GM Ignition Switch Litigation Bellwethers

Lance Cooper, a lawyer with ignition switch cases against GM, has made a motion to remove the MDL plaintiffs counsel in the ignition switch litigation.  You can find some coverage by Sara Randazzo & Mike Spector at the Wall Street Journal. (I haven't seen the motion). 

The first lawsuit to proceed to trial - of a total of six, three picked by plaintiffs, three by defendants - has ended with a dismissal with prejudice under a cloud of allegations of fraud.  What does this say about bellwethers?

I think it says nothing about the underlying cases, or not very much.  (It does tell you something about lawyer error, but that's a topic for another day).  However, the recent events at trial do show that the way bellwether trials are structured is deeply flawed.  If the cases tried are going to be meaningful, they should be randomly selected and the number of trials should be related to the variation in the underlying group of cases.  If there is a high variation, you will need more trials to tell you much about the underlying run of cases.  A number that is convenient (six, for example) is just that, convenient, but convenience should not be confused with meaningful.  Now a lot is riding on this case because one would imagine, since plaintiffs picked it and they only get three, its a really good case for plaintiffs.  So a skewed sample can tell you something - but what it ends up telling you is more about the lawyers than the underlying run of cases. 

To understand this, imagine you have a jar full of marbles.  If you know all the marbles are the same color, you can just pick one marble out of the jar to find out what the color of all the marbles is.  But if the marbles are of various colors picking one marble is not enough.  If you have a sense of the distribution of colors in the jar - say you know that there are some black and some red marbles - you can calculate how many marbles to pick out so that you will have a pretty good estimate of the proportion of red to black.   The same with cases.  If your cases are homogeneous, you can just try a few to get a sense of their value.  The more heterogeneity, the greater the number of cases that need to be tried.  This was the basis for Francis McGovern's idea of maturation of mass torts - you try lots of cases, and over time a value emerges.  It was also the basis for the structure of bellwether trials in the 9/11 First Responders' Litigation.  

Of course, its easy for me to say this. I am not running an MDL and I don't have to pay for all those trials, which are expensive. But that said, if you want to get a sample that could mean something, the sample size needs to be related to the variance of the underlying class and the method of selection needs to be random.  That's basic statistical methods.  I think that MDL judges need to partner with statisticians and have a serious conversation about what bellwether trials are trying to achieve and how best to do this.  Judges have a lot of discretion and they can use it wisely to lead to fair and equitable results for everyone. There are ways to do better, we just need to find them.