)

Since DOL opinion letters lack the force of law, they may be rejected categorically as a source of an intervening change of law under the law-of-the-case doctrine. Moreover, in our view, the DOL Opinion Letter here does not offer persuasive reasons to reconsider our decision in Bell II . First, the letter is based on distinguishable facts. The evidence in Bell II disclosed that the claims adjusters in the Personal Lines Division were “ordinarily occupied in the routine of processing a large number of small claims.” (Bell II, supra, 87 Cal.App.4th 805, 827-828 .) With few exceptions, their settlement authority was set “at $15,000 or lower and often [was] $5,000 or lower.” (Id. at p. 828.) For the most part, their work consisted of gathering facts, filling out appropriate forms, and “ ‘communicating with policy holders and third party claimants about the indemnity value of the claim ....’” (Id. at p. 826.) “On matters of relatively greater importance,” they acted as “conduits of information to supervisors” or as “ ‘go-betweens’ in conveying information to [an] attorney.” (Id. at p. 828.)

Secondly, the persuasive power of the DOL Opinion Letter is undermined by its failure to review the body of federal decisions applying the administrative/production worker dichotomy derived from 29 C.F.R. part 541.205(a). In this respect, it appears to represent a departure from past opinion letters of the DOL Wage and Hour Division that have repeatedly employed this analysis. The administrative/production worker dichotomy, as Bell II notes, offers a common frame of reference for federal decisions in this area and has led to summary judgment for the employees in á line of decisions.

Thirdly, the statement that the DOL has “long recognized” the work of claims adjusters to be “typically” administrative in nature is overly simplified. In Bell II , we noted a 1940 DOL report stating that “ ‘the term “claim agent” may cover a great variety of employees.’ ” (Bell II, supra, 87 Cal.App.4th 805, 827 , citation omitted.) As indicated in 29 C.F.R. part 541.205(c)(5), the job category includes exempt employees employed in the capacity of “advisory specialists and consultants.” But the duties of the claims representatives in Bell II closely resembled an “inspector for an insurance company,” listed as a nonexempt employee in 29 C.F.R. part 541.205(c)(2). Moreover, 29 C.F.R. part 778.405 lists insurance adjusters as an example of those nonexempt employees with irregular hours who may enter into contracts guaranteeing constant pay for varying workweeks under section 7(f) of the FLSA.

Much of the argumentation of FIE and amici curiae seeks to demonstrate that our decision in Bell II was in error. These arguments cannot easily be reconciled with the law-of-the-case doctrine. Our high court has stated that the doctrine demands that the court adhere to a principle or rule of law enunciated in an earlier appeal of the same case even though it may “be clearly of the opinion that the former decision is erroneous in that particular.” (Tally v. Ganahl (1907) 151 Cal. 418, 421 [90 R 1049].) But, stretching the limits of the change-of-law exception to the doctrine, FIE argues that the five recent federal decisions and the DOL Opinion Letter provide further clarification of the applicable state law, revealing errors in our analysis, and that a reconsideration of the decision is required to avoid injustice to California employers. We think that we have adequately answered this argument by showing that the federal appellate decisions in Webster and Bothell , as well as the district court decision in Schaefer , are entirely consistent with our analysis in Bell II . The Palacio decision also is basically compatible with our opinion because it applies the same analytical methodology on distinguishable facts. FIE’s arguments rest on the questionable relevance of the DOL Opinion Letter to interpretation of wage order No. 4 and on the single federal district court decision (Jastremski) that has cited the DOL Opinion Letter as authority.

Nevertheless, we will reply to FIE’s arguments on the merits. FIE and amici curiae pursue four lines of argument: (1) the Bell II opinion “treat[s] employees’ job duties as irrelevant,” (2) it causes relief to turn on FIE’s organizational structure and the geographic location of the plaintiffs’ jobs, (3) it misconstrues “the product of an insurance company,” and (4) it fails to recognize the “ancillary” nature of the plaintiffs’ services.

The first argument is based on an apparent misreading of the opinion. In Bell II we held that wage order No. 4 required proof both of employment in an administrative capacity and of job duties meeting certain criteria (“ ‘work which is primarily intellectual, managerial, or creative, and which requires exercise of discretion and independent judgment’ ”). Concluding that plaintiffs were not employed in an administrative capacity, we did not reach the criteria relating to job duties in the second prong of the exemption or explore the corresponding body of federal interpretative law. We did not, however, treat the plaintiffs’ job duties as irrelevant. To the contrary, we examined the plaintiffs’ jobs using the expression “duties,” as well as equivalent terms such as “activities,” “work,” “mission,” and “function,” to describe the jobs in the context of the employer’s business.

The second argument similarly reflects a misunderstanding of the opinion. We see nothing in Bell II that would suggest it “tum[s] on the happenstance of geographic location/corporate structure.” The decision engages in a review of the undisputed facts relevant to an adjudication of the plaintiffs’ administrative capacity. Holding that plaintiffs were not employed in an administrative capacity, we carefully phrased our opinion in terms of these facts, that is, on “an analysis of the peculiar nature of FIE’s business and the claims representatives’ role in that business . . . .” (Bell II, supra, 87 Cal.App.4th 805, 829 .) But the opinion does not assign to particular facts, such as those concerning corporate structure or the plaintiffs’ work location, a significance exceeding their actual relevance to the issue of administrative capacity.

Indeed, the opinion never went beyond the strictly factual observation that the plaintiff claims representatives worked in branch offices, which performed none of the activities constituting the administrative or executive functions of the corporation. These activities were delegated to other corporate entities or performed elsewhere. The work performed by the plaintiff claims representatives in branch offices was devoted entirely to carrying out one important component of the business—claims adjustment—and was limited to the routine aspects of performing that function. We see nothing in our discussion of the undisputed facts to suggest that the determination of administrative capacity turns on the location or departmental organization of the place of employment.

The third argument is based on an abstract characterization of the “product of an insurance company” and deduces from this characterization that plaintiffs were not production workers. It is argued, for example, “that the product of an insurance company is its policies, not claims adjusting,” and therefore claims adjusters cannot be production workers. Amici curiae construe the decision in Reich v. John Alden Life Ins. Co. (1st Cir. 1997) 126 F.3d 1, 9-10 , as supporting their reasoning. But as we read the John Alden decision, it explores the significance of a stipulation of facts. We decline to follow two federal district court decisions that have relied on the John Alden dicta to deny overtime claims. (Wilshin v. Allstate Ins. Co. (M.D.Ga. 2002) 212 F.Supp.2d 1360, 1376 ; Hogan v. Allstate Ins. Co. (M.D.Fla. 2002) 210 F.Supp.2d 1312, 1320 .)

The relevant issue concerns the propriety of the administrative/production worker dichotomy as an analytic tool. We recognized in Bell II that the dichotomy may lose its utility as it is applied to small or specialized segments of a business. (Bell II, supra, 87 Cal.App.4th 805, 826 ; see also Bothell v. Phase Metrics, Inc., supra, 299 F.3d 1120, 1126 .) But the federal decisions applying the dichotomy to grant relief to particular employee groups do not tie its use to a global characterization of the employer’s business. In Bell II , we found Reich v. Chicago Title Ins. Co., supra, 853 F.Supp. 1325 , to be “especially instructive.” (Bell II, supra, at p. 822.) Applying the dichotomy to grant summary judgment to escrow closers, the court observed that the employer’s “ ‘status as a title insurer does not alter the fact that escrow closings are a very real product . . . which it markets and sells separate from ... its overall title insurance operations.’ [Citation.]” (Id. at p. 823.) We consider that, under the facts of the present case, the administrative/production worker dichotomy may similarly be applied to the work of claims adjusting in “an important component of [FIE’s] business organization” notwithstanding the existence of other business activities in the organization. (Id. at p. 828.)

In a related argument, amici curiae argue that claims adjusting is ancillary to the business function of an insurance company and therefore administrative in nature. The argument relies on a short line of decisions stemming from a comment in Martin v. Cooper Elec. Supply Co., supra, 940 F.2d 896 . Affirming a lower court finding that salespersons did not come within 29 C.F.R. part 541.205(b), i.e., administrative employees “servicing” a business, the court observed, “ ‘Servicing’ a business within the meaning of 29 C.F.R. § 541.205 (b) denotes employment activity ancillary to an employer’s principal production activity .... It follows that inside salespersons do not ‘service’ Cooper’s business by making wholesale sales—wholesale sales is Cooper’s business.” (Martin v. Cooper Elec. Supply Co., supra, at pp. 904-905.) In other words, the employees did not perform ancillary activities and hence did not come within this subcategory of administrative employees.

A later case, Reich v. John Alden Life Ins. Co., supra, 126 F.3d 1 , applied this language to the converse fact situation: it found that the employees did perform an ancillary function and hence could be regarded as engaged in “administrative ‘servicing’ within the meaning of section 541.205(b).” (Reich v. John Alden Life Ins. Co., supra, at p. 10; see also Haywood v. North American Van Lines, Inc. (7th Cir. 1997) 121 F.3d 1066, 1072 .) The federal district court decision in Palacio v. Progressive Ins. Co., supra, 244 F.Supp. 1040 , contains language that can be regarded as taking the concept of “ancillary” services still further to limit the application of the “administrative/production dichotomy” in 29 C.F.R. part 541.205(a). (Palacio v. Progressive Ins. Co., supra, at p. 1047.)

In Bothell v. Phase Metrics, Inc., supra, 299 F.3d 1120, 1126 , the court recognized that the concept of “ancillary” services arises as a gloss on 29 C.F.R. part 541.205(b), though “[s]ome cases do use the term ‘ancillary’ as a short-hand description of administrative activities.” Like the Bothell court, we think the concept of “ancillary” services should be confined to its original context as an explication of 29 C.F.R. part 541.205(b). With this limited reference, it has no bearing on our finding that the plaintiffs belong “in the sphere of rank and file production workers.” (Bell II, supra, 87 Cal.App.4th 805, 828 .)

B. Amendment to Wage Order No. 4

Lastly, FEE argues that the judgment should be set aside with respect to any damages or injunctive relief applying after the amendment of wage order No. 4 on October 1, 2000. We note that FEE never raised this argument in the trial court and therefore has waived any assignment of error on appeal. (9 Witkin, Cal. Procedure (4th ed. 1997) Appeal, § 394, p. 444.)

We conclude that reconsideration of our decision in Bell II is precluded by the law-of-the-case doctrine. FEE has failed to show an intervening change in the controlling rules of law that would justify a departure from this procedural rule.

C. Class Certification

1. Legal Background

FEE next contends that the trial court abused its discretion in initially certifying the class of claims representatives on May 28, 1998, and later denying the motions for decertification filed on September 22, 1999, and May 24, 2001. The contention requires us to briefly review the principles governing the certification of class actions in California.

