Kilgo v. Bowman Transportation, Inc. , 789 F.2d 859 ( 1986 )

ANDERSON, Circuit Judge:

Bowman Transportation, Inc. appeals from the decision of the district court holding that it had violated Title VII of the Civil Rights Act of 1964, 42 U.S.C.A. §§ 2000e to 2000e-16, by discriminating against women in its hiring for over-the-road (“OTR”) tractor-trailer driving positions. In Case Nos. 84-8105 and 84-8899, Bowman, a trucking company, raises numerous challenges to the district court’s findings with respect to the disparate impact of its one-year prior experience rule, the pattern and practice of discrimination against women in hiring for OTR truck drivers, and the certification of the plaintiff class (“plaintiffs”). We reject each of Bowman’s challenges. On the cross-appeal, we also reject plaintiffs’ challenge to the district court’s refusal to include hiring goals in its remedial order. In Case No. 85-8409,¹ Bowman argues that the district court’s civil contempt order dealing with back pay is punitive, and since civil contempt may not be imposed for the purpose of punishment, the contempt order is invalid. We find no merit in this contention. Thus, we affirm.

I. BACKGROUND

In late 1974, Bowman imposed a requirement that applicants for OTR driver positions have one-year prior experience as OTR drivers. This requirement was imposed on a company-wide basis, with the exception of the Birmingham, Alabama region, which only required six-months prior experience until 1976. Under this rule, an applicant's prior experience would be considered OTR if it involved driving a tractor-trailer for a distance greater than a 75-mile radius from the dispatch terminal.

In 1976, Edna Kilgo went to the Bowman terminal in Atlanta, Georgia to apply for an OTR driver position, and when Bowman refused to allow her to file an application, she filed a complaint with the Equal Employment Opportunity Commission (“EEOC”). After receiving her right-to-sue letter, Kilgo filed this class action lawsuit under Title VII of the Civil Rights Act of 1964, 42 U.S.C.A. §§ 2000e to 2000e-16, alleging sex discrimination by Bowman in its hiring of OTR drivers.

In August 1979, Virginia Wentz was added as a party plaintiff. After the death of Edna Kilgo in November 1979, the district court held that Kilgo’s Title VII claim survived her death, and thus permitted the substitution of her husband, Oscar Kilgo, as a named plaintiff in his capacity as the representative of her estate. Kilgo v. Bowman Transportation, Inc., 87 F.R.D. 26, 27-28 (N.D.Ga.1980). The court, however, concluded that Oscar Kilgo would not be an adequate representative of the proposed class “because, although he has an interest in claims for back pay, he does not have the same interest in declaratory and injunctive relief as other members of the proposed class.” Id. at 28-29. In addition, the court found that the date of Kilgo’s filing of her EEOC charge would govern the scope of the class both because her filing had given Bowman “adequate notice of a challenge to its employment practices” and because a contrary holding might prejudice members of the proposed class who “could have been aware of Edna Kilgo’s EEOC charge and the possibility that she would represent them in a class action.” Id. at 29. Finally, the court also conditionally certified a class consisting of “[a]ll females who, since April 18, 1976, have, would have, but for the futility of doing so, or will in the future seek permanent employment as an over-the-road truck driver with [Bowman] by applying through its Atlanta, Georgia terminal and who have been, would have been, or will be refused such permanent employment due [to Bowman’s discriminatory practices].” Id. at 28-30.

In March 1982, the district court redefined the class as follows:

[T]he class includes any applicant who, meeting other class standards, (1) was hired by [Bowman] as an over-the-road driver but was terminated within the forty-five day probationary period, (2) was not hired as an over-the-road driver and her application was, for one reason or another, forwarded to the Atlanta terminal for its review, and (3) any prospective applicant, meeting other class standards, whose application, had she filed one, would have been forwarded to the Atlanta terminal.

Record on Appeal, vol. 3 at 634-35.²

After a nonjury trial, the district court found that Bowman had committed sex discrimination in its hiring of OTR drivers. Kilgo v. Bowman Transportation, Inc., 570 F.Supp. 1509 (N.D.Ga.1983). First, the court redefined the class with respect to the disparate impact claim:³

All females who since April 18, 1976, have, would have but for the futility of doing so, or will in the future seek permanent employment as an over-the-road truck driver with defendant by applying at any of the terminals, and who have been, would have been, or will be refused such employment due to the operation of [Bowman’s] commercial, over-the-road tractor-trailer driving experience requirement.

Id. at 1512-13 (emphasis in original). The court rejected Bowman’s contention that this redefinition of the class would violate the due process clause, Rules 16 and 23 of the Federal Rules of Civil Procedure, and N.D.Ga.R. 221.13. Id. The court first noted that since Bowman’s experience requirement is a company-wide policy, the asserted business justification for the requirement would apply whether this requirement is examined in terms of applicants seeking employment through the Atlanta terminal or through any of Bowman’s terminals. Id. at 1513. In addition, the court rejected Bowman’s argument that it would be prejudiced by the lack of an opportunity to introduce system-wide applicant flow data; the court found that argument to be without merit because such applicant flow data did not reflect an appropriate labor pool in this case since women were deterred from applying. Id. Finally, the court found that the introduction of evidence of the labor market in each of the regions where Bowman hires truck drivers was not necessary because its finding of adverse impact was based on a comparison of Bowman’s “women hirees for the relevant time period” with, inter alia, the general labor force and national labor pool of truck drivers “which are equally applicable to applications filed through the Atlanta terminal or through other terminals.” Id.

Turning to the merits, the court found that based on a comparison of Bowman’s hirees in the relevant time period with various national and local labor pools, Bowman’s prior experience requirement had an adverse impact upon female applicants. Id. at 1514-17, 1525-26. The court then concluded that Bowman had failed to establish that its prior experience requirement was a business necessity and that even if this requirement were a business necessity, there were less discriminatory alternatives available, and this requirement was used as a pretext for sex discrimination. Id. at 1517-21, 1526-27.

Finally, the district court held that Bowman had engaged in a pattern and practice of disparate treatment of female applicants for OTR driving positions. Id. at 1522-25, 1527-28. In reaching this conclusion, the district court relied on plaintiffs’ statistical evidence and testimony concerning Bowman’s disparate treatment of female applicants. Id.

In 1984, the district court issued a remedial order. Kilgo v. Bowman Transportation, Inc., 576 F.Supp. 600 (N.D.Ga.1984). In this order, the court refused to impose hiring goals or quotas but indicated that it would reconsider this decision “at a later time if [Bowman] fails to make a special effort to insure that women are given the opportunity to achieve OTR positions.” Id. at 601. This order also required Bowman to file a hiring plan to remedy its past discriminatory practices. Id. at 602-03.

Bowman filed its plan on June 1, 1984. Pursuant to the plan, on June 15, 1984, Bowman sent out “offers” of employment to plaintiffs, which provided in relevant part:

Bowman Transportation, Inc. is now currently hiring road drivers. There are a number of vacancies at the present time. Because you are a purported class member in the case of Kilgo v. Bowman Transportation, a case pending in the U.S. District Court for the Northern District of Georgia, in which there has been a finding of sex discrimination by Bowman in the hiring of road drivers, you are hereby offered the opportunity of employment as a road driver with Bowman Transportation.

If you are interested, you may contact the company in person or by mail at Bowman Transportation____

As required by law in the employment of all drivers, you must fill out a current job application and meet any other legal requirements and qualifications. You are invited and encouraged to submit a current application, one of which is enclosed. If you are minimally qualified, there is a good chance you will be hired, provided you contact the company within 15 days of receipt of this letter. If this letter has taken an unusually long period in reaching you, you should nevertheless contact the person and address given above, because Bowman may still be hiring.

If we do not hear from you in some manner within 15 days of receipt of this letter, we will assume that you are not interested in employment with the company.

This offer is made without prejudice to any prior claims for back pay or other relief, and your acceptance of this offer in no way constitutes a waiver of any previous claims____

Supp. Record on Appeal, Case No. 85-8409, vol. 2 at 8-9.

