In a rational world class-action suits would be brought when the probability of success is positive, and settlements would happen when the probability of success is less than one.

An alternative hypothesis could be that the probability of success is positive not because of the merits of the case, but because of defects in the law, judges, and juries.

My devil’s advocate point was that the data of few trials in class-action cases does not by itself prove either theory.

We do have some other data to inform us. The Pants suit shows defects in the law; the Anderson conviction shows defects in the litigation process, and giving a woman $11 million for hurt feelings shows defects in juries.

Class-action suits bother me because they go against market theory. For example, airplane travel is extremely safe because people will avoid airlines that have crashes, not because of class-action lawyers. The Vioxx ligation shows crazy verdicts when there was hysteria in the press, but recent results are pretty good. The data did not justify pulling the drug in the first place.

If I sue you for $100 million, and I have a lawyer willing to go all the way through, then the pressure of the system is that you WILL give me (or at least my lawyer) money, no matter how far-fetched my claim is.

That’s what “settlement” is in class action: money being transferred from the defendant to the plaintiff (or at least their lawyer).

Do you really think that all but 4 class-action suits in the last 12 years have deserved money? Well, they got money!

What is more likely is that settlements are made to avoid the irrational results from litigation.