The class action in California evolved from “the equitable doctrine of virtual representation which ‘ “rests upon considerations of necessity and paramount convenience, and was adopted to prevent a failure of justice.” ’ [Citations.]” (Daar v. Yellow Cab Co. (1967) 67 Cal.2d 695, 703-704 [ 63 Cal.Rptr. 724 , 433 P.2d 732 ].) The only statutory authorization for the class action consists of a phrase at the end of a statute dealing with compulsory joinder. Code of Civil Procedure section 382 states: “. . . and when the question is one of a common or general interest, of many persons, or when the parties are numerous, and it is impracticable to bring them all before the court, one or more may sue or defend for the benefit of all.” Reflecting a “general support of class actions” (City of San Jose v. Superior Court (1974) 12 Cal.3d 447, 459 [ 115 Cal.Rptr. 797 , 525 P.2d 701 ), the California Supreme Court “has urged trial courts to be procedurally innovative” in determining whether to allow class suits, “encouraging them to incorporate procedures from outside sources,” in particular “rule 23 of the Federal Rules of Civil Procedure .... [28 U.S.C.]” (City of San Jose v. Superior Court, supra, at p. 453, fn. omitted.)

California courts have long “held that two requirements must be met in order to sustain any class action: (1) there must be an ascertainable class [citations]; and (2) there must be a well defined community of interest in the questions of law and fact involved affecting the parties to be represented.” (Daar v. Yellow Cab Co., supra, 67 Cal.2d 695, 704 ; Linder v. Thrifty Oil Co. (2000) 23 Cal.4th 429, 435 [ 97 Cal.Rptr.2d 179 , 2 P.3d 27 ]; Richmond v. Dart Industries, Inc. (1981) 29 Cal.3d 462, 470 [ 174 Cal.Rptr. 515 , 629 P.2d 23 ]; Vasquez v. Superior Court (1971) 4 Cal.3d 800, 809 [ 94 Cal.Rptr. 796 , 484 P.2d 964 ].) While the requirement of an ascertainable class normally requires little elaboration, the community of interest requirement has been held to embody three factors: “(1) predominant common questions of law or fact; (2) class representatives with claims or defenses typical of the class; and (3) class representatives who can adequately represent the class.” (Richmond v. Dart Industries, Inc., supra, at p. 470; Linder v. Thrifty Oil Co., supra, at p. 435; Bartold v. Glendale Federal Bank, supra, 81 Cal.App.4th 816, 828 ; Civ. Code, § 1781, subd. (b).)

The certification of a class is a discretionary decision that demands the weighing of many relevant considerations. (Linder v. Thrifty Oil Co., supra, 23 Cal.4th 429, 435 .) In Collins v. Rocha (1972) 7 Cal.3d 232, 238 [ 102 Cal.Rptr. 1 , 497 P.2d 225 ], our high court observed, “The ultimate question in every case of this type is whether, given an ascertainable class, the issues which may be jointly tried, when compared with those requiring separate adjudication, are so numerous or substantial that the maintenance of a class action would be advantageous to the judicial process and to the litigants.” Following this language, numerous later cases have “admonished trial courts to carefully weigh respective benefits and burdens and to allow maintenance of the class action only where substantial benefits accrue both to litigants and the courts.” (City of San Jose v. Superior Court, supra, 12 Cal.3d 447, 459 ; see also Lockheed Martin Corp. v. Superior Court (2003) 29 Cal.4th 1096 , 1104—1105 [ 131 Cal.Rptr.2d 1 , 63 P.3d 913 ]; Washington Mutual Bank v. Superior Court (2001) 24 Cal.4th 906, 914 [ 103 Cal.Rptr.2d 320 , 15 P.3d 1071 ]; Blue Chip Stamps v. Superior Court (1976) 18 Cal.3d 381, 385 [ 134 Cal.Rptr. 393 , 556 P.2d 755 ]; Caro v. Procter & Gamble Co. (1993) 18 Cal.App.4th 644, 655 [ 22 Cal.Rptr.2d 419 ].)

By preventing “a failure of justice in our judicial system” (Linder v. Thrifty Oil Co., supra, 23 Cal.4th 429, 434 ), the class action not only benefits the individual litigant but serves the public interest in the enforcement of legal rights and statutory sanctions. (Cf. Bartold v. Glendale Federal Bank, supra, 81 Cal.App.4th 816, 830-831 .) Thus, in upholding a consumer class action, the court in Vasquez v. Superior Court, supra, 4 Cal.3d 800, 807, quoted with approval an argument for stockholder class action suits: “ ‘Modem society seems increasingly to expose men to . . . group injuries for which individually they are in a poor position to seek legal redress, either because they do not know enough or because such redress is disproportionately expensive. If each is left to assert his rights alone if and when he can, there will at best be a random and fragmentary enforcement, if there is any at all.’ ”

In a proper case, “the judicial system substantially benefits by the efficient use of its resources” afforded by class certifications. (Richmond v. Dart Industries, Inc., supra, 29 Cal.3d 462, 474 .) Class actions offer a means of avoiding “repetitious litigation” (Linder v. Thrifty Oil Co., supra, 23 Cal.4th 429, 435 ) and “a multiplicity of legal actions dealing with identical basic issues . . . .” (Vasquez v. Superior Court, supra, 4 Cal.3d 800, 810.) Nevertheless, by greatly expediting the resolution of claims, class actions can create possibilities for injustice in an individual case. (City of San Jose v. Superior Court, supra, 12 Cal.3d 477, 458 .) A line of California cases follows the principle of mle 23(b)(3) of the Federal Rules of Civil Procedure (28 U.S.C.), which “provides that, for a class action to be maintained, it must be ‘superior to other available methods for the fair and efficient adjudication of the controversy.’ This ‘superiority’ criterion has been held to be ‘manifest’ in the . . . requirement that the class mechanism confer ‘substantial benefits.’ ” (Dean Witter Reynolds, Inc. v. Superior Court (1989) 211 Cal.App.3d 758, 773 [ 259 Cal.Rptr. 789 ]; Caro v. Proctor & Gamble Co., supra, 18 Cal.App.4th 644, 662 ; Schneider v. Vennard (1986) 183 Cal.App.3d 1340, 1346-1347 [ 228 Cal.Rptr. 800 ]; Lazar v. Hertz Corp. (1983) 143 Cal.App.3d 128, 143 [ 191 Cal.Rptr. 849 ].)

“Because trial courts are ideally situated to evaluate the efficiencies and practicalities of permitting group action, they are afforded great discretion in granting or denying certification” (Linder v. Thrifty Oil Co., supra, 23 Cal.4th 429, 435 ) or motions for decertification. (Occidental Land, Inc. v. Superior Court (1976) 18 Cal.3d 355, 360 [ 134 Cal.Rptr. 388 , 556 P.2d 750 ].) “[I]n the absence of other error, this court will not disturb a trial court ruling on class certification which is supported by substantial evidence unless (1) improper criteria were used [citation]; or (2) erroneous legal assumptions were made [citation].” (Richmond v. Dart Industries, Inc., supra, 29 Cal.3d 462, 470 .)

2. Legal Challenge

Claiming that the trial court abused its discretion in initially certifying the class of claims representatives on May 28, 1998, FIE and amici curiae advance an array of arguments with little actual citation to the record. This failure to address the record defeats their challenge to the trial court’s finding of a community of interest among class members. The trial court found that the class was “composed of hundreds of similarly situated employees” employed as claims representatives and that there were “common questions of law and fact common to the class . . . with respect to such class members’ claims for overtime pay.” The finding was based on plaintiffs’ extensive evidentiary showing in support of the motion for certification. In opposing the motion and moving twice for decertification, FIE never claimed that there were differences among the various duties performed by class members that would bar class certification; and in this appeal, it does not offer any reasoned argument, based on actual citations to the record, to this effect. The trial court’s findings of fact with respect to a community of interest remain unchallenged.

Arguing that “questions specific to each individual claims representative” predominated over common legal issues, FIE points only to matters, such as overtime hours worked, pertaining to the determination of individual damages. It is well established that the necessity for an individual determination of damages does not weigh against class certification. The community of interest requirement recognizes that “ultimately each class member will be required in some manner to establish his individual damages . . . .” (Vasquez v. Superior Court, supra, 4 Cal.3d 800, 815 ; Collins v. Rocha, supra, 7 Cal.3d 232, 238 ; Acree v. General Motors Acceptance Corp. (2001) 92 Cal.App.4th 385, 397 [ 112 Cal.Rptr.2d 99 ]; Hicks v. Kaufman & Broad Home Corp. (2001) 89 Cal.App.4th 908, 916 [ 107 Cal.Rptr.2d 761 ].) As explained in Employment Development Dept. v. Superior Court (1981) 30 Cal.3d 256, 266 [ 178 Cal.Rptr. 612 , 636 P.2d 575 ], “a class action is not inappropriate simply because each member of the class may at some point be required to make an individual showing as to his or her eligibility for recovery or as to the amount of his or her damages.” In B.W.I. Custom Kitchen v. Owens-Illinois, Inc. (1987) 191 Cal.App.3d 1341, 1354 [ 235 Cal.Rptr. 228 ], the court quoted a comment from a federal decision: “ ‘In almost every class action, factual determinations [of damages] ... to individual class members must be made. [Citations.] Still we know of no case where this has prevented a court from aiding the class to obtain its just restitution. Indeed, to decertify a class on the issue of damages or restitution may well be effectively to sound the death-knell of the class action device.’ [Citations.]”

Relying on language in Ramirez v. Yosemite Water Co. (1999) 20 Cal.4th 785 [ 85 Cal.Rptr.2d 844 , 978 P.2d 2 ], amici curiae call for “an individualized, employee-by-employee assessment of the exemption standards,” involving “a fact-intensive inquiry into the duties actually performed by employees.” But close attention to the facts may require an individual adjudication of exempt status in certain cases while permitting a group adjudication in others. We do not need to pursue amici curiae’s suggestion that other cases may call for individual adjudication of exempt status. Our decision here must be based on the record in this appeal, which amici curiae do not discuss. Amici curiae offer no reason to disturb the trial court’s finding that class members here were “similarly situated,” under circumstances presenting common issues of law and fact, which permitted an adjudication of whether they were employed in an “administrative capacity” within the meaning of wage order No. 4.

In its second motion for decertification, FIE opposed class certification on the ground that 9 percent of the 295 employees in the random sample did not claim compensation for unpaid overtime. In this appeal, it urges that the discovery of this fact constituted a “changed circumstance” that justified its second motion for decertification. The record discloses, however, that most of the employees not claiming overtime had worked as claims representatives for very short periods of time—often measured in a few weeks. The nonclaimants accounted for only 4 percent of total weeks worked in the class. Moreover, a segment of the nonclaimants were .currently employed as claims representatives and thus had an interest in the injunctive relief sought by the class representatives. If these employees are excluded from the total, only 5.7 percent of the employees in the random sample lacked an interest in the objectives of the litigation.