On June 21, 1984, plaintiffs moved to have Bowman, its officers, and its attorneys held in contempt⁴ of the remedial order. Plaintiffs contended that Bowman’s plan did not comply with the remedial order in four respects: (1) the plan did not provide for a co-driver training program; (2) the plan did not contain any provision for recruiting women for city driving positions which sometimes lead to OTR driving positions; (3) an applicant for an OTR driving position was required to have no more than a six-month gap in employment during the three years preceding the date of her application; and (4) the plan referred to heavy loading and unloading which OTR driving positions allegedly entailed. With respect to the first, second and fourth deficiencies, the plan departed from the express requirements of the district court’s order.

The special master recommended that Bowman be held in civil contempt for violating the remedial order. Although he did not recommend fining Bowman or incarcerating its officers or attorneys because Bowman had purged its contempt by filing a plan that complied with the remedial order, the master did recommend that the contempt “be considered in connection with the issue of back pay, if any, due the class members and any notice sent under the noncomplying plan cannot affect or ter-mínate their right, if any, to back pay.” Supp. Record on Appeal, vol. 1 at 169. The district court adopted the recommendation of the special master, and found Bowman in civil contempt of court from June 1984 until August 1984. This appeal ensued.⁵

II. ANALYSIS

On appeal, Bowman challenges the district court’s findings of disparate impact and disparate treatment, the certification of the instant case as a class action, and the validity of the provision in the contempt order concerning potential back pay liability.⁶ On cross-appeal, plaintiffs argue that the district court abused its discretion by refusing to impose hiring goals in the instant case.

A. Disparate Impact

The Supreme Court has established a three-step analysis for evaluating the evidence in disparate impact suits. See, e.g., Connecticut v. Teal, 457 U.S. 440, 446-47, 102 S.Ct. 2525, 2530-31, 73 L.Ed.2d 130 (1982); Albemarle Paper Co. v. Moody, 422 U.S. 405, 425, 95 S.Ct. 2362, 2375, 45 L.Ed.2d 280 (1975). In order to establish a prima facie case of discrimination, a plaintiff must show by a preponderance of the evidence that a facially neutral employment practice had a substantial adverse impact upon a protected group. See, e.g., Teal, 457 U.S. at 446, 102 S.Ct. at 2530; Maddox v. Claytor, 764 F.2d 1539, 1548 (11th Cir. 1985). If the plaintiff makes this showing, the burden of persuasion shifts to the defendant to show that its practice is a business necessity. See, e.g., Teal, 457 U.S. at 446-47, 102 S.Ct. at 2530-31; Georgia State Conference of Branches of NAACP v. Georgia, 775 F.2d 1403, 1417 (11th Cir. 1985). If the defendant establishes a business necessity defense, the burden of persuasion then shifts to the plaintiff to show the availability of other alternatives with a lesser adverse impact which would serve the defendant’s legitimate needs or that the defendant was using the practice as a pretext for discrimination. See, e.g., Teal, 457 U.S. at 447, 102 S.Ct. at 2530; Walker v. Jefferson County Home, 726 F.2d 1554, 1559 (11th Cir.1984).

1. Prima Facie Case

Bowman raises several challenges to the district court’s finding that its prior experience requirement had an adverse impact on female applicants. Bowman’s first contention is that the district court improperly disregarded the applicant flow data. Bowman argues that the applicant flow data from the Atlanta terminal for the period from 1976 to 1980 would have established that its prior experience rule had not had a disparate impact upon female applicants.⁷ The district court, however, found that the applicant flow data was not an appropriate labor pool both because the experience requirement posted in Atlanta was more likely to deter women than men from applying and because the record showed that some women were not allowed to file applications with Bowman. 570 F.Supp. at 1517.

The Supreme Court has held that applicant flow data may be relevant in determining whether a facially neutral employment practice has a disparate impact in a particular case, Hazelwood School District v. United States, 433 U.S. 299, 308 n. 13, 97 S.Ct. 2736, 2742 n. 13, 53 L.Ed.2d 768 (1977), but the Court has rejected the requirement that a prima facie case of disparate impact must be based on an analysis of the characteristics of actual applicants, Do-thard v. Rawlinson, 433 U.S. 321, 330, 97 S.Ct. 2720, 2727, 53 L.Ed.2d 786 (1977). As the Court explained in Dothard, “[t]he application process might itself not adequately reflect the actual potential applicant pool, since otherwise qualified people might be discouraged from applying because of a self-recognized inability to meet the very standards challenged as being discriminatory.” 433 U.S. at 330, 97 S.Ct. at 2727. See also, e.g., Moore v. Hughes Helicopters, Inc., 708 F.2d 475, 482 (9th Cir.1983) (applicant flow data inappropriate because individuals who lack the challenged employment requirements such as a high school education “will self-select themselves out of the pool of applicants”); Wheeler v. City of Columbus, 686 F.2d 1144, 1152 (5th Cir. 1982) (applicant flow data should be treated with “[cjaution” because this data “ ‘is often distorted, by inadequate or excessive recruiting efforts, improper deterrence] of applicants, unqualified applicants, multiple applications by the same applicant, or lack of specificity or improper groupings’ ”) (quoting B. Schlei & P. Grossman, Employment Discrimination Law 320-21 (Supp. 1979)); Castaneda v. Pickard, 648 F.2d 989, 1003 (5th Cir. Unit A 1981) (applicant flow data may not “constitute an accurate picture of the relevant labor market” where “the employer’s discriminatory practices infect recruiting, the process by which applications are solicited”).⁸ See generally B. Schlei & P. Grossman, Employment Discrimination Law 1348-51 (2d ed. 1983).

In the instant case, we conclude that the district court’s rejection of Bowman’s applicant flow data is not clearly erroneous. See, e.g., Markey v. Tenneco Oil Co., 707 F.2d 172, 174-75 (5th Cir.1983) (applying “clearly erroneous” standard to district court’s finding that applicant flow data was not skewed by discriminatory recruiting practices); United States v. East Texas Motor Freight, Inc., 643 F.2d 304, 307 (5th Cir. Unit A 1981) (“clearly erroneous” standard is applicable to district court’s finding that five employees would have applied for OTR jobs “but for their knowledge that doing so would have been futile” due to racial discrimination). The posted experience requirement at defendant’s Atlanta terminal was more likely to deter women than men from filing applications because women are less likely than men to satisfy this requirement.⁹ The record also indicates that Virginia Wentz was not permitted to file a written application at Bowman’s Atlanta terminal in 1979, Record on Appeal, vol. 9 at 16, and that Edna Kilgo was not allowed to do so until after she had filed a charge of discrimination with the EEOC, id. vol. 10 at 145-51.

Bowman’s next contention is that the district court erred as a matter of law in its selection of the relevant labor market because the labor pools selected by the district court were geographically and temporally irrelevant. We disagree.

Contrary to Bowman’s assertion, we conclude that the determination of the relevant labor market in the instant case is essentially a factual inquiry, reviewable under the “clearly erroneous” standard.¹⁰ In Castaneda, 648 F.2d at 1003, the former Fifth Circuit remanded the question of the relevant labor market to the district court for further findings because “determination of the relevant labor market, the geographical area from which we might reasonably expect [the school district] to draw applicants and teachers ... is an essentially factual matter within the special competence of the district court.” Similarly, in Markey v. Tenneco Oil Co., 635 F.2d 497, 499 (5th Cir.1981), the former Fifth Circuit concluded that the trial court should be afforded “a great deal of discretion” in ascertaining the relevant labor market “because statistics come in infinite variety, and because their value depends on all of the surrounding facts and circumstances.” Other courts have held that the “clearly erroneous” standard of review is generally applicable to the trial court’s selection of the relevant labor market. See, e.g., Castaneda ex rel. Castaneda v. Pickard, 781 F.2d 456, 464 (5th Cir.1986); De Medina v. Reinhardt, 686 F.2d 997, 1004 (D.C.Cir. 1982). Since the determination of the relevant labor market in the instant case depends upon the resolution of essentially factual questions such as whether national statistics are appropriate because of Bowman’s recruiting practices, we review the district court’s determination of the relevant labor market under the “clearly erroneous” standard.¹¹

Turning to the merits of Bowman’s challenge, we find that the district court did not err in determining the relevant labor market. The district court found that the percentage of female truck drivers nationally in 1980 (2.2%) was an appropriate labor pool because Bowman “hires over-the-road drivers throughout the Southeastern United States, as opposed to only in one or two cities.” ¹² 570 F.Supp. at 1515. There is ample support in the record for this finding. For example, as of 1979, Bowman had terminals throughout the Southeastern United States,¹³ and had regional hiring terminals in Atlanta, Georgia; Birmingham, Alabama; Charlotte, North Carolina; Jackson, Mississippi; Louisville, Kentucky; Richmond, Virginia; and Savannah, Georgia, Record on Appeal, vol. 1 at 74-75. In the absence of evidence indicating that from 1976 to 1980 the percentage of women truckdrivers in the region would differ significantly from the national figure,¹⁴ we conclude that national statistics were appropriate in the instant case. We also find no error in the district court’s comparison of the number of women hired as OTR drivers from 1976 to 1980 with the number of female truck drivers nationally in 1980.¹⁵ The 1980 data is clearly relevant to the period at issue here, and the district court also relied on other statistics which bracketed this period (1970 and 1980 data from Georgia and the Atlanta SMSA).

