ORDER
Appellees’ petition for rehearing is granted. Our opinion of July 3, 2001 is withdrawn and a revised opinion is filed with this order.
The petition for rehearing en banc was transmitted to all of the judges of the court who are in regular active service as required by Fed. R.App. P. 35. As no member of the panel and no judge in regular active service on the court requested that the court be polled, the petition for rehearing en banc is denied.
OPINION
BRISCOE, Circuit Judge.
Plaintiff Gary Thiessen, an employee of defendants General Electric Capital Corporation (GE) and Montgomery Ward Credit Services, Inc., filed this putative class action under the Age Discrimination in Employment Act (ADEA) alleging that he and other similarly situated employees had been adversely affected by a pattern or practice of age discrimination on the part of defendants. Although the district court initially certified a class of twenty-three plaintiffs, it ultimately decertified the class, dismissed the opt-in plaintiffs, and granted summary judgment in favor of defendants with respect to Thiessen’s individual claims. Thiessen now appeals claiming the district court erred in (1) decertifying the class and dismissing the opt-in plaintiffs, (2) granting - summary judgment in favor of defendants with respect to his individual claims of age discrimination, (3) refusing to allow eight proposed plaintiffs to join the class, and (4) refusing to allow him to depose defendants’ corporate counsel. We exercise jurisdiction pursuant to 28 U.S.C. § 1291, and reverse and remand for further proceedings.
I.
GE is the parent company of General Electric Capital Services (GECS), which in turn owns General Electric Capital Credit (GECC). Within GECC is the Retail Financial Services unit (RFS). In 1989, GECC purchased Monogram Retail Credit Services, Inc. (MRCSI), which had been the credit division of retailer Montgomery Ward. MRCSI became a subsidiary of the RFS unit after the acquisition, and in 1997, it was renamed Montgomery Ward Credit Services, Inc.1
Thiessen, who was born on March 7, 1947, began working as a credit manager trainee for MRCSI in Í968. From 1972 to mid-1994, he held various management positions within MRCSI and progressed to Band 4 on the company’s five-band pay-grade scale. Between May 1994 and August 1996, Thiessen was placed on “special assignments” and assisted in the construction of MRCSI facilities in Kansas, Illinois, and Georgia. At the completion of those projects, Thiessen transferred to MRCSI’s Las Vegas facility and assumed his current position as Band 4 collection manager. According to statements in the record, [1100]*1100Thiessen’s position at the Las Vegas facility was to “be eliminated as of May 30, 1998,” and Thiessen allegedly was not “allowed to post for any other positions within GE.” Aplt.App. 3 at 2.
On February 5, 1996, Thiessen filed a charge of discrimination with the Kansas Human Rights Commission and the Equal Employment Opportunity Commission (EEOC). He alleged that defendants had “an express but covert ‘White Blockers’ or ‘Blockers’ policy of discriminating against older white employees ... by forcing [them] into early retirement or by eliminating [their] ... positions] through restructuring.” Id. at Tab C. He further alleged that defendants exercised this “pattern and practice of employment decisions motivated by age and/or race and related factors” against him. Id. Specifically, Thiessen alleged he was repeatedly denied promotions because of his age and race. He subsequently filed this action alleging, in pertinent part, that defendants committed willful violations of the ADEA by denying him various promotions and placing him on “special assignments.”2 Appee. Supp.App. II at 003. Thiessen further alleged in his complaint that defendants undertook a pattern or practice of systematically discriminating against older employees.
The centerpiece of both Thiessen’s individual claims of age discrimination and his claims of class-wide discrimination was a “blocker policy” allegedly adopted and implemented by defendants in the early 1990s. According to Thiessen, in 1991, “GECC management began referring to the older [executive] employees as ‘blockers,’ because in their view these employees were ‘blocking’ the advancement of younger, newly recruited employees,” in particular those younger employees who were participants in defendants’ Management Development Program (MDP). Id. at 004. On March 2, 1992, Dave Ekedahl, a vice-president at RFS, and David Ferreira, vice-president of human resources for GECC, sent a memorandum to Steve Joyce, the then president and CEO of MRCSI, and Jeff Faucette, vice-president of human resources for MRCSI, asking, “What are plans to upgrade executive talent ... remove blockers.” Appee. Supp. App. I at 119. In response to this memo, Faucette directed his subordinates to prepare severance worksheets and retirement packages for various MRCSI employees over the age of forty, including Thiessen and several of the opt-in plaintiffs (Lawson, Gwen Colwell, Bob Demartine, Melva Heid, and Bob Marsonette) none of whom had requested such packages. On March 1, 1993, Ferreira circulated a memo to Faucette and others again discussing the subject of “blockers.” Joyce and Faucette allegedly gave a presentation discussing Band 4 Blockers at a 1993 leadership review meeting attended by Ferreira. At a May 1993 meeting of MRCSI human resource managers, the concept of “blockers” was again discussed and outlined by Fau-cette. At a June 18, 1993, meeting at MRCSI’s Merriam, Kansas facility, Fau-cette distributed a list of possible blockers to the human resource managers. The managers were effectively instructed to discuss the blocker policy with their respective operational managers, and assist the operational managers in identifying blockers and carrying out the policy. Ultimately, Thiessen alleges, the operational managers implemented the “blocker policy” by taking varioüs negative employment actions (e.g., negative performance evaluations, demotions, terminations, reductions in force, placement on “special projects,” [1101]*1101etc.) towdrd older employees identified as blockers.