The presence of this marginal element of nonclaimants did not make the class less ascertainable or significantly reduce the required community of interest. Class members were identifiable by their employment in pertinent job categories. The small number of nonclaimants did not deprive plaintiffs of their representative status. The record contains nothing to suggest that nonclaimants constituted an employee segment with distinct interests conflicting with other class members. Finally, it did not affect the common issues of law and fact. The case called for individual adjudications only of the damages resulting from unpaid overtime. The question of an employee’s “eligibility for recovery” did not present a distinct issue of liability, as FIE argues, but rather depended on the same factual showing—unpaid overtime hours worked—as the question of the amount of the employee’s damages. (See Employment Development Dept. v. Superior Court, supra, 30 Cal.3d 256, 266 .)

FIE notes that proof of the fact of damages is required in individual actions for unpaid overtime (Hernandez v. Mendoza (1988) 199 Cal.App.3d 721, 726 [ 245 Cal.Rptr. 36 ]) and argues that the presence of nonclaimants demonstrated that plaintiffs could not establish this element of their cause of action for all class members. But if proof of individual damages were required by all potentially affected parties as a condition for class certification, it would go far toward barring all class actions. As noted in B.W.I. Custom Kitchen v. Owens-Illinois, Inc., supra, 191 Cal.App.3d 1341, 1354 , most class actions contemplate individual proof of damages, which necessarily entails the possibility that some class members will fail to prove damages. FIE cites no authority that class certification should be ordered only under circumstances promising universal recovery within the class.

Considerations of the efficient use of judicial resources supported the class action, despite the discovery of a marginal element of nonclaimants. On its facts, the present case is remarkably close to Richmond v. Dart Industries, Inc., supra, 29 Cal.3d 462 . The plaintiffs there filed a class action for fraud and violation of the Subdivided Lands Act against a developer of a recreational home subdivision. In the class certification hearing, the developer introduced the results of a survey in which it had asked homeowners to express their view of the subdivision by checking one of three boxes. Approximately 6 percent of the class returned the survey with a check in the box indicating satisfaction with the developer’s performance. The trial court relied on the survey to deny class certification. Reversing the order on appeal, the California Supreme Court observed, “Since the judicial system substantially benefits by the efficient use of its resources, class certifications should not be denied so long as the absent class members’ rights are adequately protected.” (Richmond v. Dart Industries, Inc., supra, at p. 474.) “The results of the flyer questionnaire show no more than approximately 6 percent of the class of some 4,000 persons antagonistic to the class action suit. This small number should not be sufficient to defeat the motion for certification.” (Id. at p. 475.)

Lastly, FIE argues that the large size of the employees’ claims—the average claim is $37,394—precludes class certification. But, addressing a similar argument in Collins v. Rocha, supra, 1 Cal.3d 232, 238, our high court observed “that while the impracticability of bringing an individual action for comparatively small potential recovery [is] a consideration in favor of permitting a class action, the converse [is] not necessarily controlling, i.e., the possibility of a potential recovery for each class member larger than a nominal sum does not militate against the maintenance of such an action.”

We note, first, that the size of individual claims does not necessarily have a bearing on the consideration of judicial efficiency favoring class actions. Whatever the size of the claims, a class action may be required “when the parties are numerous, and it is impractical to bring them all before the court,” as provided by Code of Civil Procedure section 382. FIE advocates the alternative of individual “mini-trials,” but this vaguely described concept would still involve “a multiplicity of legal actions dealing with identical basic issues” (Vasquez v. Superior Court, supra, 4 Cal.3d 800, 810), which would burden litigants and the courts with repetitious litigation. (Linder v. Thrifty Oil Co., supra, 23 Cal.4th 429, 435 .)

Moreover, class actions may be needed to assure the effective enforcement of statutory policies even though some claims are large enough to provide an incentive for individual action. While employees may succeed under favorable circumstances in recovering unpaid overtime through a lawsuit or a wage claim filed with the Labor Commissioner, a class action may still be justified if these alternatives offer no more than the prospect of “random and fragmentary enforcement” of the employer’s legal obligation to pay overtime. (Vasquez v. Superior Court, supra, 4 Cal.3d 800, 807.)

FIE maintains that where claims are as large as $37,000, the right to recover attorney fees, costs, and interest provides “ample incentive” for an individual lawsuit. But the size of the average claim in part reflects the accrual of unpaid overtime over the five-year duration of this lawsuit prior to trial. When the complaint was first filed in October 1996, the average claim would have been smaller and a large portion of the claims may not have been reasonably adequate to fund the expense of individual litigation. The length of this litigation in fact underscores the practical difficulties vindicating claims to unpaid overtime. Employees will seldom have detailed personal records of hours worked. Their case ordinarily rests on the credibility of vague recollections and requires them to litigate complex overtime formulas and exemption standards. For current employees, a lawsuit means challenging an employer in a context that may be perceived as jeopardizing job security and prospects for promotion. If the employee files after termination of employment, the costs of litigation may still involve travel expenses and time off from work to pursue the case, and the value of any ultimate recovery may be reduced by legal expenses.

In its second motion for decertification FIE argued that employees may seek administrative relief through what is commonly known as the “Berman” hearing procedure codified in Labor Code sections 98 to 98.8. (Stats. 1976, ch. 1190, §§ 4—11, pp. 5368-5371.) This procedure is “designed to provide a speedy, informal, and affordable method of resolving wage claims,” but it has several disadvantages for the employee. (Cuadra v. Millan (1998) 17 Cal.4th

T 855, 858 [ 72 Cal.Rptr.2d 687 , 952 P.2d 704 ]; Lolley v. Campbell (2002) 28 Cal.4th 367, 372 [ 121 Cal.Rptr.2d 571 , 48 P.3d 1128 ].) Injunctive relief is not available. The employer is typically represented by counsel; and if employees choose also to retain counsel, they cannot recover attorney fees. Moreover, a losing employer has a right to a trial de novo in superior court where the ruling of the hearing officer is entitled to no deference. (Lab. Code § 98.2, subds. (b) & (c); Post v. Palo/Haklar & Associates (2000) 23 Cal.4th 942, 948 [ 98 Cal.Rptr.2d 671 , 4 P.3d 928 ].) Opposing the first motion for decertification, plaintiffs submitted a declaration by a former chief counsel of the Department of Labor Standards Enforcement (DLSE), who cautioned, “Requiring two thousand or so class members to go through individual ‘Berman’ hearings would obviously be extremely inefficient as compared to a single class action. Also, a deluge of claims would simply outstrip the resources of the DLSE . . . impacting not only these claimants but others unrelated to this suit.”

If the costs to the courts of repetitious litigation and the public interest in effective enforcement of overtime laws are taken into account, the trial court clearly acted within its discretion in finding that a class action was a superior means of adjudicating the present controversy. Hence, on the present record, we conclude that FEE has not demonstrated that the trial court abused its discretion in granting the motion for class certification and later denying the two motions for decertification. Our decision is based solely on the issues raised by these motions.

D. Trial Management

1. Trial Court Discretion

FIE next challenges the use of the statistical methodology of random sampling and extrapolation for the determination of aggregate classwide damages. The issue is a novel one in California. It is true that California courts have relied on expert testimony on statistical methods to resolve certain issues in damage calculations, and Daar v. Yellow Cab Co., supra, 67 Cal.2d 695 , offers a precedent for determining aggregate classwide damages through a process of logical inference, but we must look to federal law for closer precedents. In discussing these federal precedents, we are mindful that class actions in the past have sometimes also called for judicial innovations guided by the experience of the federal courts. (City of San Jose v. Superior Court, supra, 12 Cal.3d 447, 453 .) In this area of litigation, the California Supreme Court has in fact “challenged the trial courts to develop ‘pragmatic procedural devices’ to ‘simplify the potentially complex litigation while at the same time protecting the rights of all the parties.’ [Citations.]” (State of California v. Levi Strauss & Co. (1986) 41 Cal.3d 460, 471 [ 224 Cal.Rptr. 605 , 715 P.2d 564 ].)

The record does not present the evidentiary issues that sometimes accompany the use of statistics in litigation. FIE did not challenge the scientific methodology employed by the plaintiffs’ expert witness or question his qualifications. Our analysis will be confined to two inquiries: (1) was adoption of the trial management plan within the scope of the trial court’s discretion? and (2) did the use of statistical inference violate FIE’s due process rights? While our inquiries may draw on the use of statistical methodology in disparate fields, we recognize claims for unpaid overtime compensation present certain unique and distinguishing characteristics. In the area of federal ELSA litigation, the touchstone is the seminal decision in Anderson v. Mt. Clemens Pottery Co. (1946) 328 U.S. 680 [ 90 L.Ed. 1515 , 66 S.Ct. 1187 ] (hereafter Aft Clemens).

In the Mt. Clemens case, the plaintiff employees complained that the employer failed to compensate them for time spent walking to their work stations and completing certain necessary activities before and after their scheduled shifts, but they could not accurately prove the amount of uncompensated work. The Supreme Court noted that an employee who brings suit under the FLSA “has the burden of proving that he performed work for which he was not properly compensated. The remedial nature of this statute and the

great public policy which it embodies, however, militate against making that burden an impossible hurdle for the employee.” (Mt. Clemens, supra, 328 U.S. 680, 686-687 .) Where the employer fails to keep proper records, the courts should not “penalize the employee by denying him any recovery on the ground that he is unable to prove the precise extent of uncompensated work. Such a result would place a premium on an employer’s failure to keep proper records in conformity with his statutory duty . . . .” (Id. at p. 687.) Instead, the court held: “an employee has carried out his burden if he proves that he has in fact performed work for which he was improperly compensated and if he produces sufficient evidence to show the amount and extent of that work as a matter of just and reasonable inference. The burden then shifts to the employer to come forward with evidence of the precise amount of work performed .... If the employer fails to produce such evidence, the court may then award damages to the employee, even though the result be only approximate.” (Id. at pp. 687-688, italics added.)

Following Mt. Clemens , the federal courts have consistently granted back wages to nontestifying employees on the basis of a pattern or practice adduced from the testimony of other employees within their job category. As stated in Donovan v. Bel-Loc Diner, Inc. (4th Cir. 1985) 780 F.2d 1113, 1116 , “[tjhere is no requirement that to establish a Mt. Clemens pattern or practice, testimony must refer to all non-testifying employees. ... [][].. . Courts have frequently granted back wages under the FLSA to non-testifying employees based upon the representative testimony of a small percentage of the employees. [Citation.] The requirement is only that the testimony be fairly representational.” (Fn. omitted.)

