Cecile Maddow, individually and on behalf of eight opt-in plaintiffs, brought suit against the Procter and Gamble Company and the Noxell Corporation, a subsidiary, for violation of the Age Discrimination in Employment Act, 29 U.S.C. §§ 621-634 (1994). The plaintiffs appeal the district court's 1) grant of summary judgment against them on the claim of discrimination, 2) grant of summary judgment against Maddow on the claim of adverse affect, 3) order compelling discovery and awarding attorney fees for failure to comply with discovery, and 4) denial of joinder of additional plaintiffs.

We have jurisdiction pursuant to 28 U.S.C. § 1291. We reverse in part and affirm in part.

Plaintiffs, each age forty or above, were sales representatives for Max Factor, a cosmetics company. In July 1991, Procter and Gamble acquired Max Factor and assigned it to Noxell Corporation, a wholly-owned subsidiary. Noxell decided to terminate a substantial number of the former Max Factor sales representatives but none of its own original sales force.

Noxell and Max Factor had used different sales strategies. Noxell used a data-based market strategy relying on demographics and other data. Max Factor used little data and was more relationship-based. Noxell generally hired college graduates, while most Max Factor representatives lacked a college education.

In January 1992, Larry Anderson, Noxell's Vice President of Sales, addressed the company's sales personnel. He stated that the image of the new division would be 'new and young.' He also remarked that the Noxell representatives were 'young and energetic' and that the Max Factor personnel were 'old, mature, and well-trained.'

In March 1992, a Noxell district manager interviewed each Max Factor sales representative. Based on the interview, the manager rated each interviewee in seven categories. These ratings were summed to produce a Recruiting Quality Index for each candidate. Noxell depositions and affidavits state that this Index has been reliable in predicting future sales performance.

According to Noxell depositions and affidavits, the Recruiting Quality Index was the sole factor in the selection process. Prior sales experience and performance were not considered. Employment was offered to anyone who scored sixteen or higher. In addition, Noxell decided to retain at least one former Max Factor employee at each level in each geographical sales district. To meet this goal Noxell retained some former Max Factor employees who received scores of fourteen or fifteen. The plaintiffs offered evidence that six employees receiving these scores were retained, only one of whom was over forty, and that all three employees receiving these scores who were not retained were over forty.

Maddow received an Index rating of sixteen. One other plaintiff received a rating of fourteen. The rest were each below thirteen. During Maddow's interview she stated that she would not accept a sales position that was not based in Atlanta, her current location. Maddow's interviewer, who believed she was under forty at the time of the interview, noted that Maddow's likelihood of accepting a job offer as "high."

After Maddow's interview, former Max Factor manager Wade McLelland left a message on her answering machine stating that the company needed to know her age because of an "age factor." She responded that she was forty-one. Noxell then offered Maddow a sales position in Huntsville, Alabama, which she turned down. Two younger Max Factor sales representatives in the same district as Maddow with higher Index scores were offered positions in locations other than their Max Factor districts. None of the plaintiffs besides Maddow was offered a position.

Depositions and affidavits from McLelland and other Noxell employees state that tentative decisions had been made when McLelland called about the 'age factor' and he was getting the information to assess the impact of proposed terminations for federal equal employment opportunity laws and affirmative action regulations. However, Noxell manager Strause stated that he had directed the McLelland inquiry, as the result of a request from another manager or Human Resources, and that he was not given any reason for the inquiry.

One of the plaintiffs was told at the time he was informed of his termination that he did not fit the Procter and Gamble profile and that at his age he should consider retirement.

An affidavit of Mary Nelson, a former Max Factor sales representative who is not a plaintiff, states that she was told by her former Max Factor sales manager that the interviews were a formality and she would not be hired by Noxell because she was over forty. The sales manager attributed this statement to her boss. The former sales manager was not a decision-maker regarding Nelson or any of the plaintiffs, and denies making the statement.

While Noxell was terminating former Max Factor sales representatives, it was hiring new, inexperienced sales representatives. Most were college graduates. Noxell followed a different interview procedure with these recruits than with the former Max Factor employees, part of which was more objective than the Max Factor interviews.

Sixty-eight former Max Factor sales representatives were interviewed by Noxell. Thirty-two of the sixty-eight received offers (not including Maddow and one other sales representative who claim they were constructively discharged). Of those receiving offers, twenty-seven were under age forty and five were over. Of those not receiving offers, fourteen were under forty and twenty-two were over. The plaintiffs' expert testified that the standard deviation for this distribution was 3.83, or a less than a one in one thousand chance of having randomly occurred. His statistical analysis did not account for individuals' Recruiting Quality Index scores. Ninety-one percent of Noxell's original 112 person sales force was under age forty.