Thiessen alleged defendants’ “blocker policy” impacted all of the opt-in plaintiffs, each of whom allegedly suffered some type of adverse employment action. On an individual level, Thiessen alleged he was identified by defendants as a blocker, and he was (1) denied a business center manager position in Las Vegas in 1993, (2) phased out of his position as National Attorney-Agency manager in 1994 and placed on “special assignment,” (3) downgraded on his 1993 and 1994 performance reviews, and (4) denied two remittance processing manager positions in 1995.
Defendants denied having any pattern or practice of discrimination. Although they acknowledged there had been some discussion among certain employees of blockers, they denied that any type of “blocker policy” was implemented anywhere in the organization. Further, defendants alleged that in the fall of 1994, Gail Lanik, the CEO of MRCSI and successor to Steve Joyce expressly repudiated the notion that employment decisions were based on any “blocker policy.” In response to defendants’ denials, Thiessen alleged that, notwithstanding Lanik’s purported repudiation, the “blocker policy” continued to be used on a covert basis. In support of his allegation, Thiessen pointed to alleged acts of employment discrimination against him and the other proposed plaintiffs after the fall of 1994. Thiessen also pointed to statistical evidence that allegedly demonstrated older workers were treated less favorably than younger employees.
During the course of discovery, the district court conditionally certified a class of twenty-three plaintiffs (Thiessen and twenty-two opt-in plaintiffs). Thiessen v. General Elec. Capital Corp., 996 F.Supp. 1071, 1083 (D.Kan.1998) (Thiessen I). In doing so, the district court noted:
Mr. Thiessen has not simply averred the existence of a discriminatory policy or merely argued that circumstantially one should infer its existence by virtue of numerous people in the protected category having incurred adverse employment actions. Instead, he has come for-' ward with direct evidence of an overall policy of purported age discrimination. Moreover, the names of some of the opt-in plaintiffs appear on “blocker” lists. The court believes this evidence cannot be so easily discounted just because the alleged discrimination was meted out “over a wide geographic range in a number of ways” and that it merits a threshold determination that the opt-in plaintiffs are “similarly situated” to Mr. Thiessen for purposes of something akin to conditional certification. However, the court remains troubled by a puzzling lack of any significant showing by Mr. Thiessen that there is a link between this alleged policy and the adverse job actions. Even if there was a discriminatory policy, the opt-in plaintiffs are only “similarly situated” if they can make a submissible case that they, too, were victims of it. It is because of that deficiency, at least in part, that the court is unwilling to grant final certification on the record before it.
Id. The court cautioned Thiessen that, “[i]n order to survive a motion to decertify,” he would “need to set forth what he deem[ed] to be the specific link between the blocker policy and what occurred with each opt-in plaintiff.” Id. Further, the court noted that if, in a subsequent motion to decertify, defendants could “truly convince the court that individual [defense] issues would predominate at trial, the contention that the opt-in plaintiffs [we]re similarly situated would likely be largely eviscerated.” Id. at 1084. Finally, the [1102]*1102court noted it would make a final determination regarding whether the trial of all the opt-in plaintiffs’ claims could be coherently managed and presented “in a manner that w[ould] not confuse the jury or unduly prejudice any party.” Id.
At the conclusion of discovery, defendants moved to decertify the class. Defendants also moved for summary judgment with respect to each of the twenty-three plaintiffs. The district court decertified the class, dismissed the opt-in plaintiffs’ claims -without prejudice, and granted summary judgment in favor of defendants with respect to Thiessen’s individual claims of discrimination. Thiessen v. General Elec. Capital Corp., 13 F.Supp.2d 1131 (D.Kan.1998) (Thiessen II).
II.