The district court also used 1970 and 1980 data from Georgia and the Atlanta SMSA to support its finding of disparate impact. The court compared the percentage of women hired by Bowman with the percentage of female truck drivers in Georgia in 1970 (1.96%), the percentage of female truck drivers, bus drivers, routemen, and deliverymen in Georgia in 1970 (4.06%), the percentage of female Class V license holders (the license required to drive a tractor-trailer) in Georgia in 1980 (2.44%), the percentage of female truck drivers in the Atlanta SMSA in 1970 (2.1%), and the percentage of unemployed women in motor freight occupations in Georgia in 1980 (2.15%). 570 F.Supp. at 1516. These figures are geographically relevant since Bowman hired many drivers from the Atlanta and Georgia areas. Moreover, these figures were temporally relevant since they bracketed the period at issue in the instant case and showed that the percentage of women in trucking occupations in Georgia and Atlanta had increased during this period. Thus, we find no error in the district court’s use of these figures.

Finally, Bowman contends that even if the various labor statistics relied upon by the district court were relevant, there was no evidence of adverse impact. Bowman asserts that in order for a statistical disparity to be probative of discrimination, the difference between the expected value and the observed number must be greater than three standard deviations.¹⁶ Since the comparison of the percentage of women hired by Bowman as OTR drivers with the various labor pools generally resulted in standard deviations less than three,¹⁷ Bowman claims that this disparity was insufficient to establish that its one-year experience requirement had a disparate impact.¹⁸ We find no merit in this contention.

In Maddox v. Claytor, 764 F.2d 1539 (11th Cir.1985), we held that one should be cautious in drawing inferences of discrimination from standard deviations in the range of one to three and that the significance of the statistical evidence should be evaluated in light of all the surrounding facts and circumstances:

[I]t is important to note that no bright line of two, three or more standard deviations can be drawn to distinguish automatically between fair and discriminatory employment practices. Just as a standard deviation of two or three does not necessarily exclude legitimate causes other than chance, so a [standard deviation] below that range does not necessarily exclude discrimination as the cause____ The Supreme Court has repeatedly warned, “We caution only that statistics are not irrefutable; they come in infinite variety and, like any other kind of evidence, may be rebutted. In short, their usefulness depends on all the surrounding facts and circumstances.”

Id., 1552 (quoting Teamsters, 431 U.S. at 340, 97 S.Ct. at 1856-57) (citation omitted). Other courts have also adopted this approach. See, e.g., Craik v. Minnesota State University Board, 731 F.2d 465, 475-76 n. 13 (8th Cir.1984); Gay v. Waiters’ & Dairy Lunchmen’s Union, 694 F.2d 531, 551-53 (9th Cir.1982); EEOC v. American National Bank, 652 F.2d 1176, 1192 (4th Cir.1981), cert. denied, 459 U.S. 923, 103 S.Ct. 235, 74 L.Ed.2d 186 (1982); Chang v. University of Rhode Island, 606 F.Supp. 1161, 1189 (D.R.I.1985).

After examining plaintiffs’ statistical evidence and the surrounding circumstances,¹⁹ we conclude that the district court’s finding of disparate impact is not clearly erroneous.²⁰ A comparison of the percentage of women OTR drivers hired by Bowman with the various labor pools described above reveal disparities ranging from 1.92 standard deviations to 5.3 standard deviations. The statistical evidence was also bolstered by a variety of nonstatistical evidence. For example, the record indicates that Bowman had used its prior experience requirement as a pretext for discriminating against female applicants. See infra Part II.A.2. Bowman also used various devices to discourage female applicants for OTR driving positions — e.g., informing them that they might be assigned to co-drive with a male driver of another race, even though racially integrated driving teams were extremely rare at Bowman. See infra Part II.B.

2. Pretext

Bowman contends that even if its prior experience requirement had a disparate impact upon women, the district court erred in finding that this requirement was not a business necessity, and that there were less discriminatory alternatives available. We do not reach this contention, however, because we find that even if its experience requirement were a business necessity and even if there were no less discriminatory alternatives that would serve Bowman’s legitimate business needs, Bowman used this requirement as a pretext for discrimination against women.

In the instant case, the district court found that Bowman had used the prior experience requirement “merely as a pretext for discrimination.” 570 F.Supp. at 1527. The court noted that Bowman had hired over 60 men who did not satisfy the prior experience requirement as OTR drivers, while the prior experience requirement was strictly enforced against female applicants. Id. at 1522-24.

Bowman contends that even if it had used this requirement as a pretext to discriminate against women, pretext is irrelevant in a disparate impact case. This argument, however, is without merit. Courts have been virtually unanimous in holding that even if an employer shows that its employment practice is a business necessity, the plaintiff may prevail by proving that this practice is a pretext for discrimination. See, e.g., Connecticut v. Teal, 457 U.S. 440, 447, 102 S.Ct. 2525, 2530, 73 L.Ed.2d 130 (1982); Georgia State Conference of Branches of NAACP v. Georgia, 775 F.2d 1403, 1417 (11th Cir.1985); Atonio v. Wards Cove Packing Co., 768 F.2d 1120, 1131 (9th Cir.1985); Easley v. Anheuser-Busch, Inc., 758 F.2d 251, 255 n. 7 (8th Cir.1985); Merwine v. Board of Trustees for State Institutions of Higher Learning, 754 F.2d 631, 639 (5th Cir.), cert. denied, — U.S.-, 106 S.Ct. 76, 88 L.Ed.2d 62 (1985); Lasso v. Woodmen of World Life Insurance Co., 741 F.2d 1241, 1244 n. 1 (10th Cir.1984), cert. denied, — U.S.-, 105 S.Ct. 2320, 85 L.Ed.2d 839 (1985); Walker v. Jefferson County Home, 726 F.2d 1554, 1559 (11th Cir.1984). But see Levin v. Delta Air Lines, 730 F.2d 994, 999 (5th Cir.1984).

Reviewing the district court’s finding of pretext under the clearly erroneous standard,²¹ we conclude that there is ample evidence in the record to support this finding. For example, over sixty men were hired by Bowman for OTR driver positions even though they did not satisfy the prior experience requirement, see 570 F.Supp. at 1522, but Bowman refused to hire several female applicants who met the prior experience requirement, see, e.g., id. at 1523-24, and strictly enforced the prior experience requirement against women applicants who did not fully comply with it, see, e.g., id. at 1522-23. Moreover, there are no written criteria for determining the circumstances under which individuals who do not fully comply with the prior experience requirement will be hired. Record on Appeal, vol. 10 at 24-41, vol. 16 at 1125. Finally, Mr. Jule Wood, Bowman’s director of safety and personnel, informed Mr. Tarvin, a driver for Bowman, that Bowman would hire his wife, even though she did not meet Bowman’s prior experience requirement, if Mr. Tarvin obtained a letter from someone indicating that she had at least one year of OTR driving experience. Id. vol. 9 at 62-63.

B. Disparate Treatment

Bowman’s next contention is that the district court’s finding of a pattern and practice of disparate treatment of female applicants for OTR driving positions is clearly erroneous.²² We find no merit in this contention.

