Estimating public performance bias through an MTMM model: the case of police performance in 26 European countries

Organisational performance is notoriously difficult to measure in the public sector. More often than not, objective performance measures are difficult to construct. Subjective performance is a popular alternative to, as well as, complement of objective performance measures. However, such subjective perception measures are likely to be biased. The bias tends to depend upon the specific stakeholder's position vis-à-vis the focal organisation. We show how a multi-trait-multi-method (MTMM) model cannot only determine the validity of performance measures, but is also valuable in generating estimates the potential biases in both method (e.g. respondent type) and trait (e.g. performance measure) of these subjective performance measures. To demonstrate the benefits of this methodology in public management, we apply this method to the subjective performance of Police Forces in 26 European countries. Our policing example demonstrates that single-handedly the available subjective performance measures are not reliable estimates of overall police performance. Moreover, the analysis shows that all three rater groups have significant bias, with police employees most positively biased about their own performance. Interestingly enough this bias depends on the performance measure; corporate managers are most biased about the police catching burglars, while health care professionals are more biased about policing arrival times.