Decertification of plaintiff class and dismissal of opt-in plaintiffs
Thiessen contends the district court erred in decertifying the plaintiff class and dismissing the opt-in plaintiffs’ claims without prejudice. We review for abuse of discretion a district court’s decision to certify or decertify a class under the ADEA. See Mooney v. Aramco Servs. Co., 54 F.3d 1207, 1213 (5th Cir.1995); Lockhart v. Westinghouse Credit Corp., 879 F.2d 43, 51-52 (3d Cir.1989). Under an abuse of discretion standard, we will not disturb the underlying decision unless we have a definite and firm conviction that the district court made a clear error of judgment or exceeded the bounds of permissible choice in the circumstances. See Richardson v. Missouri Pacific R.R. Co., 186 F.3d 1273, 1276 (10th Cir.1999).
Class actions under the ADEA are authorized by 29 U.S.C. § 626(b), which expressly borrows the opt-in class action mechanism of the Fair Labor Standards Act of 1938, 29 U.S.C. § 216(b) (1994). Section 216(b) provides for a class action3 where the complaining employees are “similarly situated.” Unlike class actions under Rule 23, “[n]o employee shall be a party plaintiff to any such action unless he gives his consent in writing to become such a party and such consent is filed in the court in which such action is brought.” Id.
The overriding question here is whether Thiessen and the twenty-two opt-in plaintiffs are “similarly situated” for purposes of § 216(b). Unfortunately, § 216(b) does not define the term “similarly situated,” and there is little circuit law on the subject. Federal district courts have adopted or discussed at least three approaches to determining whether plaintiffs are “similarly situated” for purposes of § 216(b). See, e.g., Mooney, 54 F.3d at 1213 (discussing two different approaches adopted by district courts); Bayles v. American Med. Response of Colo., Inc., 950 F.Supp. 1053, 1058 (D.Colo.1996). Under the first approach, a court determines, on an ad hoc case-by-case basis, whether plaintiffs are “similarly situated.” Mooney, 54 F.3d at 1213. In utilizing this approach, a court typically makes an initial “notice stage” determination of whether plaintiffs are “similarly situated.” Vaszlavik v. Storage Tech. Corp., 175 F.R.D. 672, 678 (D.Colo.1997). In doing so, a court “ ‘require[s] nothing more than substantial allegations that the putative class members were together the victims of a single decision, policy, or plan.’ ” Id. (quoting Bayles, 950 F.Supp. at 1066). At the conclusion of discovery (often prompted by a motion to [1103]*1103decertify), the court then makes a second determination, utilizing a stricter standard of “similarly situated.” Id. at 678. During this “second stage” analysis, a court reviews several factors, including “(1) disparate factual and employment settings of the individual plaintiffs; (2) the various defenses available to defendant which appear to be individual to each plaintiff; (3) fairness and procedural considerations; and (4) whether plaintiffs made the filings required by the ADEA before instituting suit.” Id.4
Under the second approach, district courts have incorporated into § 216(b) the requirements of current Federal Rule of Civil Procedure 23. See Bayles, 950 F.Supp. at 1060-61 (discussing cases adopting this approach). For example, in Shushan v. University of Colo., 132 F.R.D. 263 (D.Colo.1990), the district court concluded “that Rule 23(a)’s four prerequisites [numerosity, commonality, typicality, and adequacy of representation] and 23(b)(3)’s requirement that common questions of fact predominate should be used to determine whether plaintiffs are similarly situated.” Bayles, 950 F.Supp. at 1061 (describing Shushan holding).
In a third approach, district courts have suggested incorporating into § 216(b) the requirements of the pre 1966 version of Rule 23, which allowed for “spurious” class actions. See Bayles, 950 F.Supp. at 1064 (holding that, because Advisory Committee Notes to the 1966 amendments make clear that the amendments were not intended to affect § 216(b) actions, “if Rule 23’s standards apply ... at all, it must be through Rule 23 as it existed prior to 1966”). Under the pre 1966 version of Rule 23, the “character of the right sought to be enforced for” the class of plaintiffs must be “several,” there must be “a common question of law or fact affecting the several rights,” and “a common relief’ must be sought. Id. Further, under the old Rule 23, district courts “had inherent authority to refuse to proceed collectively where it would waste judicial resources or unfairly prejudice the party opposing the proposed class.” Id. at 1065.
In the present case, the district court adopted the ad hoc approach to determining whether the plaintiffs were “similarly situated” for purposes of § 216(b). It first conditionally certified the class during the course of discovery, applying a fairly lenient standard for what constituted “similarly situated.” During its second-stage analysis (prompted by the defendants’ motion to decertify), the court applied a stricter standard, focusing on three factors: “(1) whether a sufficient link existed between the alleged blocker policy and the challenged employment decisions; (2) whether individual issues would predominate at trial; and (3) whether a trial of the action could be coherently managed and evidence presented in a manner that would not confuse the jury or unduly prejudice any party.” Thiessen II, 13 F.Supp.2d at 1136. Ultimately, the district court concluded that all three factors warranted decertifying the class.
