Tid: 8 november 2017, kl 13-14
Plats: B705

Abstract

A common problem with ranking lists (e.g. regarding success/failure proportions of treatments at hospitals) is that smaller units (e.g. hospitals) tend to end up either at the top or at the bottom just by pure chance. To alleviate this problem, we propose a method that, for a given unit, gives a confidence interval for the position of this unit within the distribution of the other units. The method is illustrated by an empirical example, ranking hospitals in Sweden with respect to mortality during treatment after heart attack.

Joint work with Sture Holm, Göteborg University.