Alex Colvin of Cornell University has published one of the first empirical studies of arbitration in the employment context.  He looked at the reports submitted by the American Arbitration Association, one of the leading providers of arbitrations, in California.  The study looked at 3,945 arbitrations, of which 1,213 were decided by an arbitration award.  See abstract of this study. 

Key conclusions include: 1) employees win 21.4% which is considerably lower than win rates in trials, 2) among those few wins by employees, the median award amount was $36,500 and the median award was $109,858, both amounts substantially lower than that reported in litigation awards, 3) mean arbitration fees were $6,340 in cases overall and $11.070 for cases disposed of by an award following a hearing (In 97% of these, the employer paid 100% of the fees other than a small filing fee – pursuant to AAA rules), 4) in 82.4% of the arbitrations, the employee was paid less than $100,000 per year.  Note that the author must be comparing California arbitrations to California state court trials.  The success rate of employees in california state courts is generally higher than that found in federal courts.  See my prior post regarding success rates in federal courts on a national level.   

The study also examined whether there was a repeat player effect, that is, wherher employers who appear repeatedly would receive favorable treatment.  The study indicates that yes, employers who appear more than once achieve significantly lower awards.  The study indicates that when the same arbitrator decides a case with the same employer from prior arbitrations, then those employees receive lower awards and win less often. These findings support the anecdotal evidence suggesting that repeat employers do better and they do better in particular when they use the same arbitrator.  

The repeat player effect has a large impact on employees, since employees will very rarely have more than one arbitration.  Employment arbitrations are far different than labor arbitrations, in which the union would also receive some repeat player effect.