DECISION AND JUDGMENT ENTRY {¶ 1} This appeal is from the March 8, 2006 judgment of the Lucas County Court of Common Pleas, which denied the motion of appellants, Countrywide Home Loans, Inc. (hereinafter "Countrywide"), Mary Fackler, and Susan Wirick, to decertify this class action. Finding that the trial court did not abuse its discretion when it held that appellees' method of determining the members of the class was not speculative, we affirm the *Page 2 decision of the lower court. Appellants assert the following assignments of error on appeal:
{¶ 2} "1. The trial court erred by ignoring plaintiffs' failure to identify any additional class members upon the completion of discovery, as required by the Court of Appeals. (Decertification Opinion at 5-7).
{¶ 3} "2. The trial court erred by finding Rule 23's numerosity requirement satisfied by a flawed theoretical model, unrelated to the conditional class definition. (Decertification Opinion at 5-7).
{¶ 4} "3. The trial court erred by failing to consider defendants' arguments concerning the post-November 1999 class period, including that plaintiffs are neither adequate nor typical class representatives for that portion of the class period.
{¶ 5} "4. The trial court erred by finding that an identifiable class exists, notwithstanding that plaintiffs are unable to identify actual class members. (Decertification Opinion at 4)."
{¶ 6} Appellees, Eric Williams and Vonda Williams, brought suit against appellants alleging that appellants had discriminated against them pursuant to a company policy and practice of discrimination when it refused to consider Vonda Williams' income during a loan approval process. Appellants refused to consider her income because she would be on maternity leave at the time of the loan closing. As a result, appellees' loan application was denied. Appellees contended that this action constituted sexual and familial status discrimination in violation of Ohio's Fair Housing Act, R.C. 4112.02(H). Appellees later alleged that Countrywide practiced a statewide pattern of *Page 3 this type of discrimination based upon a similar incident in Dayton, Ohio, in July 1999. Appellees sought declaratory relief, compensatory and punitive damages, and attorney fees. Appellees later added the Toledo Fair Housing Center as a plaintiff and class allegations.
{¶ 7} Appellees sought class action certification based upon a class of all women and their co-applicants who, between June 1, 1997, and the date of trial, contacted appellants regarding a home mortgage loan and were or were expected to be on maternity leave at the time of their loan closing and were told that the woman's income could not be considered for purposes of the loan. Appellants opposed class certification on the basis that appellees failed to meet the numerosity requirement of Civ.R. 23(a)(1), that appellees had failed to demonstrate that the individual issues predominated over the common issues as required by Civ.R. 23(b)(3), and that the class was not identifiable.
{¶ 8} As to the numerosity issues, experts from both parties estimated that the class size to be between 1.4 and 2,480. The trial court concluded that even if only a portion of the estimated members of the class estimated by appellees were class members, there would be enough members to constitute a class action. Therefore, the trial court conditionally certified the class until discovery could be completed. This court affirmed the trial court's finding that the numerosity requirement had been met in Williams v. Countrywide Home Loans,Inc., 6th Dist. No. L-01-1473, 2002-Ohio-5499, ¶ 33.
{¶ 9} Discovery in this case initially encompassed the examination of Countrywide's loan records and depositions of Countrywide's employees. However, *Page 4 appellees assert that they later learned that Countrywide had not preserved the unfunded loan files for the first 2 ½ years of the class period. Therefore, appellees sought to locate class members through the issuance of class notice based upon a list generated from Countrywide's electronic database, which contained demographic and other information about all loan applicants.
{¶ 10} Appellants moved to decertify the class action on August 20, 2004. Appellants alleged that appellees never even inspected Countrywide's loan records that were made available in 2003. Furthermore, they argued that appellees had failed to identify any other members of the class after the completion of discovery.
{¶ 11} Because appellants challenged that the notice proposal for class notification was too broad, the trial court ordered that appellees compare a list of loan applicants taken from Countrywide's computer electronic data base against birth records from the State of Ohio Department of Health to determine whether sufficient class members could be identified who met some of the characteristics of the class.
{¶ 12} Appellees contend that this analysis identified 846 potential members of the class who should receive notice. Appellants, however, argue that this analysis did not disclose any actual class members and that if the analysis were modified to conform to the class definition, only 21 potential class members remain. Appellants argued that the comparison ordered did not consider whether the women who had a child during the time specified by the class description were employed and whether they took maternity leave. The trial court therefore applied the statistical evidence presented by the experts who testified as to the percentage of women employed in Ohio and the percentage of *Page 5 employed women who took maternity leave. Based on this evidence, the trial court determined that 411 potential class members had been identified. Furthermore, the court noted that this figure did not include additional class members that could be found once the comparison model was applied to birth records from 2004 through 2006, when such data was available. The court also recognized that there were other potential members of the class who did not complete a loan application because of information learned during their initial contact with Countrywide. Therefore, in an order journalized on March 8, 2006, the trial court held that there were at least 40 members of the class and, therefore, denied appellants' motion, and this appeal ensued.