The district court granted summary judgment for the defendants, holding that the plaintiffs had not raised a genuine issue of material fact that the defendants' use of Recruiting Quality Index scores was a pretext for age discrimination. The court also held that Maddow had not been adversely affected because she had been offered the position in Alabama.

During discovery, the defendants requested plaintiffs' tax returns and sent an interrogatory seeking detailed attorney fee arrangement information. Plaintiffs objected to the tax return request on privilege and relevance grounds, and instead provided alternative documentation of earnings such as W-2 and 1099 forms. In response to the interrogatory, the plaintiffs stated that the fee agreement was a contingency and the individual plaintiffs were responsible for costs, but did not disclose the precise contingency percentage.

The district court granted the defendants' motion to compel discovery of these two items, and granted $3,000 in attorney's fees to the defendants.

On March 2, 1994 the parties filed a stipulated Scheduling Order, which was approved by the district court on March 14. One paragraph of the Order allowed individuals similarly situated to the plaintiff to file opt-in notices within sixty days. On March 18, defendants informed plaintiffs that information on Max Factor employees could be obtained from defendants' personnel files. Plaintiffs did not obtain the necessary information until late May and early June due to their own scheduling difficulties. Plaintiffs did not seek a modification of the Scheduling Order. On July 5, 1994; August 16, 1994; and August 26, 1994 plaintiffs filed motions to join a total of eight additional persons. The August 16 motion included a motion to modify the Scheduling Order to extend the joinder deadline. The district court held that these individuals were similarly situated to plaintiffs, but denied their motions to join as untimely.

We review a district court's grant of summary judgment de novo. Jameson v. Arrow Co., 75 F.3d 1528, 1531 (11th Cir.1996). In analyzing a motion for summary judgment, we must view the facts in the light most favorable to the nonmoving party and make all factual inferences in favor of that party. Hairston v. Gainesville Sun Publishing Co., 9 F.3d 913, 918 (11th Cir.1993).

"In an employment discrimination case, the plaintiff must produce sufficient evidence to support an inference that the defendant employer based its employment decision on an illegal criterion." Jameson, 75 F.3d at 1531. We "generally ... eschew[ ] an overly strict formulation of the elements of a prima facie case, particularly in age discrimination cases." Id.

In age discrimination cases, once a plaintiff has made a prima facie case, the employer may then rebut the inference of discrimination by providing legitimate, non-discriminatory reasons for its decision. The burden than shifts back to the plaintiff to raise a genuine factual question as to whether defendant's stated reason is mere pretext. Hairston, 9 F.3d at 920. However, "The grant of summary judgment, though appropriate when evidence of discriminatory intent is totally lacking, is generally unsuitable ... [where] plaintiff has established a prima facie case because of the 'elusive factual question' of intentional discrimination." Id. at 921 (citing Texas Dep't of Community Affairs v. Burdine, 450 U.S. 248, 256, 101 S. Ct. 1089, 1095, 67 L. Ed. 2d 207 (1981)).

A plaintiff may establish a prima facie case of age discrimination in a reduction-in-force case by (1) demonstrating that he was in a protected age group and was adversely affected by an employment decision; (2) showing he was qualified for his former position or another position at the time he was adversely affected; and (3) producing circumstantial or direct evidence from which a reasonable factfinder could conclude that his employer intended to discriminate on the basis of age in reaching the decision at issue. Verbraeken v. Westinghouse Elec. Corp., 881 F.2d 1041, 1045-46 (11th Cir.1989). Plaintiffs were all in a protected age group, were adversely affected (Maddow is discussed in part II, infra), and were qualified. Therefore, the question is whether the plaintiffs produced sufficient evidence of age discrimination.

Evidence submitted by the plaintiffs satisfies their burden of establishing a prima facie case of age discrimination. The defendants rebutted the prima facie inference with the alleged legitimate, non-discriminatory reason of basing termination decisions on the Recruiting Quality Index. The plaintiffs thus have the burden of establishing that defendants' proffered reason is a pretext.

Because this is a summary judgment motion, the plaintiffs need only raise a genuine issue of material fact that the reason was a pretext; they do not need to actually prove it. Hairston, 9 F.3d at 921. The plaintiffs' evidence clears this hurdle. Any reading of prior case dicta to the effect that plaintiffs need to prove pretext at the summary judgment stage is incorrect.