With respect to the first factor, the district court concluded “the mere fact that plaintiffs suffered adverse employment actions after having been designated as ‘blockers,’ standing alone, [wa]s insufficient to support an inference that a causal connection exist[ed] between the blocker policy and the adverse actions, particularly in light of the significant time lag and critical intervening events.” Id. at 1140. . The [1104]*1104“significant time lag” was the time period between the plaintiffs’ alleged designations as “blockers” in 1993 and the filing of Thiessen’s EEOC charge in February-1996, which the district court concluded allowed Thiessen and the other opt-in plaintiffs to reach back 300 days from the charge, or until approximately April 11, 1996. The “critical intervening events” included the express repudiation of the “blocker policy” by MRCSI CEO Gail Lan-ik in the fall of 1994.5 Because there was insufficient evidence of a “causal link between the blocker policy or plaintiffs’ purported blocker status and the challenged employment actions,” the district court stated it could not “conclude that the plaintiffs [we]re similarly situated for purposes of proceeding as a collective action.” Id. at 1141.
In conducting its analysis of the first factor, the district court also rejected the suggestion forwarded by Thiessen that it should apply a summary judgment-type standard in determining whether, for purposes of the “similarly situated” analysis, there was a sufficient link between the blocker policy and the adverse employment actions taken against, plaintiffs. Instead, the court concluded that whether plaintiffs were “similarly situated” was a factual determination it was entitled to make based on the evidence presented by the parties.
With respect to the second factor, the district court noted that defendants had come forward with “highly individualized” defenses with regard to each of the various adverse employment actions alleged by plaintiffs. Id. at 1142. Allowing defendants to assert these defenses during a collective trial, the court concluded, would effectively result in “23 individual jury trials to determine defendants’ liability to each plaintiff.” Id. at 1143. Although the court acknowledged the “common thread” of the blocker policy, it nevertheless concluded that “the judicial inefficiencies” associated with resolving “what appear[ed] to be 23 distinct cases” clearly outweighed “any potential benefits in proceeding as a collective action.” Id.
With respect to the third factor, the district court noted that plaintiffs had proposed separating the trial into two phases. Plaintiffs proposed that during the first phase of the trial the court “would submit the pattern and practice allegations as well as the individual claims and damages of nine plaintiffs [seven of whom appeared on various lists of proposed blockers, and two of whom were told they were blockers].” Id. According to plaintiffs, this initial phase would “involve a determination of common class issues relating to the existence, implementation, and discriminatory nature of the Blocker Policy as well as class issues regarding wilfulness under the [1105]*1105ADEA.” Id. “If the jury determined that the blocker policy existed and that defendants had engaged in a pattern and practice of age discrimination, according to plaintiffs’ proposed plan, then the same jury would ... hear Phase Two of the trial, which would involve the remaining 14 plaintiffs.” Id. at 1143-44.
The district court concluded that plaintiffs’ proposed trial plan had “numerous, serious deficiencies.” Id. at 1144. In particular, the court concluded that plaintiffs’ plan “renderfed] individualized consideration of the claims impossible and impose[d] extraordinary burdens on the jury, both in terms of the quantity of evidence and the length of trial.” Id. Thus, the court stated that “the absence of any workable trial management plan ... reinforce[d]” its decision to decertify the class. Id.
The initial question, which we address de novo, is whether it was proper for the district court to adopt the ad hoc approach in determining whether plaintiffs were “similarly situated” for purposes of § 216(b). See Mooney, 54 F.3d at 1213 (applying a de novo standard of review in determining whether district court applied the proper legal standard in decertifying a class). Arguably, the ad hoc approach is the best of the three approaches outlined because it is not tied to the Rule 23 standards. Congress clearly chose not to have the Rule 23 standards apply to class actions under the ADEA, and instead adopted the “similarly situated” standard. To now interpret this “similarly situated” standard by simply incorporating the requirements of Rule 23 (either the current version or the pre 1966 version) would effectively ignore Congress’ directive.6 That said, however, there is little difference in the various approaches. All approaches allow for consideration of the same or similar factors, and generally provide a district court with discretion to deny certification for trial management reasons. We find no error on the part of the district court in adopting the ad hoc approach.