{¶ 13} In their first assignment of error, appellants argue that the trial court erred by ignoring the fact that appellees were still unable to name one additional class member after the completion of discovery. In their second assignment of error, appellants argue that the trial court abused its discretion by finding that the numerosity requirement of Civ.R. 23(a) had been met. We will consider these two assignments of error simultaneously.
{¶ 14} When appellants raised the numerosity issue in the last appeal, we held that a preliminary finding that appellees had met their burden of proving the numerosity requirement was not an abuse of discretion. We noted that the trial court had issued a conditional class certification pending completion of discovery.
{¶ 15} Appellants now argue that after completion of discovery appellees are still unable to identify even one other class member. Appellants contend that the information needed to identify additional class members could be ascertained by a review of *Page 6 Countrywide's loan records that appellees have never attempted to review. Therefore, appellants argue that the trial court abused its discretion by failing to decertify the class.
{¶ 16} Class certification judgments are reviewed under an abuse of discretion standard of review. Baughman v. State Farm Mut. Auto Ins.Co. (2000), 88 Ohio St. 3d 480, 483 quoting Marks v. C.P. Chem. Co.,Inc. (1987), 31 Ohio St. 3d 200, at the syllabus. Therefore, we must consider whether the trial court's judgment was the result of an unreasonable, arbitrary, or unconscionable attitude. Marks v. C.P. Chem.Co., Inc., supra at 201. The abuse of discretion standard of review is employed because of the inherent power of the trial court to manage its own docket based upon its "special expertise and familiarity with case-management problems." Hamilton v. Ohio Sav. Bank (1998),82 Ohio St. 3d 67, 70.
{¶ 17} Under Civ.R. 23(A)(1), class certification is proper where the "class is so numerous that joinder of all members is impracticable[.]" The key, therefore, to making a determination of whether an action should be maintained as a class action is to weigh the factors for and against joinder of all of the parties to the action. In re AmericanMedical Systems, Inc., (C.A. 6, 1996), 75 F.3d 1069, 1078-79 (interpreting Fed.Civ.R. 23, which is identical to Ohio's Civ.R. 23). The burden rests on the plaintiff to establish that there are a large number of putative members of the class. Currey v. Shell Oil Co. (1996),112 Ohio App. 3d 312, 317, and Plastic Surgery Assoc, Inc. v.Rachford (1983), 10th Dist. No. 83AP-358, at 2.
{¶ 18} There is no exact minimum number of potential class members needed to satisfy the numerosity requirement. The court must determine the necessary size of the *Page 7 class on a case-by-case approach. Warner v. Waste Mgt, Inc. (1988),36 Ohio St. 3d 91, 97. However, 40 class members is generally the number of members necessary to justify a class action. Id. citing Wright, Miller Kane, Federal Practice Procedure (1986), Section 22.
{¶ 19} The plaintiff must either produce some evidence from which the court can determine the class size or make a reasonable estimate of the class size. Cervantes v. Sugar Creek Packing Co., Inc. (S.D. Ohio E.D. 2002), 210 F.R.D. 611,621. The plaintiff cannot rely upon mere allegation and speculation. Mundell v. Landstyles (Sept. 6, 2001), 8th Dist. No. 78829, at 3; Podany v. Real Estate Mtg Corp. Escrow Co. (1999), 8th Dist. No. 75307, at 8; and Burrell v. Sol Bergman EstateJewelers, Inc. (1991), 77 Ohio App. 3d 766, 773.
{¶ 20} The court may take into consideration the fact that the opposing party controls the information needed to determine the class members. Cervantes v. Sugar Creek Packing Co., Inc., supra. It is also permissible for the plaintiff to make reasonable inferences from the facts in evidence. Id. The trial court may even make common sense assumptions to support a finding that the numerosity requirement was met. Evans v. U.S. Pipe Foundry Co. (C.A.11, 1983), 696 F.2d 925, 930. But, the court must be careful not to grant class certification where no class exists and, as a result, issue an advisory opinion. Rex v. Owensex rel. State of Okla. (C.A.10, 1978), 585 F.2d 432, 436.