Defendants' contention that the evidence used by the plaintiffs to establish the prima facie case may not be used to raise a genuine issue as to pretext, and their implication that a genuine issue as to pretext cannot be raised circumstantially, are unsupportable. Evidence offered in the prima facie case may be sufficient to raise a genuine issue of material fact regarding pretext. Id.

The statistical evidence plaintiffs offered is sufficient for a reasonable person to infer discrimination based on age and that the reasons given were pretextual. The plaintiffs' expert testified that the standard deviation for the distribution of terminations based on age was 3.83. This distribution has a less than one in one thousand chance of occurring randomly. A standard deviation of two or three is enough to support an inference of discrimination. Kilgo v. Bowman Transp., Inc., 789 F.2d 859, 871 n. 18 (11th Cir.1986) (citing Castaneda v. Partida, 430 U.S. 482, 496 n. 17, 97 S. Ct. 1272, 1281 n. 17, 51 L. Ed. 2d 498 (1977)). The defendants failed to rebut this evidence or the expert. Two reasonable conclusions can be drawn from statistical evidence this strong: the defendants intentionally discriminated or the Recruiting Quality Index system unintentionally favored younger people. Despite defendants' plausible claim that there was no discrimination, the former is a reasonable inference, and satisfies plaintiffs' burden at this stage.

In addition to the statistical evidence, there is also circumstantial evidence of age discrimination. This includes the Vice-President of Sales' statement that the new division would have a 'new and young' image and that the original Noxell employees were 'young' while the former Max Factor employees were 'old and mature'; the phone message for Maddow mentioning an 'age factor'; the interviewer's listing Maddow as highly likely to accept an offer and then offering her a position that she had already stated she would turn down; the employment offers to some representatives who scored only fourteen or fifteen; the statement to one plaintiff that he should consider retiring at his age; Noxell's use of an interview procedure for former Max Factor employees that was different from the usual procedure; and Noxell's decision to retain its predominantly young original sale representatives. Taken alone, any one of these may not be enough. However, as a whole, a reasonable person could infer discrimination.

Finally, Mary Nelson's statement that she was told Noxell was only hiring representatives under forty and the interviews were only a formality is direct evidence of discrimination. Viewing this evidence in the light most favorable to the plaintiff, it creates a genuine issue of material fact. The statement may not be credible, but that is a decision for the factfinder, not for summary judgment.

The defendants' explanation for their termination decisions is entirely plausible. However, the statistical, circumstantial, and direct evidence in the record, when viewed in the light most favorable to the plaintiff, creates a genuine issue of material fact that defendants proffered reason is a pretext and that there was discrimination. Summary judgment against the plaintiffs on the issue of discrimination must therefore be reversed.

To make a prima facie case of age discrimination, an employee must show that she was adversely affected by an employment decision. Verbraeken, 881 F.2d at 1045-46. The district court held that Maddow was not adversely affected because she was offered the position in Alabama, and thus was not constructively discharged.

However, requiring a transfer can be a constructive discharge. Stamey v. Southern Bell Telephone & Telegraph Co., 859 F.2d 855, 860 n. 11 (11th Cir.1988). Further, constructive discharge is not the only possible adverse affect in employment cases. Id. at 860. Under the Age Discrimination in Employment Act, it is unlawful for an employer "to fail or refuse to hire or to discharge any individual or otherwise to discriminate against any individual with respect to his compensation, terms, conditions, or privileges of employment because of such individual's age." 29 U.S.C. § 623(a) (emphasis added).

Maddow presented evidence that when her interviewer believed she was under forty he gave her a Recruiting Quality Index score of sixteen. The interviewer noted that she had a high likelihood of accepting an offer after being told explicitly that she would only accept an offer in Atlanta. Maddow was then left a message asking her age. After Maddow told Noxell she was forty-one, she was informed that the only position she was being offered was in Alabama.

Maddow has created a reasonable inference that the terms and conditions of her employment were affected because of her age. This is an adverse affect. Stamey, 859 F.2d at 860. It is inapposite that a job Maddow has since accepted involves travel to Alabama. Summary judgment against Maddow on the issue of adverse affect is reversed.

We review district court rulings on discovery motions and discovery sanctions for abuse of discretion. Harris v. Chapman, 97 F.3d 499, 506 (11th Cir.1996).

Plaintiffs initially did not comply with defendants' discovery requests for income tax forms and details of their attorney fee arrangement. Plaintiffs supplied defendants with W-2 and 1099 forms as evidence of earnings, but argued that the tax records would reveal irrelevant and qualifiedly privileged information. Plaintiffs also informed defendants that the attorney fee arrangement was contingent and plaintiffs were responsible for costs, but did not supply the contingency percentage on the grounds that it was irrelevant.

