The Priming Effect of Corruption and Bureaucracy Bashing on Citizens’ Perceptions of an Agency’s Performance

ABSTRACT

Priming theories suggest that negative stories or events can affect how citizens feel about public organizations. However, research concerning the priming impact of both corruption and bureaucracy bashing—two relevant topics today—on how citizens perceive the performance of agencies is lacking. To close this lacuna in the literature, an experimental survey was conducted where respondents in the United States were randomly assigned to one of five performance vignettes. The results demonstrate that when participants were assigned to the vignettes containing the corruption and bureaucracy bashing cues, perceptions of performance were lower. This finding is consistent with priming theories. Moreover, differences within the corruption and bureaucracy bashing vignettes as well as between the corruption and bureaucracy bashing vignettes were not found. The theoretical and practical implications of these findings are discussed in the article.

KEYWORDS:

• agency bashing
• corruption
• performance

Notes

Census data were used to weight the data because men and minorities were underrepresented in the sample. To correct for this, men were assigned a weight of 1.50 and women a weight of 0.75. Furthermore, minorities were assigned a weight of 1.76 and majority population a weight of 0.79. The weights for sex were then multiplied by the weights for minority status to achieve an overall sample weight. The sample weight corrected the data such that males and minorities represent 50% and 37.3%, respectively. The corrected percentage of employees not working is 42.5%. While this may seem high, not working includes such individuals as students, retirees as well as individuals who choose to stay home. Furthermore, the average age of individuals not working is about nine years higher, suggesting that they have a large percentage of retirees.

The data were transformed using the two-step approach used by Templeton (2011). More specifically, it involved “transforming the variable into a percentile rank,” then applying the “inverse-normal transformation to the results of the first step to form a variable consisting of normally distributed z-scores” (p. 41).

Ordinal regression is an alternative estimator to ordinary least squares when normality is violated (Lumley et al., 2002). Consequently, the model was also run using ordinal regression. The results were similar to Table 3 and 4.

Additional information

Notes on contributors

James Caillier

James Caillier is an assistant professor of political science at University of Alabama. His teaching and research interests include performance management, evaluation, public policy, and public budgeting.