In some eases, the courts have relied on statistical analyses of evidence to infer wages owed to nontestifying employees within the same group as testifying claimants. In Donovan v. Hudson Stations, Inc. (D.C. Kan., Oct. 14, 1983, Nos. 77-2172 & 77-2173) 1983 U.S. Dist. LEXIS 12751 , 26 Wage & Hour Cas. (BNA) 795 [99 Lab.Cas.P. 34,463], the Secretary of Labor deposed 572 employees out of a group of 15,000 employees, but only 50 of this number were randomly selected. The deposition data was used to calculate the average amounts of “shortages paid” and “time worked off the clock” by employees. At trial, the Secretary asked that “these averages be applied to calculate the back pay award for all defendants’ employees.” (Id. at p. 801.) While recognizing that the survey was not conducted in “a scientific manner,” the court found that it was sufficient to establish uncompensated work “ ‘as a matter of just and reasonable inference.’ ” (Ibid.)

Again, in Reich v. Waldbaum, Inc. (S.D.N.Y. 1993) 833 F.Supp. 1037, 1049 , a “computer specialist” for the Labor Department calculated actual hours worked by nontestifying employees by a process of “within-store averaging” or “across-store averaging.” Noting that “[j]ust as some employees are overcompensated by the calculations, others are undercompensated,” the court concluded that the method was “reasonable,” though “not ideal.” (Id. at p. 1049.) Similarly, in McLaughlin v. Dialamerica Marketing, Inc. (D.N.J. 1989) 716 F.Supp. 812, 816-817 , a “compliance specialist” developed “a computerized summary of all of the available data” and devised formulas to reconstruct actual hours worked from the record of units of production. The court awarded back wages to both testifying and nontestifying employees based on these calculations of the average relation of units of production to hours worked. (Id. at p. 826.)

The present case differs from these precedents only in the size of the employee group and the use of a scientific methodology to infer aggregate classwide damages. If hours worked by nontestifying employees may be inferred from the testimony of other employees within a small employee group, we see no reason why it should not be allowed for a larger employee group. And if rough approximations and statistical estimates pass scrutiny for smaller groups, a scientific methodology based on a random sampling should also qualify as a “just and reasonable inference” of uncompensated hours worked in the case at bar.

FIE vigorously argues that the use of statistical inference improperly “relieved class members of their initial burden of showing they worked overtime,” a burden recognized both by the Mt. Clemens decision and by California law applying to the recovery of damages in individual wage claims. (Hernandez v. Mendoza, supra, 199 Cal.App.3d 721, 726 ; Mt. Clemens, supra, 328 U.S. 680, 687 ; cf. Collins v. Safeway Stores, Inc. (1986) 187 Cal.App.3d 62, 72 [ 231 Cal.Rptr. 638 ].) FIE notes, and we agree, that substantive rules of law may not be altered in the interests of efficient litigation. (Cimino v. Raymark Industries, Inc. (5th Cir. 1998) 151 F.3d 297, 312 ; Granberry v. Islay Investments (1995) 9 Cal.4th 738, 749 [ 38 Cal.Rptr.2d 650 , 889 R2d 970].) However, statistical sampling does not dispense with proof of damages but rather offers a different method of proof, substituting inference from membership in a class for an individual employee’s testimony of hours worked for inadequate compensation. It calls for a particular form of expert testimony to carry the initial burden of proof, not a change in substantive law. We note that the use of statistical sampling in the present case is analogous to FLSA precedents that allow back pay to nontestifying claimants. By basing relief on evidence of a pattern or practice, these decisions have also relieved some employees of the procedural necessity of making individual proof.

FIE appears to complain that the determination of aggregate damages on the basis of statistical inference entails the possibility of awarding back wages to particular employees who are not entitled to them. But this possibility also exists in FLSA cases (see, e.g., Reich v. Chez Robert, Inc., supra, 821 F.Supp. 967 , 974-975, 988-993) and is not a unique problem of statistical sampling; it is inherent in many class action decisions. FIE’s argument for individualized proof of damages, if accepted, would challenge all class action judgments adopting reasonably expeditious means of distributing the recovery among class members. (State of California v. Levi Strauss & Co., supra, 41 Cal.3d 460, 473-476 ; Bruno v. Superior Court, supra, 127 Cal.App.3d 120, 123-130 .) We decline to adopt this point of view, preferring the more pragmatic approach characterizing federal decisions. (See generally 7B Wright, Miller & Kane, Fed. Practice and Procedure, supra, § 1784, pp. 85-88; 3 Conte & Newberg, Newberg on Class Actions, supra, § 10:12, pp. 505-509.)

Nevertheless, from the perspective of the administration of justice,' we see an important negative consequence of the use of statistical sampling to calculate damages: it necessarily yields an average figure that will overestimate or underestimate the right to relief of individual employees. As Professor Robert Bone writes, “sampling can yield an extremely accurate average damage figure and thus an accurate total damage figure for the whole aggregation when the sample average is multiplied by the total number of plaintiffs. [] . . . [Hjowever, sampling imperfectly distributes this total relative to the expected outcome of an individual trial, giving some plaintiffs more and some less than their individual entitlements.” As we will see, the plan of distribution in this case is designed to ameliorate this problem, but it still offers an imperfect method of distributing damages in comparison to individual adjudication.

Weighing against this disadvantage is the consideration that statistical inference offers a means of vindicating the policy underlying the Industrial Welfare Commission’s wage orders without clogging the courts or deterring small claimants with the cost of litigation. In a particular case, the alternative to the award of classwide aggregate damages may be the sort of random and fragmentary enforcement of the overtime laws that will fail to effectively assure compliance on a classwide basis. In Mt. Clemens , the court held that “the remedial nature of this statute and the great public policy which it embodies” justified a reduced standard of proof of damages. (Mt. Clemens, supra, 328 U.S. 680, 687 .) The same consideration militates in favor of a reasonably expeditious means of calculating and distributing classwide aggregate damages if individual adjudication of the entitlements of all class members, or a substantial portion of the members, would impose impossible burdens on the courts and litigants.

In our view, it was within the discretion of the trial court to weigh the disadvantage of statistical inference—the calculation of average damages imperfectly tailored to the facts of particular employees—with the opportunity it afforded to vindicate an important statutory policy without unduly burdening the courts. As stated in In re Chevron U.S.A., Inc. (5th Cir. 1997) 109 F.3d 1016, 1018 , “[o]ur review of a trial court’s plan for proceeding in a complex case is a deferential one that recognizes the fact that the trial judge is in a much better position than an appellate court to formulate an appropriate methodology for a trial.”

2. Due Process

FIE maintains, however, that the use of statistical sampling and extrapolation violated its right to due process in the determination of damages. Under Connecticut v. Doehr (1991) 501 U.S. 1, 11 [ 115 L.Ed.2d 1 , 111 S.Ct. 2105](Doehr), the question whether a procedural device used in judicial proceedings to deprive a defendant of property comports with due process is determined by a balancing of interests: “[Fjirst, consideration of the private interest that will be affected by the [procedure]; second, an examination of the risk of erroneous deprivation through the procedures under attack and the probable value of additional or alternative safeguards; and third, . . . principal attention to the interest of the party seeking the [procedure], with, nonetheless, due regard for any ancillary interest the government may have in providing the procedure or forgoing the added burden of providing greater protections.” (See also Mathews v. Eldridge (1976) 424 U.S. 319, 335 [ 47 L.Ed.2d 18 , 96 S.Ct. 893 ]; Hilao v. Estate of Marcos (9th Cir. 1996) 103 F.3d 767, 786 ; In re Simon II Litigation, supra, 211 F.R.D. 86, 152-153 .)

In this case, the interest of FIE that was affected by the use of statistical sampling is its total or aggregate liability to the plaintiff class for unpaid overtime compensation. FIE may object to statistical sampling on due process grounds only to the extent that the procedure affected its overall liability for damages. (Sanders v. City of Los Angeles (1970) 3 Cal.3d 252, 263 [ 90 Cal.Rptr. 169 , 475 P.2d 201 ] [defendant lacked standing to object to a common fund attorney fee award]; Hilao v. Estate of Marcos, supra, 103 F.3d 767, 786 [the defendant’s “interest is only in the total amount of damages for which it will be liable”]; In re Agent Orange Product Liability Litigation (E.D.N.Y. 1984) 597 F.Supp. 740, 839 [defendant has no valid objection so long as it is “liable for no more than the aggregate loss fairly attributable to its tortious conduct”].)

FIE’s aggregate liability was not affected by the method of determining individual entitlements to members of the plaintiff class. This issue pertains only to the exercise of the trial court’s discretion in ordering a trial management plan. Similarly, the presence of nonclaimants in the class, which may be relevant to the trial court’s discretion in certifying the class, has no relevance to FEE’s due process claim so long as the inclusion of these class members did not increase the total amount of damages. The record shows that the damage award was in fact calculated on the basis of an average weekly overtime figure that factored in the presence of nonclaimants.

Turning to the third interest in the Doehr analysis, the interest of the plaintiffs in using statistical inference as a basis for an aggregate classwide recovery “is enormous, since adversarial resolution of each class member’s claim would pose insurmountable practical hurdles.” (Hilao v. Estate of Marcos, supra, 103 F.3d 767, 786 .) Relatively few class members have a realistic capability of assuming the costs and personal risks involved in judicial and administrative remedies. The ancillary interest of the government (i.e., the Industrial Welfare Commission and the courts) in using statistical inference for the award of damages lies in the means it affords to enforce the objectives of the Industrial Welfare Commission’s wage orders, advance the judicial policy of giving small claimants access to the courts, and avoid imposing unreasonable burdens on the judicial system in the enforcement of the law.

The second part of the Doehr test—the risk of erroneous deprivation— leads to the complex body of law and scholarship dealing with the use of statistical inference in adjudication, but in the end it reveals a serious constitutional issue only with respect to the plaintiffs’ claim for double-time compensation. We will discuss, first, the due process issue pertaining to the $88,798,871.12 award for time-and-a-half overtime compensation.

(a) Time-and-a-half Compensation

FIE argues that the statistical methods used to determine its aggregate liability for time-and-a-half compensation involved a risk of erroneous deprivation of property because “[t]he extrapolation process produced a range of overtime compensation figures between $80 million and $100 million—an arbitrary swing of $20 million.” More accurately, the record discloses that the statistical methodology employed by the plaintiffs’ expert, Richard Drogin, with the apparent concurrence of FIE’s expert, Roy Weinstein, not only estimated the average weekly unpaid overtime of the class on the basis of random sampling of 295 employees, but provided a measure of the possible inaccuracy of the estimate. This measure of possible inaccuracy consisted of a “confidence interval,” based on a 95 percent degree of confidence. The term “interval” can be expressed in more familiar language in terms of a margin of error, which is one-half of the interval. Drogin calculated that, with a 95 percent degree of confidence, the average weekly hours of unpaid overtime of the sample of 295 employees, i.e., 9.4 hours, reflected “the true average overtime hours” of the entire class of 2402 employees, subject to a margin of error of 0.9 hours.

