An Alternative to Bemmels's Method of Investigating Biases in Arbitration

The effects of arbitrators' and grievants' characteristics on the arbitration decision have been the focus of several recent studies. In an effort to reveal biases, Bemmels (1988a, 1988b, 1988c, 1990) introduced a method for evaluating arbitrators' decisions. Within Bemmels's framework, the arbitral decision is statistically evaluated as three separate decisions, and models are estimated for each. This approach has statistical problems. Specifically, since two of the decisions are conditional on previous decisions, the estimates are inconsistent due to selectivity bias. In this paper we re-estimate Bemmels's models using ordinal logit. An ordinal logit model not only eliminates selectivity bias (since only one equation is estimated), but also takes account of the natural ordering of alternatives. Our results show that eliminating selectivity bias leads to some modification of the original study's conclusions

Arbitration · Econometrics · Economics · Focus (optics) · Law and economics · Logit · Ordered logit · Political science · Statistics · Advanced Statistical Methods and Models · Computer Science · Law · Law, Economics, and Judicial Systems · Mathematics · Multi-Criteria Decision Making · Psychology

• When Can an Employee Refuse Unsafe Work and Expect to Be Protected from Discipline? Evidence from Canada

  Mark Harcourt, Sondra Harcourt•Industrial and Labor Relations…•2000

  Add to list

• The Influence of Work History Factors on Arbitration Outcomes

  Patricia A Simpson, Patricia Simpson et al.•Industrial and Labor Relations…•1997

  Add to list