{¶ 21} Beyond the mere number of potential class members, the court must consider other factors such as: the practicability of joinder, the geographical dispersion of class members, the efficiency or economy from a class action, the nature of the relief *Page 8 sought, the theory of liability, the ease of identifying the class members and their location, the relatively small size and scope of individual claims, and impediments that prevent individual members of the class from coming forward. Plastic Surgery Assoc, Inc. v.Rachford, supra at 3; Communities for Equity v. Michigan High SchoolAthletics Assn. (W.D.Mich.S.D.1999), 192 F.R.D. 568, 571; and Farm LaborOrganizing Commt. v. Ohio State Highway Patrol (N.D. Ohio W.D. 1998),184 F.R.D. 583, 587.
{¶ 22} The United States Supreme Court has clearly indicated that the requirements of Civ.R. 23(A) must be met to certify a class action even in cases alleging civil rights discrimination. JAT, Inc. v. Natl. CityBank of the Midwest (E.D.Mich.S.D. July 9, 2007), No. 06-11937,2007 WL 2004877, at 2; General Telephone Co. of the Southwest v. Falcon (1982),457 U.S. 147, 155-156; and East Texas Motor Freight System, Inc. v.Rodriguez (1977), 431 U.S. 395, 405. Nonetheless, discrimination cases present their own unique considerations.
{¶ 23} Courts have permitted a few known class members to suffice when injunction or declaratory relief was sought to prevent future discrimination. Plastic Surgery Assoc, Inc. v. Rachford, supra, andHorn v. Associated Wholesale Grocers, Inc., (C.A.10, 1977),555 F.2d 270, 276. Joinder of actions would be impracticable under such scenarios because future plaintiffs would never be known.
{¶ 24} Also, once the plaintiff has presented a colorable claim of discrimination and the plaintiff is part of a large class with potentially similar claims, the court may infer that there is a large enough group to justify a class action. Almendares v. Palmer (N.D. Ohio W.D. 2004), 222 F.R.D. 324, 330;Horn v. Associated Wholesale Grocers, *Page 9 Inc., supra at 276; and Weathers v. Peters Realty Corp. (C.A. 6, 1974),499 F.2d 1197, 1200. The court can infer that persons similarly situated were discriminated against as well.
{¶ 25} Statistical analysis often plays a greater role in cases where individual class members are unknown or unlikely to come forward because of intimidation or lack of knowledge concerning their rights.Armstrong v. Whirlpool Corp. (M.D. Tenn. Nashville D. 2007), 99 Fair Empl. Prac. Cas. (BNA) 1682, 2007 WL 676694, and Senter v. General MotorsCorp. (C.A. 6, 1976), 532 F.2d 511, 523. But the statistics must be specific enough that they permit logical conclusions based upon plaintiffs claims. JAT, Inc. v. Natl. City Bank of the Midwest, supra at 2, and Golden v. City of Columbus (C.A.6, 2005), 404 F.3d 950, 966, certiorari denied (2005), 546 U.S. 1032.
{¶ 26} Appellants contend in their first and second assignments of error that the trial court abused its discretion by failing to find that appellees never met their burden of establishing that there actually were any other members of the class. They contend that the use of a theoretical model results in speculative members of the class rather than actual class members. Furthermore, they argue that the model employed was defective because (1) it was based on testimony about averages which are now 5 to 16 years old; (2) it fails to consider critical class components (such as whether the women were working, planned to take a maternity leave at the time of the loan closing, intended for their income to be considered, and were denied a loan because of their maternity leave); and (3) its resulting "matches" are quite often erroneous. Finally, appellants argue that application of the same model to the period from June 1997 through November 1999 *Page 10 (when Countrywide instituted a company policy to avoid any future similar discriminatory practice from reoccurring) to applications after November 1999 would result in an over-inclusive class. This final argument is addressed under appellants' third assignment of error.
{¶ 27} As we stated in our prior order, the use of expert testimony to establish a reasonable estimate of the number of class members is appropriate in some cases. Williams v. Countrywide Home Loans,Inc., 6th Dist. No. L-01-1473, 2002-Ohio-5499, 32. Appellees did not have an obligation to prove an exact number of class members. In ReRogers Litigation, 6th Dist. No. S-02-042, 2003-Ohio-5976,¶ 23 and ¶ 24. The model developed by the expert and modified by the trial court based upon appellants' objection is a reasonable method for estimating the number of members in the class.
{¶ 28} There is no perfect method to determine the exact number of potential class members in this case. Countrywide's records are incomplete. Loan records might not even have been generated if the person never even completed a loan application after hearing of the alleged discriminatory practice. Future class members cannot be known at this time. Furthermore, potential class members might not even realize that that have been subject to a discriminatory practice.
{¶ 29} We recognize that the expert's model is not perfect since it might include some individuals who would not meet the class criteria (women who did not work or did not take maternity leave). However, the trial court considered this possibility and reduced the total number of class members by the percentage of women who work outside the home and those women who did take maternity leave. While the statistics *Page 11 may have been entered into evidence in another context, they are still facts that the trial court could utilize in its analysis. The statistical assumptions applied by the court were acceptable for the time period in which the discrimination allegedly occurred.