In an action alleging class-wide disparate treatment, the plaintiffs must “establish by a preponderance of the evidence that racial discrimination was the company’s standard operating procedure — the regular rather than the unusual practice.” International Brotherhood of Teamsters v. United States, 431 U.S. 324, 336, 97 S.Ct. 1843, 1855, 52 L.Ed.2d 396 (1977). A prima facie case of class-wide disparate treatment may be established by statistics alone if they are sufficiently compelling. See, e.g., Griffin v. Carlin, 755 F.2d 1516, 1525 (11th Cir.1985); Eastland v. Tennessee Valley Authority, 704 F.2d 613, 618 (11th Cir. 1983), cert. denied, 465 U.S. 1066, 104 S.Ct. 1415, 79 L.Ed.2d 741 (1984). The prima facie case will be enhanced if the plaintiffs offer anecdotal evidence to bring “the cold numbers convincingly to life.” Teamsters, 431 U.S. at 339, 97 S.Ct. at 1856.

Once the plaintiffs have established a prima facie case of disparate treatment, the burden of production shifts to the defendant to rebut the inference by showing that the plaintiffs’ statistics are misleading or by presenting legitimate nondiscriminatory reasons for the disparity. See, e.g., Griffin, 755 F.2d at 1525—26; Eastland, 704 F.2d at 618-19. The defendant does not have to persuade the court that it was actually motivated by the proffered reasons; it is sufficient if the defendant raises a genuine issue of fact as to whether it discriminated. See, e.g., Griffin, 755 F.2d at 1526; Perryman v. Johnson Products Co., 698 F.2d 1138, 1143-44 (11th Cir.1983). If the defendant carries this burden of production, the plaintiff must prove that the reasons offered by the employer were pretextual. See, e.g., Griffin, 755 F.2d at 1526; Perryman, 698 F.2d at 1145.

In the instant case, we find ample evidence in the record to support the district court’s finding of a pattern and practice of sex discrimination in Bowman’s hiring of OTR drivers. As discussed above in Part II.A.l, plaintiffs offered statistical evidence showing that its hiring rate for women applicants for OTR driving positions differed from several different labor pools in a range from 1.92 to 5.3 standard deviations. This statistical evidence is also relevant to the question of whether Bowman had engaged in a pattern and practice of discrimination.²³

Plaintiffs offered a variety of anecdotal evidence of discrimination to bolster the statistical evidence. For example, Bowman used many different devices to discourage women from becoming OTR drivers. Bowman refused to provide separate sleeping, shower and bathroom facilities for its women drivers, see, e.g., Record on Appeal, vol. 10 at 123-25, vol. 15 at 1014, and female applicants were informed about the lack of separate facilities, see, e.g., id. vol. 9 at 81. In fact, at a meeting between Bowman’s president, Garfield Sal-yer, and union representatives, the union representatives raised the subject of the lack of separate facilities for women drivers, and Salyer responded that Bowman would “under no circumstances” provide separate facilities for women drivers. Id. vol. 12 at 441-42.

Moreover, women were not hired as OTR drivers because of Bowman’s policy against women driving with men other than their husbands. For example, Jon Martin was told that “because I was a female that I could not be put into a truck with a male driver and that they did not have any female drivers that I could be put with, therefore, they could not hire me.” Id. vol. 9 at 32. Female applicants were also informed that they might be assigned to drive with a male driver of a different race, see, e.g., id. vol. 10 at 125-26, 131, vol. 11 at 308, vol. 12 at 444-45, even though racially integrated driving teams are extremely rare at Bowman, see, e.g., id. vol. 9 at 65-66, vol. 11 at 308, 310-11, vol. 12 at 427, 442. The district court found that such devices had been used to discriminate against women, 570 F.Supp. at 1527, and this finding is not clearly erroneous.

The district court also concluded that Bowman’s attempts to articulate nondiscriminatory reasons for its denial of employment to plaintiffs were “unconvincing” because “[t]he record is replete with instances where [Bowman’s] representatives testified at trial that a particular class member was denied employment because of a legitimate reason which contradicted the reasons stated originally for such denial.” Id. at 1528. The district court’s finding that these explanations were pretextual is amply supported by the record. For example, Mr. Wood testified at trial that Carolyn Hewitt, who had nine months of OTR experience and had graduated from a truck driving school, was rejected because her past employer stated that she could not unload a truck.²⁴ Record on Appeal, vol. 16 at 1035. In an earlier affidavit, however, Mr. Wood had stated that she was not hired because of a lack of prior experience, and did not mention her inability to unload trucks. Id. at 1035-36. Mr. Wood also conceded that he had approved the hiring of less qualified applicants. Id. at 1036-37.

Finally, as discussed above in Part II.A.2, Bowman used its prior experience requirement as a pretext for sex discrimination. For example, Bowman hired over sixty men who did not satisfy the prior experience requirement while it strictly enforced this requirement against female applicants. Under these circumstances, we hold that the statistical and anecdotal evidence in the instant case amply supports the district court’s finding of a pattern and practice of disparate treatment of female applicants for OTR driving positions.

C. Class Certification

Bowman raises three principal challenges to the class certified in the instant case. Bowman first argues that Edna Kilgo’s EEOC filing cannot be used to determine the temporal scope of the class. Bowman’s next contention is that the district court abused its discretion by expanding the disparate impact class after trial. Finally, Bowman asserts that the conditionally certified class failed to satisfy the numerosity requirement of Fed.R. Civ.P. 23. We address each of these arguments in turn.

Bowman offers two arguments in support of its contention that Kilgo’s EEOC filing date cannot be used to determine the temporal scope of the class. Bowman first contends that because Kilgo’s Title VII cause of action did not survive her death, her EEOC filing date cannot constitute the cut-off date for the class. Because we find that her Title VII cause of action survives under both federal and state law, this contention is without merit. Since her Title VII cause of action survives under either state or federal law, we do not decide whether the survival of a Title VII cause of action is governed by state or federal law.²⁵

In Brazier v. Cherry, 293 F.2d 401 (5th Cir.), cert. denied, 368 U.S. 921, 82 S.Ct. 243, 7 L.Ed.2d 136 (1961), the former Fifth Circuit held that under Georgia law, a decedent’s § 1983 claim for damages survives his death.²⁶ Thus, if state law were applicable, Kilgo’s Title VII cause of action would survive.

A Title VII cause of action would also survive under federal common law. In James v. Home Construction Co. of Mobile, 621 F.2d 727, 729-30 (5th Cir.1980), the former Fifth Circuit held that under federal common law, a federal cause of action generally survives the death of the plaintiff unless it is an action for penalties. In determining whether the plaintiff's cause of action under 15 U.S.C. § 1635 survived her death, the court looked to three factors: (1) whether the purpose of the action was to redress individual wrongs or more general wrongs to the public; (2) whether the recovery runs to the individual harmed or to the public; and (3) whether the recovery was wholly disproportionate to the harm suffered. 621 F.2d at 730. Accord Smith v. No. 2 Galesburg Crown Finance Corp., 615 F.2d 407, 414 (7th Cir. 1980).

Applying this analysis to the instant case, we find that Title VII of the Civil Rights Act of 1964 is a remedial, rather than a penal statute for the purpose of survival. Turning to the first consideration (the purpose of the action), the primary purposes of Title VII are to prevent discrimination and achieve equal employment opportunity in the future, see, e.g., International Brotherhood of Teamsters v. United States, 431 U.S. 324, 364, 97 S.Ct. 1843, 1869, 52 L.Ed.2d 396 (1977); Albe-marle Paper Co. v. Moody, 422 U.S. 405, 417, 95 S.Ct. 2362, 2371, 45 L.Ed.2d 280 (1975), and to make whole the victims of discrimination, see, e.g., Teamsters, 431 U.S. at 364, 97 S.Ct. at 1869; Franks v. Bowman Transportation Co., 424 U.S. 747, 764, 96 S.Ct. 1251, 1264, 47 L.Ed.2d 444 (1976). Although Title VII thus benefits both the individual and the public, we conclude that the primary purpose of Title VII is to remedy individual wrongs, not to punish employers. Recovery under Title VII also runs to the individual, not to the public at large. Finally, the authorized recovery under Title VII is not disproportionate to the harm suffered since, as described above, Title VII is intended to make whole the victims of discrimination.