The district court held that the tax records would reveal income information not contained in W-2 forms and that the information would be protected by the Protective Order entered into by the parties. The court held that the contingency percentage was relevant because plaintiffs sought attorney's fees, and the percentage would be relevant to calculating reasonable attorney's fees. The court therefore compelled discovery of these two items, and sanctioned plaintiffs $3,000 in attorney's fees for their initial failure to comply.

The court's decision to compel discovery was not an abuse of discretion: both items of information are arguably relevant to the case. The award of sanctions, however, was.

A court must impose attorney's fees and expenses when compelling discovery unless the party was substantially justified in resisting discovery. Fed.R.Civ.P. 37(a)(4)(A). Substantially justified means that reasonable people could differ as to the appropriateness of the contested action. Pierce v. Underwood, 487 U.S. 552, 565, 108 S. Ct. 2541, 2550, 101 L. Ed. 2d 490 (1988).

Here, the plaintiffs were substantially justified in relying on Supreme Court dictum regarding the attorney's fees issue, and relying on out-of-circuit district court caselaw, where there was no in-circuit caselaw, regarding the tax form issue. See Blanchard v. Bergeron, 489 U.S. 87, 93, 109 S. Ct. 939, 944, 103 L. Ed. 2d 67 (1989) (evidence of specific details of contingency fee arrangement is not necessary for a court to calculate reasonable attorney's fees; it is at most one factor); City of Burlington v. Dague, 505 U.S. 557, 565, 112 S. Ct. 2638, 2643, 120 L. Ed. 2d 449 (1992) (contingency fee is not grounds for a fee enhancement); Lemanik v. McKinley Allsopp, Inc., 125 F.R.D. 602, 609 (S.D.N.Y.1989) (there is a public policy of confidentiality of tax returns; to require discovery, a party must establish relevancy and the court must find a compelling need for the returns because the information is not otherwise obtainable); Biliske v. American Live Stock Inc., 73 F.R.D. 124, 126 n. 1 (W.D.Okla.1977) (public policy against unnecessary disclosure of tax returns).

The plaintiffs were substantially justified in initially refusing discovery. The award of attorney's fees was an abuse of discretion. We affirm the district court's decision to compel discovery but reverse its sanction of attorney's fees.

IV. Joinder or Intervention of Additional Opt-in Plaintiffs

We review the denial of a motion to join additional plaintiffs for abuse of discretion. See Nolin v. Douglas County, 903 F.2d 1546 (11th Cir.1990), overruled on other grounds, McKinney v. Pate, 20 F.3d 1550 (11th Cir.1994) (ruling on motion to amend pretrial order is reviewed for abuse of discretion). We review the denial of a motion for permissive intervention for abuse of discretion. United States v. Dallas County Comm'n, 850 F.2d 1433, 1443 (11th Cir.1988).

The district court has authority to manage the process of joining additional parties. Hoffmann-La Roche, Inc. v. Sperling, 493 U.S. 165, 170-71, 110 S. Ct. 482, 486-87, 107 L. Ed. 2d 480 (1989). Federal Rule of Civil Procedure 16(b) requires a district court judge to enter a scheduling order that limits the time to join additional parties. Here, the parties entered a Scheduling Order on March 2, 1994 that stated all parties would be joined within sixty days. The additional plaintiffs did not file their motion to join until more than two months after this deadline. No motion to extend the time in Scheduling Order was filed until even later.

In their reply brief to the district court on the motion to join additional opt-in plaintiffs, the plaintiffs apparently tried to convert it into a motion to intervene under Federal Rule of Civil Procedure 24. However, no official intervention motion was ever filed. Further, whether a motion to intervene is timely is within the district court's discretion. Stallworth v. Monsanto Co., 558 F.2d 257, 263 (5th Cir.1977). Stallworth requires the district court to explicitly rule on four different factors to deny a motion to intervene. However, this is not required where it is not even clear that the motion was one to intervene.

The district court did not abuse its discretion in denying the additional opt-in plaintiffs' motion to join.

We reverse the district court's summary judgment order against Maddow and the opt-in plaintiffs on both discrimination and adverse affect grounds; we affirm the district court's order compelling discovery, but reverse its sanction of attorney's fees against plaintiffs; we affirm the district court's order denying joinder or intervention of the additional opt-in plaintiffs.

REVERSED in part and AFFIRMED in part.