A class of case-mix adjusted probability-based indices for performance evaluation

A family of indices for the assessment of the performance of a group of agencies providing a certain service is introduced to deal with settings where binary variables are used as outcome measures. The indices are built by aggregating groupspecific success probability contrasts in order to account for case-mix, that is, users’ characteristics influencing the outcome and thereby acting as confounders. The proposed family encompasses many particular cases allowing to weight differently performance improvements or worsenings as well as performance shifts occurring at distinct average levels. Therefore, it represents a flexible tool adapting to various applied evaluation contexts. A general framework for estimation is also discussed.