Even if Kilgo’s cause of action survives her death, Bowman asserts that since she died before the certification of the class and since her husband was determined to be an inadequate class representative, her EEOC filing cannot be used to determine the temporal scope of the class. We also find no merit in this contention.

Section 2000e-5(e) requires a charge to be filed with the EEOC “within one hundred and eighty days after the alleged unlawful employment practice occurred.” 42 U.S.C.A. § 2000e-5(e) (West 1981). The timely filing requirement serves two primary purposes: to give notice to the charged party and to give the EEOC an opportunity to settle the grievance. See, e.g., Laffey v. Northwest Airlines, 567 F.2d 429, 472 (D.C.Cir.1976), cert. denied, 434 U.S. 1086, 98 S.Ct. 1281, 55 L.Ed.2d 792 (1978); Bowe v. Colgate-Palmolive Co., 416 F.2d 711, 729 (7th Cir.1969). The Supreme Court has held that “filing a timely charge of discrimination with the EEOC is not a jurisdictional prerequisite to suit in federal court, but a requirement that, like a statute of limitations, is subject to waiver, estoppel, and equitable tolling.” Zipes v. Trans World Airlines, 455 U.S. 385, 393, 102 S.Ct. 1127, 1132, 71 L.Ed.2d 234 (1982).

In cases involving Title VII class actions, we have long held that every member of the class is not required to file an EEOC charge; it is sufficient that at least one named plaintiff has filed such a charge. See, e.g., Crawford v. United States Steel Corp., 660 F.2d 663, 665 (5th Cir. Unit B 1981);²⁷ Oatis v. Crown Zellerbach Corp., 398 F.2d 496, 498-99 (5th Cir.1968). This rule does not frustrate either purpose of the timely filing requirement. First, the employer cannot claim to be surprised by the Title VII class action since he was given notice by the EEOC filing. See, e.g., Laffey, 567 F.2d at 472 & n. 325; Bowe, 416 F.2d at 720. Moreover, the filing of a timely EEOC charge by at least one named plaintiff satisfies the conciliation purpose of the timely filing requirement:

It would be wasteful, if not vain, for numerous employees, all with the same grievance, to have to process many identical complaints with the EEOC. If it is impossible to reach a settlement with one discriminatee, what reason would there be to assume the next one would be successful. The better approach would appear to be that once an aggrieved person raises a particular issue with the EEOC which he has standing to raise, he may bring an action for himself and the class of persons similarly situated....

Oatis, 398 F.2d at 498; see also Wheeler v. American Home Products Corp., 582 F.2d 891, 897 (5th Cir.1977).

However, Bowman argues that this rule only applies where the timely EEOC filing is made by one of the class representatives. Since Edna Kilgo died before the class was certified, and since her husband was determined to be an inadequate class representative, it claims that her EEOC filing cannot be used as the cut-off date for the class. We find no merit in this distinction. Kil-go’s filing of her EEOC charge satisfied both purposes of the timely filing requirement since Bowman had notice of her claim and the EEOC was given an opportunity to settle the grievance. Moreover, Kilgo’s Title VII cause of action survived her death, and the district court permitted the substitution of her husband as a party in this case, even though the court determined that her husband would not be an adequate class representative. On the facts of this case, we hold that the date of her EEOC filing may be used to determine the temporal scope of the class.

Bowman’s next contention is that the district court abused its discretion in expanding the disparate impact class after trial on the merits.²⁸ Rule 23(c)(1) grants the trial court broad discretion in the initial certification and subsequent amendment of a class. See, e.g., Perryman v. Johnson Products Co., 698 F.2d 1138, 1147 (11th Cir.1983). While it is apparent that expansion of a class after trial should be viewed with caution, we find no abuse of discretion in the instant case. The district court’s post-trial expansion of the disparate impact class to include company-wide applicants simply conformed to what had been proved at trial. Bowman has also not shown that it was prejudiced by the post-trial expansion of the class. Bowman argues that it was prejudiced because it could have introduced system-wide applicant flow data which it alleges would have been favorable to its position.²⁹ Bowman’s argument is without merit because, inter alia, we have held that Bowman’s deterrence of female applicants rendered such applicant flow data an inappropriate labor pool. See supra Part II.A.l. In addition, Bowman cannot assert that it was prejudiced because it did not have the opportunity and incentive to introduce evidence on the percentage of female truck drivers in the area covered by the expanded class. It was clear early in the litigation that the class would include persons who applied initially at all other Bowman terminals when the applications were processed through the Atlanta terminal; thus, Bowman was on notice that the relevant labor pool included all of Bowman’s terminals. Moreover, plaintiffs introduced a variety of evidence going beyond the Atlanta and Georgia areas (national labor force statistics and the national labor pool of truck drivers), and Bowman itself introduced national data on the number of self-employed female truck drivers. Finally, the composition of the class at issue could not have affected Bowman’s evidence, or the district court’s finding, with respect to whether or not Bowman’s one-year prior experience requirement was pretextual.

Bowman’s final contention is that the conditionally certified class failed to satisfy the numerosity requirement of Fed.R.Civ.P. 23. We will not reverse a district court’s finding that joinder is impracticable absent an abuse of discretion. See, e.g., Zeidman v. J. Ray McDermott & Co., 651 F.2d 1030, 1038-39 (5th Cir. Unit A 1981). Practicability of joinder depends on many factors, including, for example, the size of the class, ease of identifying its numbers and determining their addresses, facility of making service on them if joined and their geographic dispersion. See, e.g., Zeidman, 651 F.2d at 1038; Garcia v. Gloor, 618 F.2d 264, 267 (5th Cir.1980), cert. denied, 449 U.S. 1113, 101 S.Ct. 923, 66 L.Ed.2d 842 (1981).

In the instant case, we conclude that the district court did not abuse its discretion in finding that the numerosity requirement had been met. Plaintiffs have identified at least thirty-one individual class members, and the class includes future and deterred job applicants, which of necessity cannot be identified. The certified class also includes applicants from a wide geographical area.³⁰

D. Contempt Order

On appeal, Bowman claims that the provision in the contempt order regarding back pay — i.e., indicating that the contempt would be considered in connection with the back pay issue — is neither compensatory nor coercive, and thus must be deemed punitive.³¹ Since punitive remedies may be imposed only for criminal contempt, and since the procedural requirements for criminal contempt were not followed in this case, Bowman concludes that the court’s contempt order is invalid. We disagree.

Bowman’s contention that the back pay provision in the contempt order is punitive rests on the assumption that the “offers” of employment under the noncomplying plan constituted a bona fide unconditional offer of employment that would toll potential back pay liability under the Supreme Court’s decision in Ford Motor Co. v. EEOC, 458 U.S. 219, 102 S.Ct. 3057, 73 L.Ed.2d 721 (1982). This assumption, however, is flawed. In Ford Motor Co., the Supreme Court held that the accrual of potential back pay liability would be tolled by a claimant’s rejection of the employer’s unconditional offer of a job substantially equivalent to the one he was denied:

An unemployed or underemployed claimant ... is subject to the statutory duty to minimize damages set out in [42 U.S.C. § 2000e-5(g) ]".... Although the unemployed or underemployed claimant need not go into another line of work, accept a demotion, or take a demeaning position, he forfeits his right to back pay if he refuses a job substantially equivalent to the one he was denied. Consequently, an employer charged with unlawful discrimination often can toll the accrual of back pay liability by unconditionally offering the claimant the job he sought, and thereby providing him with an opportunity to minimize damages.