{¶ 30} There are discrepancies between some of the "matches" in that appellees have included in their final list persons whose names or addresses were not identical to data supplied by Countrywide. Appellees justified use of these "imperfect matches" because they could not be definitively ruled out as potential class members. We do not find that these "imperfect matches" are so significant so as to alter the court's final conclusion that at least 40 persons could be included in the class. While the minimum class size remains at one known couple, the court could reasonably infer that there would be enough other class members located throughout Ohio to warrant a class action. Joinder of even a few suits filed across the state would be impractical. This is also the type of case where some potential class members might never come forward because they do not realize that appellants' actions were discriminatory. We find that the model developed by the expert and modified by the trial court based upon appellants' objection is a reasonable method for estimating the number of members in the class. Therefore, under the circumstances of this case, we cannot find that the trial court's rejection of appellants' motion to decertify the class reflected an arbitrary, unreasonable, or unconscionable attitude.
{¶ 31} Appellants' first and second assignments of error are not well-taken.
{¶ 32} In their third assignment of error, appellants argue that the trial court should have at least decertified the class as of November 1999, which is the date when appellants *Page 12 instituted a company-wide policy against the type of discrimination that occurred in this case. Appellants argue that appellees could not serve as representatives of the entire class period since one class of discrimination occurred under the alleged policy of discrimination in force prior to November 1999 and one class of discrimination occurred contrary to company policy after November 1999. Furthermore, appellants argue that appellees have no evidence to prove a claim of expected future discrimination.
{¶ 33} We are inclined to agree with appellees that appellants failed to present these issues to the trial court. The trial court did consider whether the claims of the class members after 1999 could survive summary judgment on the basis that their claims of discrimination could not be proven since there was a company policy against such discrimination after 1999. The court concluded that such discrimination could still have occurred after the change in company policy and that it was not the burden of the class representative to prove each individual claim.
{¶ 34} Appellants did not assert in their initial memorandum in support of the motion to decertify the class action that appellees were not adequate representatives of the entire class or that there was no evidence to support an expectation of future discrimination. The adequacy and typicality arguments were merely mentioned in footnote No. 8 of appellants' third supplemental memorandum in support of the motion and were reiterated in the concluding statement of appellants' fourth supplemental memorandum. Such issues could have been and should have been clearly set forth in the motion to decertify. Because of the subtleness of the presentation of these issues, we cannot presume that they were not entirely overlooked by the trial court. Rather than find *Page 13 that the issues were waived for purposes of appeal and rather than presuming that the trial court's silence was a rejection of appellants' argument, we conclude in this case that the more appropriate action is to remand this case to the trial court for consideration of these reasons for decertifying the class action. Therefore, appellant's third assignment of error is well-taken on grounds not asserted by appellant.
{¶ 35} In their fourth assignment of error, appellants alleged that the trial court abused its discretion when it found that the conditional class was identifiable. Appellants argue that Countrywide's loan files are the only source for determining if discrimination on this basis occurred in other applications and that its electronic databases are not sufficient for identifying class members. Appellees contend that this issue has already been reviewed by this court and should not be reconsidered under the law of the case doctrine.
{¶ 36} We addressed the issue of whether appellees had presented an identifiable class in our earlier decision of this case. Williams v.Countrywide Home Loans, Inc., 6th Dist. No. L-01-1473, 2002-Ohio-5499, 32. This issue was also not raised in appellants' motion to decertify. Therefore, we will not address it again. Nolan v. Nolan (1984),11 Ohio St. 3d 1, 3. Appellant's fourth assignment of error is not well-taken.
{¶ 37} Having found that the trial court did commit error prejudicial to appellants in part, the judgment of the Lucas County Court of Common Pleas is affirmed in part and reversed in part. The judgment of the trial court is affirmed in all respects except that the trial court should have addressed the issue of whether appellees adequately represent the class and whether their claims are typical of the class. Therefore, this case is remanded to *Page 14 the trial court for further proceedings on the motion to decertify. Appellants are ordered to pay the costs of this appeal pursuant to App.R. 24. Judgment for the clerk's expense incurred in preparation of the record, fees allowed by law, and the fee for filing the appeal is awarded to Lucas County.
JUDGMENT AFFIRMED IN PART AND REVERSED IN PART.
A certified copy of this entry shall constitute the mandate pursuant to App.R. 27. See, also, 6th Dist.Loc.App.R. 4.
Peter M. Handwork, J., Mark L. Pietrykowski, P.J., Thomas J. Osowik, J. CONCUR. *Page 1