We note, first, that, by a process of rounding upward to the nearest 10, FIE overstates the possibility of error. The margin of error of 0.9 weekly hours in fact yields a range of $17 million, or $8.5 million above or below the estimated aggregate classwide damages Secondly, FIE’s argument that repetition of the sampling would result in an “arbitrary swing” between the limits of the margin of error implicitly assumes that there would be an even distribution of outcomes within this range. But there is no basis for such an assumption, or, for that matter, for the contrary assumption that repetition would result in a clustering of outcomes close to the value of the initial sample. The confidence interval provides a quantitative measure only of the limits of the possible inaccuracy of the estimated average value. It does not convey any information regarding the probable distribution of outcomes within the margin of error, and the record contains no testimony on this point.

A due process critique of the accuracy of statistical inference implicitly assumes that it is a less reliable method of reaching a correct verdict than individual adjudication. But the proof of damages in an individual claimant’s trial may also involve estimates, inferences and other sources of error. The proof of damages by statistical inference chiefly differs in that it openly acknowledges the possibility of error and offers a quantitative measure of possible inaccuracy. In other words, it is overtly probabilistic.

By attacking the statistical proof of damages as “arbitrary” and “speculative,” FEE takes a position at odds with the growing acceptance of scientific statistical methodology in judicial decisions and scholarship. Forty years ago, the courts indeed displayed some reluctance to admit survey data, but today “[statistical assessments are prominent in many kinds of cases, ranging from antitrust to voting rights.” Citing varied uses of statistics, the court in In re Chevron U.S.A., Inc., supra, 109 F.3d 1016, 1020 , observes, “The applicability of inferential statistics have long been recognized by the courts.” Underlying the contemporary reliance on the methodology of inferential statistics is a recognition that “[ejxperts have developed appropriate modeling techniques for reaching statistically significant and reliable conclusions.” (In re Simon II Litigation, supra, 211 F.R.D. 86, 153 ; see also Ratanasen v. California Dept. of Health Services (9th Cir. 1993) 11 F.3d 1467, 1469-1472 ; Michigan Dept. of Educ. v. U.S. Dept. of Educ. (6th Cir. 1989) 875 F.2d 1196, 1205-1206 ; Pfizer, Inc. v. Lord (2d Cir. 1971) 449 F.2d 119 ; Blue Cross & Blue Shield of N.J. v. Philip Morris, supra, 178 F.Supp.2d 198, 247-248 ; Long v. Trans World Airlines, Inc. (N.D.Ill. 1991) 761 F.Supp. 1320, 1327 ; In re Antibiotic Antitrust Actions (S.D.N.Y. 1971) 333 F.Supp. 278, 289 , mandamus denied sub nom.; Avery v. State Farm Mut. Auto. Ins. Co. (2001) 321 Ill.App.3d 269 [ 746 N.E.2d 1242, 1260-1261 , 254 Ill.Dec. 194 ]; Tomlin v. Department of Social Services (1986) 154 Mich.App. 675 [ 398 N.W.2d 490, 497 ].) We find little basis in the decisional law for a skepticism regarding the appropriateness of the scientific methodology of inferential statistics as a technique for determining damages in an appropriate case.

In the hearings on the trial management plan on August 4, 2000, FIE made representations to the court that appeared to concede that a one-hour margin of error satisfied its due process concerns. In a memorandum proposing a 0.75-hours margin of error, FIE noted, “case law that supports the position that statistical sampling is appropriate and does not violate [FIE’s] due process rights only when the appropriate margin of error is used” and concluded that a margin of error of 0.75 hours “helps ensure that the aggregate damages [FIE] is required to pay do not exceed [FIE’s] actual liability in this action.” At the hearing on August 4, 2000, when the trial court suggested a one-hour margin of error as a compromise, FIE’s counsel responded, “We would be agreeable to a 1.0 number.” The sampling ultimately achieved a slightly better result—a 0.9-hours margin of error.

At trial, the plaintiffs’ statistical methodology withstood FIE’s challenges. The evidence revealed that over 90 percent of the employees randomly selected for depositions were actually deposed, a response rate regarded as producing a reliable and accurate sample. Drogin testified that the depositions of the second half of the sample did not vary much from the figures yielded by the first 150 depositions. Moreover, the elimination of the largest claimants, asserting claims for unpaid hours worked over 25 hours per week, would have a negligible impact on the average weekly figure.

We conclude that the proof of aggregate damages for time-and-a-half overtime by statistical inference reflected a level of accuracy consistent with due process under the Doehr balancing test. This conclusion is supported by persuasive authority. (In re Chevron U.S.A., Inc., supra, 109 F.3d 1016, 1020 ; Hilao v. Estate of Marcos, supra, 103 F.3d 767, 786-787 ; Yorktown Medical Laboratory, Inc. v. Perales (2d Cir. 1991) 948 F.2d 84, 89-90 ; Blue Cross & Blue Shield of N.J. v. Philip Morris, supra, 178 F.Supp.2d 198, 249 ; Long v. Trans World Airlines, Inc., supra, 761 F.Supp. 1320, 1327 ; In re Sugar Industry Antitrust Litigation (E.D.Pa. 1976) 73 F.R.D. 322, 351-355 ; In re Bailey (1989) 64 OhioApp.3d 291 [ 581 N.E.2d 577, 579-580 ]; see generally 3 Conte & Newberg, Newberg on Class Actions, supra, § 10:5, p. 483.)

(b) Double-time Compensation

The average weekly figure of 9.4 hours of unpaid overtime work included 0.37 hours of work qualifying for double-time compensation. The calculation of this figure was based on deposition testimony of the sample group regarding hours worked over 12 hours in a day or over 8 hours on the seventh day of a work week. The judgment awarded plaintiffs $1,210,337 as compensation for unpaid double-time hours.

FIE’s expert, Roy Weinstein, calculated that the use of the sample average of 0.37 hours of unpaid double-time compensation to determine unpaid double time for the entire class was subject to a margin of error of 0.12 hours or about 32 percent. Moreover, the distribution of employees claiming unpaid double time was highly skewed. Some 83 employees claimed some amount of unpaid double-time compensation, but only 54 employees claimed 0.37 hours or more, and a group of only 16 employees accounted for half of the double-time award.

In our view, the possible inaccuracy of the estimate of unpaid double-time compensation presents an issue of constitutional dimension. We do not mean to suggest that the margin of error alone may afford a bright-line constitutional distinction. The reliability of an estimate subject to a large margin of error might conceivably be bolstered by evidence of a high response rate, probable distribution within the margin of error, absence of measurement error, or other matters. We have been unable to find in the present record, however, anything that addresses the issues of reliability involved in the calculation of double-time damages. The estimate of unpaid double-time pay appears as a kind of afterthought in the trial management plan. The parties’ experts did not offer foundational calculations for the determination of double-time or propose an appropriate class size, margin of error, or sampling methodology.

We recognize that there are circumstances when a rough approximation of damages is preferable to a failure of justice resulting from no award at all. The record, however, reveals no consideration of possible alternatives or refinements to the statistical methodology and trial procedures used for calculation of the double-time element of damages. Under the Doehr analysis, the absence of evidence regarding possible alternatives is relevant not only to the risk of error but also to the ancillary government interest in the procedure. The government cannot have an interest in a procedure if superior alternative procedures are available.

Employing the balancing test mandated by Doehr , we conclude that the award of double-time compensation does not withstand due process challenge. We therefore reverse the portion of the judgment awarding unpaid double-time compensation of $1,210,337.

(c) Implementation of Trial Management Plan

Without any actual citation to the record, FIE makes a series of representations that it was barred from contesting the plaintiffs’ proof of damages. For example, it states: “Allowing proof of damages by a statistical showing . . . deprived FIE of its right to contest [plaintiffs’] evidence.” We agree that the trial management plan would raise due process issues if it served to restrict FIE’s right to present evidence against the claims (Mullane v. Central Hanover Tr. Co. (1950) 339 U.S. 306, 314 [ 94 L.Ed. 865 , 70 S.Ct. 652 ]; People v. Sandoval (1989) 206 Cal.App.3d 1544, 1550 [ 254 Cal.Rptr. 674 ] [restitution]), but we have searched the record without finding any indication that this occurred.

Although FIE originally proposed individual trials to establish damages, it appears that it acquiesced to statistical proof of damages at the hearing on April 21, 2000, which led to the initial order adopting a trial management plan. This interpretation tends to be confirmed by the parties’ collaboration in preparation of joint declarations of their experts and their agreement on a margin of error of one hour at the hearing on August 4, 2000. There was no mention of individual adjudication of damages at either hearing. It was not until shortly before trial that FIE raised the contention in a footnote to a status conference statement that it had not waived its demand for individual trials of damages.

Following the depositions of the 295 employees in the sample, FIE and plaintiffs agreed to work calendars, i.e., tabulations of data, that reflected the content of each employee’s testimony. FIE later waived the right to impeach the employees’ testimony at trial. Under the guidance of the special master, the parties agreed to a series of deadlines by which FIE was required to identify any deponent whom it wished to impeach. FIE initially challenged the testimony of two deponents and then resolved differences with plaintiffs with respect to these two witnesses. In a pretrial hearing before the special master, FIE reserved the right to introduce testimony of class members outside the sample, but we find no indication that it pursued this option. It never included individual employees in its witness list or sought to offer their testimony at trial.

We find only one pretrial ruling that significantly restricted FIE’s right to contest plaintiffs’ proof of damages: the order precluding its expert witness, Roy Weinstein, from testifying on probable measurement error in the deposition testimony of sample members. By measurement error, he meant a tendency for deponents to overstate average hours worked. After an extensive argument and offer of proof, the trial court excluded the testimony on the ground, among others, that it involved an asserted psychological phenomenon outside the scope of Weinstein’s expertise as an economist and statistician. FIE has not assigned error to this ruling on appeal.