458 U.S. at 231-32, 102 S.Ct. at 3065-66 (footnotes omitted).

The instant case differs from Ford Motor Co. An examination of Bowman’s “offer” shows that it was not unconditional. For example, the offer stated that the claimant was invited “to submit an application” for one of an undisclosed “number of vacancies,” and that there was a “good chance” the claimant would be hired if she were “minimally qualified.” Supp. Record on Appeal, Case No. 85-8409, vol. 2 at 8-9. Since the plaintiffs were not offered employment, but simply the opportunity to apply for an unspecified number of jobs, we conclude that Bowman’s offer did not satisfy the requirements of Ford Motor Co. See Rasimas v. Michigan Department of Mental Health, 714 F.2d 614, 625 (6th Cir.1983) (an interview letter is not an unconditional offer of employment), cert. denied, 466 U.S. 950, 104 S.Ct. 2151, 80 L.Ed.2d 537 (1984); Orzel v. City of Wauwatosa Fire Department, 697 F.2d 743, 757 (7th Cir.) (offer of reinstatement conditioned upon plaintiff’s taking and passing a physical exam did not toll back pay liability), cert. denied, 464 U.S. 992, 104 S.Ct. 484, 78 L.Ed.2d 680 (1983); cf. Romasanta v. United Air Lines, Inc., 717 F.2d 1140, 1158-59 (7th Cir.1983) (rejecting argument that the failure of the plaintiffs to seek available rehire by United weighed in favor of United on the award of seniority to the plaintiffs, because airline’s “offer” of employment consisted of a publicity statement “to the effect that priority would be given to applicants who had previously worked in positions involving ‘customer service’ ”), cert. denied, 466 U.S. 944, 104 S.Ct. 1928, 80 L.Ed.2d 474 (1984). Since Bowman’s “offer” of employment could not toll the accrual of potential back pay liability, we hold that the statement to that effect in the contempt order is not punitive.

E. Hiring Goals

On cross-appeal, plaintiffs contend that the district court abused its discretion in refusing to impose hiring goals in its remedial order.³² They assert that the court did not carefully consider the appropriateness of hiring goals in the instant case, but rather denied plaintiffs’ request merely because of a “personal predeliction against goals and quotas.” Appellees’ Brief at 77. We find no merit in this contention.

Under Title VII, courts have been granted broad equitable powers to remedy statutory violations by fashioning the most complete relief possible. See, e.g., 42 U.S.C.A. § 2000e-5(g); Franks v. Bowman Transportation Co., 424 U.S. 747, 763-64, 96 S.Ct. 1251, 1263-64, 47 L.Ed.2d 444 (1976); Albemarle Paper Co. v. Moody, 422 U.S. 405, 421, 95 S.Ct. 2362, 2373, 45 L.Ed.2d 280 (1975); James v. Stockham Valves & Fittings Co., 559 F.2d 310, 354 (5th Cir. 1977), cert. denied, 434 U.S. 1034, 98 S.Ct. 767, 54 L.Ed.2d 781 (1978). The district courts have “ ‘not merely the power but the duty to render a decree which will so far as possible eliminate the discriminatory effects of the past as well as bar like discrimination in the future.’ ” Albemarle Paper Co., 422 U.S. at 418, 95 S.Ct. at 2372 (quoting Louisiana v. United States, 380 U.S. 145, 154, 85 S.Ct. 817, 822, 13 L.Ed.2d 709 (1965)). The particular remedy granted is left largely to the broad discretion of the district court. See, e.g., Franks, 424 U.S. at 764, 96 S.Ct. at 1264; Williams v. Owens-Illinois, Inc., 665 F.2d 918, 931-32 (9th Cir.), cert. denied, 459 U.S. 971, 103 S.Ct. 302, 74 L.Ed.2d 283 (1982); Association Against Discrimination in Employment, Inc. v. City of Bridgeport, 647 F.2d 256, 278-79 (2d Cir.1981), cert. denied, 455 U.S. 988, 102 S.Ct. 1611, 71 L.Ed.2d 847 (1982); NAACP v. Allen, 493 F.2d 614, 617 (5th Cir.1974). However, we have cautioned that hiring goals are “a form of relief which should be reserved for those situations in which less restrictive means have failed or in which the chancellor could reasonably foresee that they would fail.” Allen, 493 F.2d at 621; see also Segar v. Smith, 738 F.2d 1249, 1294 (D.C.Cir.1984), cert. denied, — U.S.-, 105 S.Ct. 2357, 86 L.Ed.2d 258 (1985); Dawson v. Pastrick, 600 F.2d 70, 77 (7th Cir.1979).

In the instant case, we conclude that the district court did not abuse its discretion. Although no hiring goals were imposed, the order does eliminate the discriminatory practices in the hiring of OTR drivers, and places an affirmative duty on Bowman to recruit women for its OTR driving positions. 576 F.Supp. at 602-04. The order also required Bowman to make periodic progress reports to the special master, and the court retained jurisdiction over the case to insure compliance with its decree. Id. at 603-04. Finally, the court noted that the question of hiring goals would be “reconsidered at a later time if [Bowman] fails to make a special effort to insure that women are given the opportunity to achieve OTR positions.” Id. at 601. Thus, contrary to plaintiffs’ assertion, the district court did not deny hiring goals because of any personal opposition to such remedies, but rather because he found that less restrictive means would be effective.³³ The determination of the effectiveness of other remedies is committed to the sound discretion of the trial court, and we find no abuse of discretion in the instant case.³⁴ See, e.g., Dawson, 600 F.2d at 77 (no abuse of discretion in denying hiring goals where less restrictive means would be effective).

III. CONCLUSION

For the foregoing reasons, the judgment of the district court is

AFFIRMED.

1 . On our own motion, we have consolidated Case No. 85-8409 with Case Nos. 84-8105 and 84-8899. Fed.R.App.P. 3(b).

2 . Unless otherwise indicated, the term "Record on Appeal” will refer to the record in Case Nos. 84-8105 and 84-8899.

3 . The district court did not, however, redefine the class with respect to the disparate treatment claim. This class continued to include (1) those applicants who were either hired by defendant as an over-the-road driver but were terminated within the forty-five day probationary period, or were not hired as an over-the-road driver but their applications were forwarded to the Atlanta terminal for review; and (2) prospective applicants whose applications, if they had been filed, would have been forwarded to the Atlanta terminal.

4 . Plaintiffs and Bowman disagree about whether plaintiffs had given Bowman adequate notice that they were seeking criminal as well as civil contempt. Since the district court only found Bowman to be in civil contempt, we will assume that plaintiffs have not complied with the notice requirements for criminal contempt.

5 . We raised on our own motion the question of our jurisdiction over the appeal in Case No. 84-8105. In light of our consolidation of this case with Case No. 84-8899, we withheld adjudication on the issue of jurisdiction "until the two appeals are considered jointly, which may render the jurisdictional issue moot.” Kilgo v. Bowman Transportation, Inc., No. 84-8105 (11th Cir. Mar. 6, 1985) (unpublished order). We now hold that we have jurisdiction over the instant appeal.

The relevant chronology of events is as follows. On January 5, 1984, the district court entered a remedial order, which, inter alia, enjoined Bowman from using certain discriminatory employment practices and required Bowman to file a plan for the recruitment of women to be OTR drivers. On January 12, Bowman timely served a motion for a new trial under Fed.R.Civ.P. 59. On February 3, while this motion was still pending, Bowman filed its first notice of appeal (Case No. 84-8105). On October 5, the district court denied Bowman’s motion for a new trial, and Bowman filed a timely second notice of appeal (Case No. 84-8899).

Federal Rule of Appellate Procedure 4(a)(4) provides that a notice of appeal filed before the disposition of a Rule 59 motion for a new trial is a nullity, and that a new notice of appeal must be filed within the prescribed time from the denial of this motion:

If a timely motion under the Federal Rules of Civil Procedure is filed in the district court by any party: ... (iv) under Rule 59 for a new trial, the time for appeal for all parties shall run from the entry of the order denying a new trial.... A notice of appeal filed before the disposition of any of the above motions shall have no effect. A new notice of appeal must be filed within the prescribed time measured from the entry of the order disposing of the motion as provided above.

Fed.R.App.P. 4(a)(4); see also Bolden v. Odum, 695 F.2d 549, 550 (11th Cir.1983); Martin v. Campbell, 692 F.2d 112, 114 (11th Cir.1982). Bowman contends that its motion for a new trial did not nullify its first notice of appeal because Rule 4(a)(4) applies only to appeals from final orders, not to appeals from interlocutory injunctions and because an appeal from an interlocutory injunction, unlike an appeal from a final order, does not divest the district court of jurisdiction. We do not reach the question of whether Bowman’s first notice of appeal was effective because we find that Bowman complied with the refiling requirement of Rule 4(a)(4). Since Bowman filed its second notice of appeal within thirty days after the denial of its motion for a new trial, we hold that we have jurisdiction over the instant appeal. See, e.g., Great American Insurance Co. v. Rush, 670 F.2d 995, 996 (11th Cir.1982).

