The Empirical Importance of Randomization Bias in the Bergen Experiments.

This paper examines the empirical importance of randomisation bias in a Norwegian randomised field trial on a rehabilitation programme for sick listed worders. Inclusion of participants in the trial was base on information obtained from administrative social insurance records. Professional judgements, which are decisive for treatment assignment under normal operating conditions, is assumed to be captured by a standardised screening test of participants prior to the randomisation.