In short, we find nothing in the record that substantiates FIE’s claim that the trial management plan restricted its opportunities to contest the evidence of damages or to present rebuttal evidence relating to hours worked by individual employees. Plaintiffs proved damages by a process of “just and reasonable inference” as in other wage claim actions, and FIE was afforded the opportunity to “come forward with evidence of the precise amount of work performed or with evidence to negative the reasonableness of the inference to be drawn from the employee’s evidence.” (Mt. Clemens, supra, 328 U.S. 680, 687-688 .) In doing so, FIE appears to have chosen to rely on its own statistical evidence as a matter of trial tactics. To the extent that the use of statistical sampling led to rough or expedient justice, it was in the adjudication of the relative entitlements of individual claimants to share in the aggregate award. FIE itself was not prejudiced by this process.

E. Postjudgment Plan of Distribution

In light of our discussion of FIE’s due process objections to the trial management plan, we see little need to comment on its contention that “the postjudgment claims procedure does not cure the due process violations.” Since we have found no due process violations in the award of aggregate classwide damages for time-and-a-half overtime compensation, the judgment does not present any due process infirmity to be cured. “A class action which affords due process of law to the defendant through the time when the amount of [its] liability is calculated cannot suddenly deprive [it] of [its] constitutional rights because of the way the damages are distributed.” (Bruno v. Superior Court, supra, 127 Cal.App.3d 120, 129 .)

It is well established that “the allocation of that aggregate sum [of the judgment] among class members is an internal class accounting question that does not directly concern the defendant. . . .” (2 Conte & Newberg, Newberg on Class Actions, supra, § 4:26, p. 233.) The trial court here followed the accepted practice of establishing a non-adversary proof-of-claim procedure. After “the defendant’s total damage liability is paid over to a class fund[,] . . . individual class members are afforded an opportunity to collect their individual shares by proving their particular damages, usually according to a lowered standard of proof.” (State of California v. Levi Strauss & Co., supra, 41 Cal.3d 460, 472 ; see also In re Sugar Industry Antitrust Litigation, supra, 73 F.R.D. 322, 353 ; In re Antibiotic Antitrust Actions, supra, 333 F.Supp. 278, 287 .) The record does not show that the class members, who are directly affected by this procedure, have raised any complaint. FIE may raise a due process objection only by showing that the procedure affected the total amount of its liability. (See p. 752, ante.)

We also note that FIE has raised only two narrow issues concerning the specific provisions of the postjudgment orders adopting a plan of distribution and awarding common fund attorney fees and costs: the application of Code of Civil Procedure section 384 to disposition of any residue in the damages fund and the need for a mechanism “to question an apparently excessive claim.” It has not assigned error to the award of common fund attorney fees. Moreover, it does not question the timing of the postjudgment notice requiring class members to submit proof-of-claim forms. Indeed, FIE did not request the mailing of such notices to class members prior to the damage phase of the trial but rather sought a second opt-out notice and pursued motions for class decertification. We thus have no occasion to consider the feasibility, or possible merit, of requiring submission of claim forms prior to the adjudication of aggregate classwide liability. (See 7B Wright, Miller & Kane, Fed. Practice and Procedure, supra, § 1784, p. 79; In re Antibiotic Antitrust Actions, supra, 333 F.Supp. 278, 288 .)

FIE raises a potentially difficult issue regarding the disposition of any residue in the damages fund, which turns on the interpretation and application of Code of Civil Procedure section 384. The plan of distribution logically entails the possibility that the total sum of damage claims submitted by class members, plus prejudgment interest, will be less than the judgment for aggregate classwide damages and interest. The judgment expressly provides for this possibility with respect to prejudgment interest; if the total amount of prejudgment interest paid to class members upon payment of their claims should be less than the actual award of prejudgment interest, FIE “may seek to amend the Judgment to allow it to recover the difference (if any) between the total amount awarded by the Court . . . and the total lesser amount payable to claiming class members . . . .” The plan of distribution, however, deferred a ruling on the disposition of any unclaimed damages to a later proceeding under Code of Civil Procedure section 384. It directs the plaintiffs or the claims administrator to submit “a postdistribution report stating the total amount that was actually paid to class members . . . .” After receiving class counsel’s recommendation on “how any such residue should be distributed,” the court is to determine how “to distribute the residue consistent with C.C.P. § 384(b).”

Subdivision (a) of Code of Civil Procedure section 384 states the legislative intent “to ensure that the unpaid residuals in class action litigation are distributed, to the extent possible, in a manner designed either to further the purposes of the underlying causes of action, or to promote justice for all Californians.” Subdivision (b) contemplates a report to the court followed by an order paying the residue to a charity: “The court shall... set a date when the parties shall report to the court the total amount that was actually paid to the class members. After the report is received, the court shall amend the judgment to direct the defendant to pay the sum of the unpaid residue ... to nonprofit organizations or foundations to support projects that will benefit the class ... or that promote the law consistent with the objectives and purposes of the underlying cause of action, to child advocacy programs, or to nonprofit organizations providing civil legal services to the indigent. . . .” (See Kraus v. Trinity Management Services, Inc. (2000) 23 Cal.4th 116, 128 [ 96 Cal.Rptr.2d 485 , 999 P.2d 718 ].)

FIE acknowledges that the distribution of the unpaid residue of a class action judgment to a nonprofit organization may be appropriate where the judgment requires a defendant to disgorge money that was fraudulently or unjustly acquired. (See Kraus v. Trinity Management Services, Inc., supra, 23 Cal.4th 116, 127 .) Such a distribution may be appropriate for tort damages that serve a broad remedial purpose and are incapable of precise measurement. But FIE reasonably argues that the sole purpose of the judgment here is to provide for payment of overtime compensation that is due and payable and the amount of this unpaid compensation is subject to precise calculation. Under these circumstances, FIE maintains, it would constitute a taking in violation of due process to distribute any excess in the damage fund to a third party as authorized by Code of Civil Procedure section 384.

FIE’s constitutional objections, however, are directed at a contingency that has not yet occurred and indeed may be unlikely. By deferring a decision on the disposition of unpaid residue in the damage fund, the plan of distribution not only followed the procedure mandated by Code of Civil Procedure section 384 but also adopted a course of action consistent with the principle that “we do not reach constitutional questions unless absolutely required to do so to dispose of the matter before us.” (People v. Williams (1976) 16 Cal.3d 663, 667 [ 128 Cal.Rptr. 888 , 547 P.2d 1000 ]; Santa Clara County Local Transportation Authority v. Guardino (1995) 11 Cal.4th 220, 230 [ 45 Cal.Rptr.2d 207 , 902 P.2d 225 ]; Walker v. Kiousis (2001) 93 Cal.App.4th 1432, 1441 [ 114 Cal.Rptr.2d 69 ].) The application of section 384 to the present case may require the trial court to consider constitutional issues bearing on the interpretation and enforceability of the statute. These issues can best be deferred to a later stage in the proceedings when it is known whether such a residue exists and the issues regarding its disposition can receive a full and adequate briefing. In so holding, we construe the plan of distribution as calling for a hearing before the trial court in which both parties will be given an opportunity to brief the full range of statutory and constitutional issues involved in the disposition of any residue in the damages fund.

FEE next complains that the plan of distribution “does not provide a mechanism for the claims administrator to question an apparently excessive claim.” The plaintiffs question whether FIE has standing to object to excessive claims, but we think it necessarily follows that, if FIE can assert a claim on due process grounds for an undistributed residue in a postjudgment hearing, it has standing to question the procedure that may affect the amount of this residue, even though it may be unlikely that the residue will actually materialize. We do not reach the question whether FEE’S interest in the morale of its employees is sufficient to give it standing to question the plan of distribution.

Plaintiffs properly object that FIE is attempting to inject at this late stage in the proceeding a costly and time-consuming procedure that will undermine the advantages of a class action for aggregate damages. In In re Agent Orange Product Liability Litigation, supra, 597 F.Supp. 740, 838 , the federal court observed, “If the judicial and monetary economies of the class action are not to be lost through lengthy and expensive individual trials on damages, some mechanism must be devised to decide damage claims without the need for a full-fledged trial for each plaintiff.” The plaintiffs here proposed the simple mechanism of distributing damages to all class members based on average overtime hours for the class. At FIE’s request, the trial court ordered a proof-of-claim procedure that relies on the accuracy and honesty of individual employees to achieve its purposes and involves considerably more administrative complexity. We reject FIE’s effort to transform this proof-of-claim procedure into a quasi-adversary proceeding that will cause the economies of the class action to be “lost through lengthy and expensive trials on damages.” (Ibid.)

Nevertheless, FIE’s assignment of error embraces a much narrower concern relating to precautions against fraud. We may appropriately consider whether this narrower issue has been addressed, or is capable of being addressed, in ways that will not unduly burden the claims procedure. Our high court’s discussion of the issue in State of California v. Levi Strauss & Co., supra, 41 Cal.3d 460 , is the only authority on point that we have found.

In Levi Strauss the court considered a plan to distribute $9.3 million in damages to consumers who were entitled to an average individual recovery of $2.60 to $3.00 for each purchase of the defendant’s jeans. Notices of the settlement were mailed to 8.6 million households and other forms were made available at post offices. The trial court ordered that the top 5 percent of claims would be treated as suspect and the claim forms would be “returned with instructions to resubmit with a notarized statement confirming their veracity.” (State of California v. Levi Strauss & Co., supra, 41 Cal.3d 460, 470 .) Over 80 percent of these claims were not resubmitted. (Ibid.) The intervener argued that the plan rewarded “inflated or baseless claims” and was afflicted with “widespread fraud.” (Id. at p. 476.) The Supreme Court noted that “there appears to be no reported decision in any jurisdiction approving a plan so lacking in safeguards against fraud” (id. at p. 477), but still affirmed the judgment “[d]ue to . . . practical considerations . . . .” (Id. at p. 479.)

Though the court affirmed the plan of distribution, we read the Levi Strauss decision as saying that precautions against fraud are a legitimate concern in crafting a plan of distribution, and a failure to provide adequate precautions against fraud may constitute an abuse of discretion. The plan of distribution here includes four features that may be regarded as precautions against fraud: (1) the claims of the 343 employees who were deposed (14 percent of the total) will be based on their deposition testimony; (2) other claims must be signed under penalty of perjury; (3) employees who claim to have worked an average of more than 40 overtime hours per week will have their payments based on 40 overtime hours per week; (4) if claim forms are not properly completed, the claims administrator or class counsel may contact class members and provide them with an opportunity to file amended claims, and, if necessary information is still lacking, submit a recommendation to the court concerning the appropriate distribution.

