sample The analysis is based on survey data from 101 organi-zational units within the Dutch public sector. In building the questionnaire, we relied as far as possible on instru-ments validated in previous studies. Also, we pre-tested the survey instrument by asking five public sector man-agers to complete the questionnaire and to discuss theirexperiences with us. This has led to some minor adjust-ments in the original survey design. The relevant parts of the questionnaire are reproduced in the Appendix. Respondents are managers of organizational units. We define an organizational unit as a more or less unifiedadministrative entity within the larger organization in which the unit’s manager has at least some degree of authority over the set of tasks and processes of the unit(cf. Cavalluzzo and Ittner, 2004). The organizational unit is the appropriate level of analysis for our project. Espe-cially in larger organizations, performance measurementpractices within the organization and the degree of con-tractibility may differ widely. This intra-organizationalvariety, however, is less of a problem at the level of the organizational unit, where activities and performance measurement choices are generally more homogeneous.Our dataset contains only observations from units engaged in activities that are somehow typical of the public sector.Thus, we have excluded more generic support units suchas human resource management or finance departments from our sample. We have also excluded small units (num-ber of employees