摘要：The United States Supreme Court, in its 1974 decision, Eisen v. Carlisle & Jacquelin, held that judges should not conduct a preliminary inquiry into the merits of a suit as part of the decision whether to certify a class. The federal courts have struggled ever since to honor Eisen's bar while still conducting a credible certification analysis-a task complicated by the fact that merits-related factors are often relevant to Rule 23 requirements. The result is a muddled body of case law in which courts tend to certify generously and avoid inquiring into the merits of substantive issues even when those issues are crucial to the certification analysis. This approach creates high social costs by inviting frivolous and weak class action suits. This Article argues that the Eisen rule should be abolished. Trial judges should assess competing evidence, not just allegations, and should evaluate case strength whenever the specific requirements of Rule 23 call for an inquiry into merits-related factors. For example, a party relying on a substantive issue to show commonality or predominance should have to demonstrate a likelihood of success on the issue. The Article also goes further and recommends that judges always conduct a preliminary inquiry into the merits before certifying a class, regardless of whether merits-related factors are directly relevant to a specific requirement of Rule 23. The Article first reviews the history of the Eisen rule and surveys the current state of the law, before turning to a policy analysis of the rule's effects. The policy discussion criticizes the traditional arguments and then offers a systematic evaluation of error and process costs. Error costs must be evaluated in light of the extremely high probability of postcertification settlement. Eisen's liberal approach creates a substantial risk of erroneous certification grants that cannot be corrected later when a case settles. This risk coupled with the high likelihood of settlement invites frivolous and weak class action suits. The result is a serious error-cost problem with regard to certification. At the same time, requiring a merits review at the certification stage increases the risk of erroneous certification denials. But for several reasons this risk is not likely to increase dramatically, and the associated costs are not likely to be large. The net result therefore supports a merits inquiry, and this conclusion remains valid even after process costs are added to the policy mix.