Turning to the merits of Thiessen’s arguments, the question is whether the district court abused its discretion in applying the ad hoc approach and decertifying the class. If plaintiffs were simply attempting to collectively assert their individual claims of discrimination, the decision to decertify would appear to be entirely proper. The problem, however, is that plaintiffs were asserting a pattern-or-practice claim modeled on International Brotherhood of Teamsters v. United States, 431 U.S. 324, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977). In particular, plaintiffs alleged that defendants, by way of the MDP and the blocker policy, had a company-wide policy of age discrimination. Stated differently, plaintiffs alleged that it was defendants’ standard operating procedure, implemented through the MDP and the blocker policy, to discriminate against employees (particularly executive employees) on the basis of age. Although the district court acknowledged that all of the plaintiffs relied on the existence of the “blocker policy” to support their claims, it failed to recognize the pattern-or-practice nature of the class claim. For the reasons discussed below, we conclude the district court’s failure in this regard adversely impacted its “similarly situated” analysis and resulted in an abuse of discretion.
[1106]*1106Pattern-or-practice cases differ significantly from the far more common cases involving one or more claims of individualized discrimination. In a case involving individual claims of discrimination, the focus is on the reason(s) for the particular employment decisions at issue. Cooper v. Fed. Reserve Bank of Richmond, 467 U.S. 867, 876, 104 S.Ct. 2794, 81 L.Ed.2d 718 (1984). In contrast, the initial focus in a pattern-or-practice case is not on individual employment decisions, “but on a pattern of discriminatory decisionmaking.” Id. Thus, the order and allocation of proof, as well as the overall nature of the trial proceedings, in a pattern-or-practice case differ dramatically from a case involving only individual claims of discrimination. Teamsters, 431 U.S. at 357-62, 97 S.Ct. 1843.
Pattern-or-practice cases are typically tried in two or more stages. During the first stage of trial, the plaintiffs’ “burden is to demonstrate that unlawful discrimination has been a regular procedure or policy followed by an employer or group of employers.” Id. at 360, 97 S.Ct. 1843. Thus, “[a]t the initial, ‘liability’ stage of a pattern-or-practice suit the [plaintiffs are] not required to offer evidence that each person for whom [they] will ultimately seek relief was a victim of the employer’s discriminatory policy.” Id. Instead, plaintiffs’ “burden is to establish that such a policy existed.” Id. “The burden then shifts to the employer to defeat the prima facie showing of a pattern or practice by demonstrating that the [plaintiffs’] proof is either inaccurate or insignificant.” Id. “If an employer fails to rebut the inference that arises from the [plaintiffs’] prima facie case,” the finder of fact can conclude “that a violation has occurred” and the trial court can award prospective equitable relief. Id. at 361, 97 S.Ct. 1843. If the plaintiffs also seek “individual relief for the victims of the discriminatory practice,” the case moves into the second or subsequent stages. Id. In these additional proceedings,7 it must be determined whether each individual plaintiff was a victim of the discriminatory practice. Importantly, by having prevailed in the first stage of trial, the individual plaintiffs reap a significant advantage for purposes of the second stage: they are entitled to a presumption that the employer had discriminated against them. See Cooper, 467 U.S. at 875, 104 S.Ct. 2794.8
Returning to the district court’s decision to decertify, we conclude the district court erred in its determination of “whether a sufficient link existed between the alleged blocker policy and the challenged employment decisions,” Thiessen II, 13 F.Supp.2d at 1136. Given the pattern-or-practice nature of plaintiffs’ claim, this factor necessarily encompasses factual issues relevant to both the first and second stages of trial, e.g., whether the blocker policy continued after Lanik’s alleged repudiation and, if it did, whether a link existed between that policy and the individual employment decisions affecting the named plaintiffs. The problem is that the district court effectively made findings regarding these issues in [1107]*1107the guise of determining whether plaintiffs were “similarly situated.” See Adamson v. Bowen, 855 F.2d 668, 676 (10th Cir.1988) (holding that district court, in making class certification decision, should avoid focusing on merits of underlying class action); Anderson v. City of Albuquerque, 690 F.2d 796, 799 (10th Cir.1982) (reversing and remanding denial of class certification where district court indicated its belief that plaintiff could not prevail on individual claims). By doing so, the district court essentially deprived plaintiffs of their right to have the issues decided by a jury, or to at least have the court determine, under summary judgment standards, whether there was sufficient evidence to send the issue to the jury. Further, the district court failed to take into account the fact that, if plaintiffs were able to establish a pattern or practice of discrimination, they would be entitled to a presumption that the individual employment actions taken against them were the result of such discrimination. Indeed, the district court effectively deprived plaintiffs of this procedural advantage as well.