6 . Bowman also contends that the district court abused its discretion in denying Bowman’s motion for a new trial based on a newly discovered Department of Transportation report ("DOT report”) which concluded, inter alia, that lack of experience as a truck driver is “the best predictor of accident involvement.” Record on Appeal, vol. 7 at 1530. Bowman claims that since this report would have affected the district court’s finding that Bowman’s prior experience requirement was not a business necessity, the district court should have granted its motion for a new trial. However, in the instant case, we conclude that the district court properly denied the motion for a new trial because the new evidence would not have affected the result of the trial. The DOT report at best would have supported Bowman’s contention that its prior experience requirement was a business necessity, but since the district court found that even if this requirement were a business necessity, Bowman had used it as a pretext for discrimination, and since we hold that this finding is not clearly erroneous, see infra Part II.A.2, the DOT report would not have changed the result.

7 . Bowman offered evidence that its applicant flow logs for the Atlanta terminal during this period showed that, without considering reapplications, it had hired 21.4% of the female applicants and 16.2% of the male applicants, and that, considering reapplications, 20.7% of female applicants and 16.1% of male applicants had been hired. Record on Appeal, vol. 13 at 672-76.

8 . In Bonner v. City of Prichard, 661 F.2d 1206 (11th Cir.1981) (en banc), this court adopted as binding precedent ail of the decisions of the former Fifth Circuit handed down prior to the close of business on September 30, 1981. Id. at 1209.

9 . During the period from 1976 to 1980, only 1.27% of the applicants whose files had been forwarded to Atlanta were women, see Record on Appeal, vol. 3 at 529, even though women comprised over 2% of all truck drivers nationwide and in Atlanta and Georgia, see infra at 870-871. Moreover, only 56% of the female applicants satisfied the prior experience requirement while 87% of the male applicants met this requirement, thus indicating that women are less likely than men to satisfy this requirement. Record on Appeal, vol. 19 at 1485.

10 . Relying on Hazelwood School District v. United States, 433 U.S. 299, 97 S.Ct. 2736, 53 L.Ed.2d 768 (1977) and Griffin v. Carlin, 755 F.2d 1516 (11th Cir.1985), Bowman asserts that the selection of the relevant labor market is a question of law subject to de novo review on appeal. See Appellant’s Reply Brief at 2, 9-19. These cases, however, do not support Bowman. In Hazel-wood, the Supreme Court remanded the case to the district court for further findings of fact with respect to the relevant labor market because the district court had failed to determine whether the relevant labor market included teachers in all of St. Louis County or only those teachers in St. Louis County excluding the city of St. Louis. Indeed, the Supreme Court emphasized the factual nature of the inquiry into the relevant labor market:

In determining [the relevant labor market], it will be necessary to evaluate such considerations as (i) whether the racially based hiring policies of the St. Louis City School District were in effect as far back as 1970, the year in which the census figures were taken; (ii) to what extent those policies have changed the racial composition of that district’s teaching staff from what it would otherwise have been; (iii) to what extent St. Louis’ recruitment policies have diverted to the city, teachers who might otherwise have applied to Hazelwood; (iv) to what extent Negro teachers employed by the city would prefer employment in other districts such as Hazelwood; and (v) what the experience in other school districts in St. Louis County indicates about the validity of excluding the City School District from the relevant labor market.

It is thus clear that a determination of the appropriate comparative figures in this case will depend upon further evaluation by the trial court. As this Court admonished in Teamsters: "[Statistics ... come in infinite variety____ [T]heir usefulness depends on all the surrounding facts and circumstances.” ... Only the trial court is in a position to make the appropriate determination after further findings____

433 U.S. at 311-12, 97 S.Ct. at 2743-44 (footnote omitted) (quoting International Brotherhood of Teamsters v. United States, 431 U.S. 324, 340, 97 S.Ct. 1843, 1856, 52 L.Ed.2d 396 (1977)).

Griffin is also distinguishable from the instant case. In Griffin, we held that the appropriate labor pool was, as a matter of law, the craft work force, rather than those employees on the supervisory registers because "[w]hen promotions to supervisory positions are made almost exclusively from the internal work force and when the primary qualification for promotion is experience in the craft work force, the appropriate comparison is to this work force rather than to those on the supervisory register, who have already been screened by the agency through the use of various procedures.” 755 F.2d at 1526. The criteria used for determining the employees who would be placed on a supervisory register — e.g., written examinations and supervisory evaluations — were also challenged as being discriminatory. Id. at 1526-28. In the instant case, however, the determination of the relevant labor market depended upon the resolution of essentially factual questions — e.g., whether national statistics were appropriate because of Bowman’s recruiting practices — and did not involve the use of labor pools that could never be appropriate — e.g., supervisory registers created by discriminatory practices.

11 . We do not hold that the selection of the relevant labor market is always governed by the “clearly erroneous” standard, see, e.g., Griffin, 755 F.2d at 1526, but rather conclude that this standard is appropriate in the instant case.

12 . The district court also compared the number of women hired as OTR drivers with national labor force statistics. Bowman challenges the relevance of such statistics to the instant case because it contends that, as the result of a number of factors, women are less interested than men in becoming truck drivers. Because we find that the other labor pools relied upon by the district court and the nonstatistical evidence establish a prima facie case of disparate impact, we do not reach the question of whether general labor force statistics were relevant in the instant case.

13 . As of 1979, Bowman had company-owned and cartage terminals in Alabama, Arkansas, Florida, Georgia, Indiana, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Ohio, Rhode Island, South Carolina, Tennessee, Texas, and Virginia. Record on Appeal, vol. 1 at 73-74.

14 . The only evidence that arguably shows that the national and regional statistics were different during the period from 1976 to 1980 is a 1970 study of the six standard metropolitan statistical areas ("SMSA”) where Bowman has regional hiring terminals. This study revealed that 1.65% of the truck drivers in these six SMSA’s were women. We conclude, however, that the district court properly discounted the significance of this evidence. First, this data is from 1970, and there is no evidence in the record as to what the percentage would have been for the period at issue here (1976-1980). Moreover, the significance of this evidence is limited by the fact that it only covered the six cities where Bowman has regional hiring terminals, even though Bowman operates and hires individuals throughout the Southeastern United States.

15 . Bowman contends that the district court erred by not relying upon 1977 data indicating that women comprised approximately 1.3% of all truck drivers nationwide. However, the 1977 study is not part of the record on appeal, although it is referred to in the testimony. The district court did not mention the 1977 figure. While we do not know why the district court discounted this evidence, we cannot review that matter when the record on appeal does not include it.

16 . Standard deviation analysis tests the hypothesis that underrepresentation of a protected minority group in any sample made up of a protected and a nonprotected group (binomial distribution) might be attributable to normal fluctuations of chance rather than to discriminatory design. The standard deviation is the measure of the predictable fluctuation in a random selection process. The difference between actual ("observed”) numbers of the protected group in such a sample and the number that would be "expected” in a perfectly proportional process of selection from the appropriate pool can then be expressed in numbers of standard deviation. In turn, standard deviations can be expressed in terms of the mathematical probability that chance is the cause of the disparities (differences between "observed” numbers and "expected” numbers) measured. See B. Schlei & P. Grossman, supra, at 1372 n. 331.

17 . The comparison of the percentage of women hired by Bowman (1.26%) with the various labor pools resulted in standard deviations ranging from 1.92 to 5.3. 570 F.Supp. at 1516.

18 . Bowman argues that the Supreme Court in Castaneda v. Partida, 430 U.S. 482, 496 n. 17, 97 S.Ct. 1272, 1281, n. 17, 51 L.Ed.2d 498 (1977), and Hazelwood School District, 433 U.S. at 311 n. 17, 97 S.Ct. at 2743, n. 17, established a per se rule that a statistical disparity must be greater than three standard deviations in order to support an inference of discrimination. We disagree.

