A multicriteria blockmodel for performance assessment

This paper considers a network-based view of performance differences between organisations. Each organisation is considered in terms both of its size--the level of performance--and its shape--the means whereby the performance is achieved. Since many organisations prefer to monitor their performance via a number of performance measures rather than a single efficiency this is the approach adopted. A weighted additive multicriteria model is used to give an overall measure. The weights, inevitably imprecise, are modelled probabilistically resulting in correspondingly probabilistic estimates of the difference between pairs of organisations. Significant differences are identified and a binary network of relations built showing pairs for which performances are not significantly different from each other. Similarly, correlations between sets of measures leads to a second network showing pairs with similar shape. Blockmodels are built to show the extent to which differentiation between organisations can be found. This structural description is used to examine the changes in performance differences over time. The data used for illustration describe the performance of some airports over a nine year period.