The district court’s consideration of the second factor (whether individual issues would predominate at trial), was likewise adversely affected by its failure to recognize that plaintiffs were proceeding under a pattern-or-practice theory. Although it is true that defendants asserted “highly individualized” defenses to each of the instances of individual discrimination asserted by plaintiffs, those defenses would not become the focal point until the second stage of trial and could be dealt with in a series of individual trials, if necessary. With respect to the first stage of trial and the initial issues of whether they had in place a pattern or practice of discrimination and whether it continued after the fall of 1994, defendants had only a few common defenses. Specifically, defendants alleged that the blocker policy, though perhaps mentioned by Faucette at various times, was never implemented anywhere in the organization and was expressly repudiated by Lanik. Further, with respect to the discriminatory nature of the MDP program, defendants claimed it was open to all employees, including those over age forty. Thus, the presence of the “highly individualized” defenses clearly did not, as the district court concluded, outweigh “any potential benefits in proceeding as a collective action.” As previously noted, there was a significant procedural advantage for plaintiffs to proceed in a collective action: if they prevailed in the first stage of trial, they would be entitled to a presumption of discrimination in subsequent proceedings to decide the merits of their individual claims. By bowing to the individualized defenses relevant only to the second stage of trial, the district court deprived plaintiffs of this opportunity-
The district court’s consideration of the third factor (trial management concerns) was also adversely affected by its failure to recognize the pattern-or-practice nature of plaintiffs’ claim. Most notably, the district court failed to acknowledge that plaintiffs’ proposed trial plan, though perhaps deficient in some respects, was consistent with the framework outlined in Teamsters for pattern-or-practice claims. The court also failed to make any effort to modify the. plaintiffs’ proposed trial plan. Finally, the district court was wrong in concluding that trying the case in two phases, as suggested by plaintiffs, “rendered] individualized consideration of the claims impossible.” Thiessen II, 13 F.Supp.2d at 1143.
We conclude the district court abused its discretion in' decertifying the class. Taking into consideration the “pattern or practice” nature of plaintiffs’ lawsuit, as 'discussed above, plaintiffs were, in fact, “similarly situated” for purposes of [1108]*1108§ 216(b). In particular, many of the plaintiffs were specifically listed or classified by defendants as “blockers” and suffered adverse employment consequences during the relevant time period. The remainder of the plaintiffs, though never specifically classified by defendants as “blockers,” fell within the alleged definition of “blockers” and were likewise subjected to adverse employment consequences during the relevant time period.
A district court considering a motion to certify [or decertify] is entitled to look past the pleadings and examine the evidence produced during discovery, but when an ADEA plaintiff relies upon a “pattern or practice” theory and comes forward with legitimate evidence to support that theory, the district court must take into account the framework for pattern-or-practice cases outlined by the Supreme Court in International Brotherhood of Teamsters v. United States, 431 U.S. 324, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977). That did not occur in this case, and hence our application of those factors using the correct legal framework. We do not hold that whenever there is evidence of a pattern-or-practice, a class must be certified. Whether certification or decertification is appropriate depends upon application of the factors we have identified in the ad hoc approach.
Grant of summary judgment
In connection with their motion to de-certify, defendants moved for summary judgment with respect to all of Thiessen’s claims for relief. After granting defendants’ motion to decertify, the district court granted defendants’ summary judgment motion in its entirety. In doing so, the court treated Thiessen’s claims as non-class, individual claims of discrimination. The court concluded that “three of Mr.' Thiessen’s claims, (i.e., his 1993 failure-to-promote claim; his ‘phase-out’ claim); and his downgraded performance claim” were time barred because they occurred more than 300 days prior to the filing of Thies-sen’s EEOC charge. Thiessen II, 13 F.Supp.2d at 1147. As for the remaining two claims (the denial of the two 1995 RPC manager positions), the district court concluded, after analyzing them under the McDonnell Douglas framework, that Thiessen had “failed to produce evidence from which a reasonable factfinder could conclude that defendants’ proffered explanations for [their] actions were ‘unworthy of credence.’ ” Id. at 1149.
On appeal, Thiessen contends the district court erred in granting summary judgment in favor of defendants. We review the district court’s grant of summary judgment de novo. Kidd v. Taos Ski Valley, Inc., 88 F.3d 848, 851 (10th Cir.1996). Summary judgment is proper only when “the pleadings, depositions, answers to interrogatories, and admissions on file, together with the affidavits, if any, show that there is no genuine issue as to any material fact and that the moving party is entitled to judgment as a matter of law.” Fed.R.Civ.P. 56(c). We view the evidence and all reasonable inferences derived therefrom in the light most favorable to the nonmoving party. Kidd, 88 F.3d at 851.
