The article explains why the “in-sample prediction approach” is unreliable.
A key question in class certification decisions in the United States is whether class-wide evidence can be used to determine whether the challenged conduct resulted in harm to all or almost all proposed class members. In this article, coauthors Celeste Saravia and Daniel Ramsey discuss and evaluate an approach that has been used by a plaintiff expert to address this question in more than ten recent antitrust cases.
The approach, which plaintiff experts sometimes call the “in-sample prediction approach,” claims to estimate the impact of the challenged conduct on each transaction. That is, in a price fixing case, it purports to provide an estimate of the overcharge on each transaction, which we refer to as transaction specific overcharges. In every class action we identified in which a plaintiff expert used this approach, the expert found that almost all class members were harmed, and the judge certified the class. In light of recent usage of this approach, and its success in achieving class certification, plaintiff experts are likely to use the approach in future cases as well.
The article explains that the approach is unreliable because it fails two fundamental tests of reliable econometric methods:
- It is not a consistent estimator, meaning that its estimate will not converge to the true value of the overcharge as the data grows large.
- It has a false positive error rate far beyond typical thresholds, meaning that it will find that a substantial percentage of class members were harmed even if they were not harmed.
This article was originally published by Law360 in October 2025.
San Francisco 