Under this procedure, if an undeposed employee properly completes a proof-of-claim form and reports less than 40 hours per week of average overtime, the only procedural safeguard against fraud is the requirement of signing the form under penalty of perjury. We have no way of knowing whether the fear of peijury, as well as personal honesty and peer pressure, will prove sufficient to guarantee the integrity of the claims process. We are troubled, however, that the plan does not empower the claims administrator to take action if it receives a quantity of large claims that markedly exceeds the frequency of such claims reported in depositions of the random sample. The most obvious response would be to give the claims administrator or his designee discretionary authority to challenge a small and strictly limited number of claims presenting indicia of possible fraud. The challenge could be governed by the existing procedures applying to employee challenges to preprinted information on the claim form, which may lead ultimately to a decision by the special master. If combined with a conspicuous notice informing employees that the claims administrator has authority to investigate claims for fraud and with a cap on allowable average overtime hours well below the unrealistic limit of 40 hours per week, a strictly limited power of challenge might significantly reduce the possibility of fraud without unduly burdening the claims administration process. We remand the case to the trial court for a limited amendment of the plan of distribution and the order approving claim forms and instructions, which will authorize the claims administer to challenge suspect claims.

F. Improper Argument

Lastly, FIE maintains that plaintiffs’ counsel improperly argued to the jury in his opening argument that their verdict would be “an important message,” which would be heard by FIE’s corporate headquarters and by other employers. FIE waited until the end of counsel’s argument to object that the remark was an appeal to “the jury’s passion” and represented an attempt to impose the equivalent of punitive damages. The trial court asked plaintiffs’ counsel if he was willing to state to the jury that he did not request punitive damages. Counsel agreed to do so, and the court concluded, “I think that will clear it.” FIE again raised its objection in its motion for new trial. Rejecting the claim of error, the trial court found that counsel’s remarks “were not misconduct and in any event were cured by counsel’s other statements to the jury, and there is no indication that the jury was influenced by passion or prejudice . . . .”

We consider that any arguable claim of impropriety disappears if the argument is considered as a whole. (Nishihama v. City and County of San Francisco (2001) 93 Cal.App.4th 298 , 304—306 [ 112 Cal.Rptr.2d 861 ].) Plaintiffs’ counsel repeatedly stated that the jury should award plaintiffs what they were entitled to. He explained, “[T]he reason it’s so important is because the overtime laws embody a fundamental public policy of the State of California. And people are entitled to be paid overtime at premium rates.” In this context, counsel simply appealed to the jury to vindicate the public policy underlying the overtime laws by holding FIE accountable for the full amount of overtime compensation owing to plaintiffs. We do not view this argument as suggesting that the jury should inflate the damage award or award the equivalent of punitive damages.

On the present record, we do not need to invoke the rule giving counsel wide latitude in arguing to the jury (Tingley v. Times Mirror (1907) 151 Cal. 1, 23 [ 89 P. 1097 ]; Menasco v. Snyder (1984) 157 Cal.App.3d 729, 732 [ 203 Cal.Rptr. 748 ]) or to rely on the deference to be given the trial court’s ruling on alleged misconduct. (Sabella v. Southern Pac. Co. (1969) 70 Cal.2d 311, 319 [ 74 Cal.Rptr. 534 , 449 P.2d 750 ]; Cope v. Davison (1947) 30 Cal.2d.193, 203 [ 180 P.2d 873 , 171 A.L.R. 667 ].) We do not consider that counsel exceeded the bounds of proper argument. Moreover, we see no possibility of prejudice to FIE. Plaintiffs’ counsel dispelled any possible suggestion of impropriety by disclaiming a request for punitive damages in his later statement to the jury. The jury verdict was properly based on the quantitative calculations of plaintiffs’ expert witness, rather than any subjective measure of damages.

DISPOSITION

The portion of the judgment filed September 24, 2001, based on the special verdict of $1,210,337 for unpaid double-time hours worked by the class is reversed and the judgment is otherwise affirmed. The order re plan of distribution filed September 20, 2001, and the order approving claim forms and instructions filed April 23, 2002, are reversed and the case remanded for further proceedings consistent with this opinion. In all other respects, the postjudgment orders subject to appeal are affirmed. The parties are to bear their own costs on appeal.

Marchiano, P. J., and Stein, J., concurred.

A petition for a rehearing was denied February 9, 2004, and the opinion was modified to read as printed above. Appellant’s petition for review by the Supreme Court was denied May 12, 2004.