As with defendants’ motion to decertify, proper consideration of defendants’ motion for summary judgment must take into account the fact that Thiessen and the opt-in plaintiffs were asserting a pattern-or-practice claim. Although there is little case authority discussing summary judgment motions in pattern-or-practice cases, we see no reason why summary judgment motions cannot be aimed at both the first and second stage issues. Presumably, however, such motions must be analyzed in light of the orders of proof pecu[1109]*1109liar to pattern-or-practice cases, and must be filed and considered at an appropriate stage of the proceedings. During the first stage of a pattern-or-practice case, for example, a summary judgment motion (whether filed by plaintiffs or defendants) must focus solely on whether there is sufficient evidence demonstrating that defendants had in place a pattern or practice of discrimination during the relevant limitations period.9 See Teamsters, 431 U.S. at 357-61, 97 S.Ct. 1843. Until the first stage is resolved, we question whether it is proper for a court to consider summary judgment motions regarding second stage issues (i.e., whether individual plaintiffs are entitled to relief). Even assuming, arguendo, such motions can properly be considered prior to resolution of the first stage, it is clear they would not be analyzed under the typical McDonnell-Douglas framework. See id. (rejecting use of McDonnell-Douglas framework in pattern-or-pattern cases); see also Davoll v. Webb, 194 F.3d 1116, 1147 (10th Cir.1999) (discussing Teamster’s rejection of the McDonnell-Douglas test in pattern-or-practice cases). Instead, they would operate under the presumption that (1) defendants had in place a pattern or practice of discrimination, and (2) all employment decisions regarding the class plaintiffs were made pursuant to that pattern or practice. See Teamsters, 431 U.S. at 362, 97 S.Ct. 1843.
We conclude the district court erred in granting summary judgment in favor of defendants with respect to Thiessen’s claims for relief. Consistent with our discussion of the class decertification decision, it is apparent that the first stage issues in this pattern-or-practice case have yet to be resolved. Thus, Thiessen’s claims cannot at this point be treated as non-class claims of individual discrimination or analyzed under the McDonnell Douglas framework. Refusal to alloiv certain plaintiffs to join class
When it initially certified the plaintiff class during discovery, the district court addressed several issues, including whether the proposed opt-in plaintiffs, none of whom filed EEOC charges, could piggyback on the EEOC charge filed by Thies-sen.10 The district court first concluded that Thiessen’s EEOC charge asserted allegations of class-wide discrimination and satisfied the notice requirements of 29 U.S.C. § 626(d). The district court then concluded that only “those individuals who could have filed timely EEOC charges at [1110]*1110the time Mr. Thiessen actually filed his charge” could join as opt-in plaintiffs. More specifically, the court concluded that only those individuals who suffered adverse employment actions in the 300-day period prior to the filing of Thiessen’s EEOC charge could opt in. Accordingly, the district court excluded from the class eight proposed opt-in plaintiffs, seven of whom “resigned or were terminated from employment long before the start of the 300 day filing period,” and one of whom “failed to allege any discriminatory treatment in the 300-day filing window.” Thiessen I, 996 F.Supp. at 1078.
On appeal, Thiessen contends the district court erred in excluding these eight individuals from the opt-in class. According to Thiessen, the district court failed to consider evidence that defendants actively concealed the existence of the blocker policy. This evidence, Thiessen suggests, should have equitably tolled the time period for filing EEOC charges based on the blocker policy, and should have allowed all the proposed plaintiffs to opt in to the class'. As previously noted, we apply an abuse of discretion standard in reviewing a district court’s decision to certify a class. Mooney, 54 F.3d at 1213.
Federal courts “ ‘universally hold that an individual who has not filed an administrative charge can opt-in to a suit filed by any similarly situated plaintiff under certain conditions.’” Mooney, 54 F.3d at 1223 (quoting Anson v. Univ. of Tex. Health Sci. Ctr., 962 F.2d 539, 540 (5th Cir.1992)); see Mistretta v. Sandia Corp., 639 F.2d 588, 594-95 (10th Cir.1980) (adopting single-filing rule in ADEA action). “This so-called ‘single-filing rule’ generally allows a plaintiff, who did not file an EEOC charge, to piggyback on the EEOC complaint filed by another person who is similarly situated.” Mooney, 54 F.3d at 1223; see Gitlitz v. Compagnie Nationale Air France, 129 F.3d 554, 558 (11th Cir.1997) (“A plaintiff who has not filed an individual EEOC charge may invoke the single-filing rule where such plaintiff is similarly situated to the person who actually filed an EEOC charge, and where the EEOC charge actually filed gave the employer notice of the collective or class-wide nature of the charge.”). “The policy behind the single filing rule is that ‘[i]t would be wasteful, if not vain, for numerous employees, all with the same grievance, to have to process many identical complaints with the EEOC.’ ” Mooney, 54 F.3d at 1223 (quoting Oatis v. Crown Zellerbach Corp., 398 F.2d 496, 499 (5th Cir.1968)). “As long as the EEOC and the company are aware of the nature and scope of the allegations, the purposes behind the filing requirement are satisfied and no injustice or contravention of congressional intent occurs by allowing piggybacking.” Id. at 1223.
