Abstract
Performance review and assessment (PRA) of public managers has been adopted in several countries, but we still lack detailed knowledge of what affects the effectiveness of PRA systems. Based on interview data and on a survey of Italian public managers, this article aims to investigate this issue by developing and testing a preliminary model of perceived effectiveness of PRA. The results suggest that PRA perceived effectiveness seems related to the clarity of the organizational design, the quality of the PRA process, the involvement of public managers in the PRA process, and the use of PRA results for tightening control.
Keywords:
Notes
1On the whole, no single definition exists of “management control tightness” in the scholarly literature. In broad terms, a tight management control system is one in which a manager's performance is evaluated primarily on her ability to attain budgetary objectives during each reporting period (CitationAnthony & Govindarajan, 1998: pp. 436–437). Other definitions of tight control systems were provided in CitationMerchant, 1998; CitationMerchant & Van der Stede, 2003; CitationVan der Stede, 2001. Here, we understand tightness of management control systems as the degree to which the control system enhances the responsibility of the managers through focused attention on the definition of measurable goals and performance review centered on the ex post control of results achieved.
2Uni-dimensional scale means that the construct can be scaled in one dimension only (i.e., one-dimensional) (CitationDeVellis, 2003).
3The relatively low response rate poses serious concerns about the representativeness of the data collected. Bearing this in mind, the research findings are carefully and cautiously weighted in the concluding discussion
4Means and standard deviations refer to the distribution of the values for each of the six variables constructed as described in section 2. Because each variable is measured by summing up the scores of answers to three questionnaire items (for the perceived PRA design quality) with values ranging between 1 and 5, mean values can potentially vary up to 15 (20 for the perceived PRA design quality).
5The Adjusted Goodness of Fit Index (AGFI) computes the degree of fitness of the model to the data taking into account the degrees of freedom available for testing the model (CitationJöreskog & Sörbom, 1994).
6The Normed Fit Index (NFI) computes the discrepancy of the proposed model with respect to the model in which all the covariances are zero (CitationBentler & Bonnett, 1980).