Heads I Win, Tails The Coin Was Loaded

Sound the alarms!

As consumer, employee and other groups carefully build momentum in Congress for changes in the nation’s arbitration landscape and business groups just as carefully organize their opposition, a new empirical study reports a “disturbing trend” at the state level: state courts vacating many arbitration awards for employees, but not for employers.

See, here’s the thing: if the statistics were the opposite, these same people would be arguing (as they do in with other types of arbitration) that the arbitration panels are biased against them, and they’d cite the reversal rates by the courts as evidence that the arbitrators were less fair than a court would be.
In fact, overall statistics of this nature are famously uninformative because they assume a static universe in which the cases decided by arbitrators or the courts are a representative, evenly divided sample. But there are numerous ways in which data can be biased – just for example:
*If a forum is more favorable to plaintiffs, it may attract more weak or frivolous cases, and thus end up with a higher rate of defense victories – sort of the way outfielders with weak arms get a lot of assists because a lot of people run on them (between 1993 and 2003, Mike Piazza threw out 384 base thieves, Pudge Rodriguez threw out 387 – if you looked just at the total number thrown out, you might draw a very bad conclusion).
*Highly meritorious cases are much more likely to settle, especially in arbitration where plaintiffs are less likely to hold out for massive punitive damages. But the prevalance of nuisance-value settlements means it’s also impossible to use settlement data as a reliable proxy for the merits, especially if you lack the means to assess the value of the settlement.
*Defendants who are repeat players (in employment litigation, that’s pretty much every business) may be more likely to go to court to challenge awards they are dissatisfied with than employees represented by attorneys working on commission.
*Cases can settle at any stage of the process, so these numbers also don’t include cases where a settlement is reached somewhere between the arbitration award and the court decision reviewing it. A defendant who wins in arbitration but faces a likelihood of reversal in court may very well decide to settle the case while the getting is good.
That’s even before you get into the asymmetries here – in most employment cases the employee is the plaintiff, who has the burden of proof, a fact that will impact review of the award. The fact is, there are many points in litigation at which decisions can be made by one or both sides about what avenue to pursue next, and each of those decision-points can skew the sample.