For explanation of the concept “95 percent level of confidence,” see page 753, post. The mean is the average value of a set of numbers, and the median is the middle value of a set of numbers arranged in order of size. (American Heritage Diet. (3d college ed. 1993) pp. 841, 843.) The prejudgment interest was to be calculated on individual distributions. It was set at a maximum of $32,303,048 plus $24,584 per day for each day after the verdict and until filing of the judgment. A later award of statutory attorney fees is the subject of a separate appeal and is not reviewed herein. “Under the law of the Ninth Circuit, we must follow Bell absent ‘convincing evidence’ that California’s Supreme Court would decide the issue differently. [Citation.] Based on the record before us, we do not find compelling reasons that would lead us to believe that the California Supreme Court would reach a different conclusion.” (Palacio v. Progressive Ins. Co., supra, 244 F.Supp.2d 1040, 1050 .) We regard Bell II as being consistent with an earlier opinion letter of the Wage and Hour Division issued March 14, 1967. It observes that “[i]nsurance adjusters have generally been found to be within the general coverage of the act,” but recognizes that adjusters must “meet all the pertinent tests for exemption relating to duties, responsibilities and salary .... If an employee fails to meet any of the applicable tests contained in the regulations, he will not qualify for this exemption.” (See also DOL opn. letter issued Oct. 29, 1985.) (We take judicial notice of DOL opinion letter (May 17, 1999) [“loan officers are engaged in carrying out the employer’s day-to-day activities rather than in determining the overall course and policies of the business”]; DOL opinion letter (Feb. 18, 1999) [same language with respect to store managers]; DOL opinion letter (May 28, 1998) [boarding agents’ duties “related more to the ongoing day-to-day production operation[] of your client’s shipping business rather than to its management policies or general business operation[]’’]; DOL opinion letter (Jan. 23, 1998) [medical investigator is “carrying out the employer’s day-to-day affairs rather than running the business itself or determining its overall course and policies”]; DOL opinion letter (Jan. 8, 1998) [“sales duties fall into the category of ‘production’ rather than ‘management,’ ” relying on Martin v. Cooper Elec. Supply Co. (3d Cir. 1991) 940 F.2d 896 ]; DOL opinion letter (Sept. 12, 1997) and DOL opinion letter (July 23, 1997) [discussing and applying leading federal decisions cited in Bell II that employ the dichotomy analysis]; DOL opinion letter (Oct. 13, 1994) [“duties of the transportation brokers involve basic tasks of the ‘production’ work of the company”]; DOL opinion letter (March 16, 1992) [“duties of the safety director involve basic tasks of the electricity department, i.e., the ‘production’ work of the department”]; DOL opinion letter (May 9, 1988) [“activities performed by a Child Protective Investigator II appear to be related more to the ongoing day-to-day ‘production’ operations of the department than to management policies or ‘general business operations’ ”].) See Bell II, supra, 87 Cal.App.4th 805, 821-823, 826 , and Bothell v. Phase Metrics, Inc., supra, 299 F.3d 1120, 1127 , footnote 6. The amici curiae briefs add only one decision subsequent to Bell II , and it is entirely distinguishable on its facts. The plaintiff in McAllister v. Transamerica Occidental Life Ins. (8th Cir. 2003) 325 F.3d 997, 998 , handled “ ‘the most complex life claims’ ’’ of up to $250,000 and her duties involved training other claims examiners, interpretation of contracts and insurance statutes, and supervision of investigative work. Amici curiae, moreover, rely on selective characterizations of the insurance business. It is not difficult to find judicial recognition of claims adjusting as an important component of the insurance business. See, e.g., Illinois Commercial Men’s Assn. v. State Bd. of Equalization (1983) 34 Cal.3d 839, 849 [ 196 Cal.Rptr. 198 , 671 P.2d 349 ] (“the investigation and settlement of claims is an integral and crucial aspect of the business of insurance”); Davis v. Continental Ins. Co. (1986) 178 Cal.App.3d 836, 841 [ 224 Cal.Rptr. 66 ] (from the perspective of the insured, “the most essential part of the business of insurance is the . . . investigation and payment of their claims”); Bodenhamer v. Superior Court (1986) 178 Cal.App.3d 180, 184 [ 223 Cal.Rptr. 486 ] (“insurance adjusters . . . conduct an important aspect of the business regulated”); Proctor v. State Farm Mut. Auto. Ins. Co. (D.C. Cir. 1982) 218 U.S. App. D.C. 289 [ 675 F.2d 308, 313 ] (settlement of claims lies “within the ‘core’ of the ‘business of insurance’ ” for purposes of the McCarran-Ferguson Act ( 15 U.S.C. § 1011 et seq.)). The pertinent passage reads: “As stated in the stipulation of facts, John Alden is in the business of designing, creating, and selling insurance policies to the public. It follows, as the district court properly recognized, that the ‘products’ generated by John Alden are these insurance policies themselves. As the marketing representatives are in no way involved in the design or generation of insurance policies, the very product ‘that the enterprise exists to produce and market,’ Dalheim [v. KDFW-TV, supra,] 918 F.2d [1220,] 1230, they cannot be considered production employees.” (Reich v. John Alden Life Ins. Co., supra, 126 F.3d 1, 9 .) See, for example, Reich v. State of New York (2d Cir. 1993) 3 F.3d 581 [investigators in the New York State Police Bureau of Criminal Investigation]; Martin v. Cooper Elec. Supply Co., supra, 940 F.2d 896 [inside sales persons in a wholesale sales business]; Shockley v. City of Newport News (4th Cir. 1993) 997 F.2d 18 [police sergeants staffing telephone lines]; Dalheim v. KDFW-TV, supra, 918 F.2d 1220 [producers for the news portions of television newscasts]; Bratt v. County of Los Angeles, supra, 912 F.2d 1066 [deputy probation officers]; Reich v. American Intern. Adjustment Co., Inc. (D.Conn. 1994) 902 F.Supp. 321 [automobile damage appraisers for business providing service of resolving damage claims]; Fleming v. Carpenters/Contractors Cooperation Com. (S.D.Cal. 1993) 834 F.Supp. 323 [field investigators for labor-management cooperation committee]; Reich v. Chicago Title Ins. Co., supra, 853 F.Supp. 1325 [escrow closers for title insurance company]. The interpretative regulation that enunciates the administrative/production worker dichotomy, 29 C.F.R. part 541.205(c), states that administrative work may relate “to the operation of a particular segment of the business.” It may reasonably be read to suggest by implication that the converse is true; production work may also relate to a segment of the business. The California Supreme Court has granted review in a case raising the issue of class certification of overtime claims by salaried managers. (Sav-on Drug Stores, Inc. v. Superior Court, review granted July 17, 2002, S106718.) For a discussion of the policy generally favoring class actions, see Howard Gunty Profit Sharing Plan v. Superior Court (2001) 88 Cal.App.4th 572, 578 [ 105 Cal.Rptr.2d 896 ], Bartold v. Glendale Federal Bank (2000) 81 Cal.App.4th 816, 828 [ 97 Cal.Rptr.2d 226 ], offers a brief explication of the requirement. We note that FIE did not raise the issues of a limited certification or bifurcated trials in opposing the initial motion for certification or in its first motion for decertification. The second motion for decertification was filed a month before trial, when the trial court had developed, with FEE’S cooperation, a trial management plan for proof of aggregate classwide damages, which obviated the need for separate trials of individual damages. Accordingly, we do not reach the issues relating to partial class certifications or split trials raised by the Federal Rules of Civil Procedure and a body of federal decisions (see Fed. Rules Civ.Proc., rule 23(c)(4), 28 U.S.C. [“an action may be brought or maintained as a class action with respect to particular issues”]; 3 Conte & Newberg, Newberg on Class Actions (4th ed. 2002) § 9:53, pp. 429-436; and 7B Wright, Miller & Kane, Fed. Practice and Procedure (2d ed. 1986) § 1784, pp. 78-79) as well as by dicta in B.W.I. Custom Kitchen v. Owens-Illinois, Inc., supra, 191 Cal.App.3d 1341, 1354 (“A bifurcated trial . . . may be employed to simplify the proceedings”). Johns, California Damages: Law and Proof (5th ed. 2003) section 1.16, page 1-40. See Bruno v. Superior Court (1981) 127 Cal.App.3d 120, 129 , footnote 4 [ 179 Cal.Rptr. 342 ] (“Due process does not prevent calculation of damages on a classwide basis. The Supreme Court has assumed the use of such a method (Daar v. Yellow Cab Co., supra, 67 Cal.2d at pp. 706, 714, 716). In many cases such an aggregate calculation will be far more accurate than summing all individual claims.”). We cite the phrase “potentially complex litigation” for its descriptive meaning, without intending a reference to a procedural classification. The record does not disclose that the case was designated complex litigation under local rules. See In re Simon II Litigation (E.D.N.Y. 2002) 211 F.R.D. 86, 149-152 ; Diamond, Reference Guide on Survey Research in Federal Judicial Center, Reference Manual on Scientific Evidence (2d ed. 2000) page 229. For a discussion of leading recent cases in the field of mass torts, see Walker and Monahan, Sampling Damages (1998) 83 Iowa L.Rev. 545. See, e.g., Grochowski v. Phoenix Const. (2d Cir. 2003) 318 F.3d 80 , 88 [“the plaintiffs correctly point out that not all employees need testify in order to prove FLSA violations or recoup back-wages”]; Reich v. Gateway Press, Inc. (3d Cir. 1994) 13 F.3d 685 , 701 [22 out of 70 employees testified]; Martin v. Tony and Susan Alamo Foundation (8th Cir. 1992) 952 F.2d 1050, 1052 [back pay to be awarded “to the nontestifying employees based on the fairly representative testimony of the testifying employees”]; Martin v. Selker Bros., Inc. (3d Cir. 1991) 949 F.2d 1286 , 1298 [4 out of 6 employees testified]; McLaughlin v. Ho Fat Seto (9th Cir. 1988) 850 F.2d 586, 589 [5 out of 28 employees testified]; Donovan v. Simmons Petroleum Corp. (10th Cir. 1983) 725 F.2d 83, 86 [testimony of 12 former employees supported award to all former employees]; Donovan v. Williams Oil Co. (10th Cir. 1983) 717 F.2d 503, 504-505 [testimony of 19 attendants established pattern for 34 employees at nine stations]; Donovan v. New Floridian Hotel, Inc. (11th Cir. 1982) 676 F.2d 468, 472 [23 employees testified out of 207 employees receiving award]; Donovan v. Burger King Corporation (1st Cir. 1982) 672 F.2d 221 , 224—225 [26 of 246 employees]; Brennan v. General Motors Acceptance Corporation (5th Cir. 1973) 482 F.2d 825, 829 [16 out of 26 employees testified]; Reich v. Chez Robert, Inc. (D.NJ. 1993) 821 F.Supp. 967 , 984, 988-993 [28 out 246 testified]; Reich v. New Mt. Pleasant Bakery (N.D.N.Y., Sept. 13, 1993, No. 89-CV-581) 1993 U.S. Dist. LEXIS 12937 [15 of 46 drivers testified]; Martin v. Petroleum Sales, Inc. (W.D. Tenn., July 9, 1992, No. 90-2453-4A) 1992 U.S. Dist. LEXIS 21339 [10 out of 34 store managers testified]; Marshall v. Brunner (W.D.Pa. 1980) 500 F.Supp. 116, 122 [48 out of 93 employees testified]. Bone, Statistical Adjudication: Rights, Justice, and Utility in a World of Process Scarcity (1993) 46 Vand. L.Rev. 561, 572-573. See 3 Conte & Newberg, Newberg on Class Actions, supra, section 10:5, page 487 (“aggregate proof of the defendant’s monetary liability promotes the deterrence objectives of the substantive laws underlying the class actions and promotes the economic and judicial access for small claims objectives of Rule 23”). See the excellent survey of the decisional law and scholarly literature in In re Simon II Litigation, supra, 211 F.R.D. 86, 148-154 . Kaye and Freedman, Reference Guide on Statistics in Federal Judicial Center, Reference Manual on Scientific Evidence (2d ed. 2000) pages 83, 119, footnote 118. FIE’s use of the figure $20 million appears to be based on a leading question addressed to Drogin, which received a response that it was “approximately correct.” Finkelstein and Levin, Statistics for Lawyers (2d ed. 2001) pages 169-171. See U.S. v. Fior D’ltalia, Inc. (2002) 536 U.S. 238 [ 153 L.Ed.2d 280 122 S.Ct. 2117, 2125 ] [“After all, individual audits will be plagued by some of the same inaccuracies Fior D’ltalia attributes to the aggregate estimation method, because they are, of course, based on estimates themselves”]; Blue Cross & Blue Shield of N.J. v. Philip Morris (E.D.N.Y. 2001) 178 F.Supp.2d 198, 247 ; see also Saks & Blanck, Justice Improved: The Unrecognized Benefits of Aggregation and Sampling in the Trial of Mass Torts (1992) 44 Stan. L.Rev. 815, 833-837. In re Simon II Litigation, supra, 211 F.R.D. 86, 147 (quoting Rosenberg, The Causal Connection in Mass Exposure Cases: A "Public Law” Vision of the Tort System (1984) 97 Harv. L.Rev. 851, 870); Koehler and Shaviro, Veridical Verdicts: Increasing Verdict Accuracy Through the Use of Overtly Probabilistic Evidence and Methods (1990) 75 Cornell L.Rev. 247, 248; Saks & Kidd, Human Information Processing and Adjudication: Trial by Heuristics (1980) 15 Law & Society Rev. 123, 151. Diamond, Reference Guide on Suivey Research in Federal Judicial Center, Reference Manual on Scientific Evidence, supra, page 233. Kaye and Freedman, Reference Guide on Statistics in Federal Judicial Center, Reference Manual on Scientific Evidence, supra, page 85. Similarly, the memorandum states that the depositions required to reach a 0.75-hours margin of error “will ensure that neither [FEE] nor plaintiffs are denied due process in a calculation of aggregate damages.” See Diamond, Reference Guide on Survey Research in Federal Judicial Center, Reference Manual on Scientific Evidence, supra, page 245 [notes that according to the guidelines for statistical surveys issued by the former U.S. Office of Statistical Standards, “response rates of 90% or more are reliable and generally can be treated as random samples of the overall population”]. “MR. ZIEFF [plaintiffs’ counsel]: ‘I’m here to report that there’s been progress. Kind of a meeting not on all of the details but on the general game plan on the initial approach. Both sides agree the statistical [garbled] is a reasonable, practical approach.’ ” He then proceeded to describe plans for the initial pilot study. Turning to FEE’S counsel, Lee Paterson, the court asked, “Is this going to be in writing so we have some idea of exactly what’s going to happen?” “MR. PATERSON: T think we can. We’ll do it, if you want. What we’ll say: there will be an initial pilot of 50 people.’ ” It was instead the plaintiffs who proposed to present trial testimony of employee witnesses outside the sample group. Their witness list included 15 class members, who were identified as being able to testify regarding FIE’s “willfulness” within the meaning of Labor Code section 203, an issue that was excluded from the trial and is not involved in this appeal. In motions in limine, FIE sought to exclude these witnesses and insisted only on the right to call rebuttal witnesses if the plaintiffs’ witnesses should be allowed to testify. The judgment provides: “the judgment shall include pre-judgment interest in the maximum amount of $32,303,048, plus an additional $24,584 per day for each day after June 26, 2001 until and including the date on which judgment is entered." The order re plan of distribution provides: “A class member may challenge the pre-printed information on the Claim Form as to dates of employment as a personal lines claims representative in California as inaccurate, if and only if he or she presents written information or documents verifying, or if Farmers’ personnel records explicitly verify, that the pre-printed information as to dates of employment is inaccurate. Such challenge shall be reviewed and determined by Class Counsel. If Class Counsel and the class member cannot reach agreement as to dates worked, any such dispute shall be resolved by informal, binding hearing before a Special Master appointed by the Court. The Special Master will resolve such disputes either by expedited, in-person hearing, telephonic conference, or upon consideration of concise written submissions, whichever the Special Master deems to be most appropriate at his or her sole discretion, without review by the Court.” The disputed portion of the argument is the following: “And the verdict that you render in this case is an important message. The message will be heard by Farmers at [its] corporate headquarters in Los Angeles. The message will be heard by other employers who do not pay people entitled to overtime all of the overtime they’re entitled to. These people haven’t been paid overtime. They’re entitled to be paid for all of the overtime hours. Your message will be heard.” Counsel opened his closing argument by saying: “What the plaintiffs have told you in this case ... is what they’re asking for is to be paid for all their overtime hours. Nothing more. Nothing less. . . . [W]e’re not asking for exemplary damages. . . . We’re not asking for you to award punitive damages .... [A]ll we want is the full back pay fund that will pay all the class members for all of the total hours worked.” See, for example, the remark preceding the disputed argument: “What we’re talking about here is making Farmers pay for the overtime hours that they benefited from. What we’re talking about is making Farmers pay for all of the hours that the class members worked, the undisputed hours that they worked.” After the disputed argument, counsel again urged, “the key issues to focus on are that class members are entitled to be paid for all of their overtime hours.” Respondents’ alternative request for judicial notice filed April 17, 2003, and FTE’s request for judicial notice filed January 3, 2003, are granted. FIE’s request for judicial notice filed December 8, 2003, is denied.