Our threshold task is to determine the time frame that gave rise to Thiessen’s claims of discrimination. See Jones v. Firestone Tire and Rubber Co., 977 F.2d 527, 532 (11th Cir.1992) (holding that, for purposes of single-filing rule, the first issue is the “effective scope of’ the EEOC charge actually filed). Thiessen is relying on the continuing violation doctrine to bring in claims that arose more than 300 days prior to the filing of his EEOC charge. More specifically, Thiessen alleges that defendants implemented the blocker policy and began taking adverse action against him in pursuit of that policy beginning in September 1993. Further, Thiessen alleges that, notwithstanding Lanik’s repudiation in the fall of 1994, defendants continued the policy and applied it against him in 1994 and 1995. If Thiessen can substantiate his allegations, the time frame for his claims would generally be from [1111]*1111September 199311 through 1995. Accordingly, it would be this time frame during which the district court should make its determination of which opt-in plaintiffs to include in the class. In other words, the class should, at this stage of the proceedings, include all those plaintiffs whose related claims (i.e., adverse employment actions resulting from application of the blocker policy) arose between September 1993 and 1995.12 Necessarily, this would include at least four of the eight excluded individuals, all of whom resigned or were laid off during or after September 1993.13
To view the problem in a different way, the question could be posed as whether each of the proposed opt-in plaintiffs could have filed a timely EEOC charge at the time Thiessen filed his charge. Because all of the proposed opt-in plaintiffs rely on the continuing violation doctrine, all could conceivably have filed timely charges at the time Thiessen filed his charge.14 See Purrington v. Univ. of Utah, 996 F.2d 1025, 1028 (10th Cir.1993) (noting that the continuing violation exception can be invoked by showing “the maintenance of a company-wide policy of discrimination both before and during the limitations period”). Accordingly, on remand, the district court is directed to include in the opt-in class all those plaintiffs who allegedly suffered adverse employment actions from September 1993 through 1995.
Having reached this conclusion, we find it unnecessary to address Thiessen’s active concealment/equitable tolling argument. Even assuming, arguendo, defendants actively concealed the existence and implementation of the blocker policy from plaintiffs, only those plaintiffs who suffered adverse employment actions after implementation of the policy could be included within the class.
Deposition of defendants’ corporate counsel
Thiessen contends the district court erred in refusing to allow him to depose defendants’ corporate counsel, Sarah Gor-[1112]*1112man, regarding her investigation of the blocker policy and the alleged purging of corporate documents pertaining to the blocker policy. According to Thiessen, Gorman was the person who, in the spring of 1994 in response to a charge of age discrimination filed by Rick Richards, directed employees to purge company files of documents pertaining to the “blocker policy.” Thiessen contends he has a right to depose Gorman regarding these “coverup” activities, notwithstanding her status as corporate counsel and her involvement as counsel in this case.
We review a district court’s determinations regarding waiver of attorney-client privilege and work-product protection for abuse of discretion. Frontier Ref., Inc. v. Gorman-Rupp Co., 136 F.3d 695, 699 (10th Cir.1998). In doing so, however, we review the district court’s underlying factual findings for clear error, and its rulings on purely legal questions de novo. Id.; see In re Grand Jury Proceedings, 156 F.3d 1038, 1042 n. 1 (10th Cir.1998).
When the issue of deposing Gorman was raised in the district court, the parties and the court agreed it was controlled by the rule announced in Shelton v. American Motors Corp., 805 F.2d 1323, 1327 (8th Cir.1986).15 In Shelton, the Eighth Circuit held that depositions of opposing counsel should be limited to circumstances where the party seeking to take the deposition has shown that: (1) no means exist to obtain the information other than to depose opposing counsel; (2) the information sought is relevant and nonprivileged; and (3) the information is crucial to the preparation of the case. After hearing arguments on the issue, the district court concluded that Thiessen had failed to demonstrate the first two requirements of the Shelton rule, i.e., that the information sought from Gorman was unavailable elsewhere and that the information was nonprivileged.
After reviewing the record on appeal, we find no abuse of discretion on the part of the district court in refusing to allow Thiessen to depose Gorman. In particular, Thiessen has failed to demonstrate that the district court’s factual findings regarding the first Shelton requirement (i.e., the availability of other avenues for obtaining the information) were clearly erroneous. It is therefore unnecessary to address the district court’s ruling on the second Shelton requirement (i.e., that the information sought was privileged).
III.
The decision of the district court decerti-fying the class of opt-in plaintiffs is REVERSED; the decision of the district court granting summary judgment in favor of defendants with respect to plaintiff Thiessen’s individual claims of age discrimination is REVERSED; the decision of the district court excluding eight individuals from the opt-in class is REVERSED; the case is REMANDED to the district court for further proceedings consistent with this opinion.