In discussing whether a prima facie case of discrimination in grand jury selection had been established, the Court in Castaneda noted that "if the difference between the expected value and the observed number is greater than two or three standard deviations, then the hypothesis that the jury drawing was random would be suspect to a social scientist.” 430 U.S. at 496 n. 17, 97 S.Ct. at 1281 n. 17. In Castaneda, the standard deviations ranged from more than twelve to approximately twenty-nine. Id.

In Hazelwood, the Court explained that the selection of the relevant labor market could have a significant effect on the question of whether the plaintiffs had established a pattern or practice of racial discrimination in the school district's employment practices:

[U]nder the statistical methodology explained in Castaneda v. Partida, ... the difference between using 15.4% and 5.7% as the area-wide figure would be significant. If the 15.4% figure is taken as the basis for comparison, ... the difference between the observed number of 15 Negro teachers hired (of a total of 405) would vary from the expected number of 62 by more than six standard deviations. Because a fluctuation of more than two or three standard deviations would undercut the hypothesis that decisions were being made randomly with respect to race, ... each of these statistical comparisons would reinforce rather than rebut the Government's other proof. If, however, the 5.7% areawide figure is used, the expected number of Negro teachers hired in 1972-73 would be ... less than two standard deviations from the observed number of 10; for 1973-74, the expected value would be ... less than one standard deviation from the observed value of 5; and for the two years combined, the expected value of 23 would be less than two standard deviations from the observed total of 15.

433 U.S. at 311 n. 17, 97 S.Ct. at 2743 n. 17. The Court, however, cautioned that "[t]hese observations are not intended to suggest that precise calculations of statistical significance are necessary in employing statistical proof, but merely to highlight the importance of the choice of the relevant labor market area.” Id.

We conclude that Castaneda and Hazelwood do not require a disparity of more than three standard deviations before an inference of discrimination is permissible. See, e.g., Maddox v. Claytor, 764 F.2d 1539, 1552 (11th Cir.1985); Segar v. Smith, 738 F.2d 1249, 1283 n. 28 (D.C. Cir.1984), cert. denied, — U.S.-, 105 S.Ct. 2357, 86 L.Ed.2d 258 (1985); EEOC v. American Nat'l Bank, 652 F.2d 1176, 1191-93 & n. 11 (4th Cir.1981), cert. denied, 459 U.S. 923, 103 S.Ct. 235, 74 L.Ed.2d 186 (1982). First, the huge difference between two and three standard deviations (a level of two standard deviations corresponds to a 5% probability that chance would account for the disparity between the observed and the expected number, and a level of three standard deviations corresponds to approximately a .1% probability that chance would account for the disparity) suggests that the Court did not intend to fix a specific test for statistical significance. Moreover, it was unnecessary for the Court to set a precise lower bound in either Castaneda or Hazelwood because the statistical disparities in Castaneda greatly exceeded two or three standard deviations and those in Hazelwood were less than two.

19 . We express no opinion on the question of whether plaintiffs’ statistics standing alone would have been sufficient to establish a prima facie case of disparate impact.

20 . We have applied the “clearly erroneous” standard to a trial court’s finding of disparate impact. See, e.g., Claytor, 764 F.2d at 1556.

21 . We have held that a district court's finding of pretext may be reversed on appeal only if it is clearly erroneous. See, e.g., Conner v. Fort Gordon Bus Co., 761 F.2d 1495, 1500 (11th Cir. 1985); Fowler v. Blue Bell, Inc., 737 F.2d 1007, 1012 (11th Cir.1984).

22 . We have held that a district court’s finding of a pattern and practice of disparate treatment is reviewable under the "clearly erroneous" standard. See, e.g., Griffin v. Carlin, 755 F.2d 1516, 1526 (11th Cir.1985).

23 . Because we find that the statistical evidence in combination with the anecdotal evidence established a pattern and practice of sex discrimination, we do not consider whether the statistical evidence would have been sufficient by itself to prove a pattern and practice of disparate treatment.

24 . The legitimacy of this reason for not hiring Ms. Hewitt is also undermined by the fact that Bowman drivers generally are able to obtain assistance in unloading their trucks, Record on Appeal, vol. 10 at 127-28, and that the collective bargaining agreement between Bowman and the United Steelworkers requires a driver to be given $38 to pay for the unloading of each truck, id. vol. 12 at 424.

25 . We note that other courts have reached different conclusions on this issue. Compare Scott v. University of Delaware, 601 F.2d 76, 81-82 n. 8 (3d Cir.) (applying federal law), cert. denied, 444 U.S. 931, 100 S.Ct. 275, 62 L.Ed.2d 189 (1979), with Hamilton v. Rogers, 573 F.Supp. 452, 453-54 (S.D.Tex.1983) (applying state law). The survival of a cause of action under the Age Discrimination in Employment Act, 29 U.S.C.A. §§ 621-634, has been held to be governed by federal common law. See, e.g., Fariss v. Lynchburg Foundry, 769 F.2d 958, 962 n. 3 (4th Cir. 1985); Asklar v. Honeywell, Inc., 95 F.R.D. 419, 422-24 (D.Conn.1982).

26 . The Georgia survival statute relied upon in Brazier has been recodified at O.C.G.A. § 9-2-41 (1982), which provides in relevant part:

No action for a tort shall abate by the death of either party, whether wrongdoer received any benefit from the tort complained of; nor shall any action or cause of action for the recovery of damages for homicide, injury to the person, or injury to property abate by the death of either party. The cause of action, in case of the death of the plaintiff and in the event there is no right of survivorship in any other person, shall survive to the personal representative of the deceased plaintiff____

27 . In Stein v. Reynolds Securities, Inc., 667 F.2d 33 (11th Cir.1982), this court adopted as binding precedent all of the post-September 30, 1981, decisions of Unit B of the former Fifth Circuit. Id. at 34.

28 . Bowman also asserts that the post-trial expansion of the disparate impact class violated Fed.R.Civ.P. 23 and N.D.Ga.R. 221.13. These arguments, however, are frivolous. Rule 23(c)(1) simply requires that "[a]s soon as practicable after the commencement of an action brought as a class action, the court shall determine by order whether it is to be so maintained," Fed.R.Civ.P. 23(c)(1), and N.D.Ga.R. 221.13 merely requires a plaintiff to move for class certification within 90 days after the filing of a complaint. Both requirements were satisfied in the instant case.

29 . In the district court, Bowman did not argue with any clarity that it was deterred from offering evidence, other than applicant flow data, which would have been relevant to appropriate system-wide labor pools.

30 . For example, class members applied at Bowman terminals in such geographically dispersed cities as Jacksonville, Florida; Atlanta, Georgia; and Birmingham, Alabama.

31 . Alternatively, Bowman contends that the court’s contempt order is invalid because the court order on which it is based (the January 4, 1984 remedial order) was erroneously issued. This argument, however, is without merit since we have upheld the validity of the district court’s finding of discrimination and entry of a remedial order in the instant appeal. See, e.g., ITT Community Development Corp. v. Barton, 569 F.2d 1351, 1356 (5th Cir.1978).

32 . In their proposed remedial order, plaintiffs requested the district court to require Bowman to have, within five years, an OTR work force that would be no less than 10% female. Record on Appeal, vol. 7 at 1330. Under this proposed order, Bowman could be excused from reaching this goal if it showed "that it has made all reasonable efforts to recruit, retain and hire qualified female OTR drivers and that notwithstanding those affirmative steps, it was not able to satisfy the hiring goals.” Id. at 1331.

33 . Plaintiffs’ claim that the district court simply denied hiring goals because of his personal opposition to such remedies rests on the following language in the court’s opinion:

In this remedial order the Court has not provided hiring goals or quotas____ Preferential treatment on the grounds of sex, by means of a quota, is destructive of self respect and merely substitutes one form of discrimination for another.

576 F.Supp. at 601. This claim, however, ignores the fact that the court expressly stated that it would consider hiring goals if the relief granted were ineffective.

34 . If the relief granted in the instant case proves to be ineffective, plaintiffs may apply to the district court for additional relief.