Research on organizations’ decisions to adopt evidence-based programs exclusively

draws upon self-selected samples. Consequently, research may have overlooked any first order barriers to program adoption. Using four embedded natural field experiments in

a random sample of all public schools in Denmark (N=368), we find that capturing the

attention of decision-makers raises adoption of an evidence-based book-reading program

by about 50 %-points. Once decision-makers pay attention to the program, their primary

interest is the immediate transaction costs of adopting the program, and not—as usually

expected—the evidence about the benefits for clients nor the running costs of operating

the program.

Behavioral decision-making theories portray decision-makers as informationally and cognitively constrained problem solvers (Cyert and March, 1963). A central assertion is that performance below aspiration levels initiates problemistic search, which is failure-induced and solution-oriented. Search behaviors may also aim to identify relevant characteristics and causal origins of low performance (Posen et al., 2018). Empirical assessment of attention remains scarce, but recent studies suggest that negative performance increases decision-makers’ solution-generation but not problem-definition search (Van der Voet, 2023).

We use eye-tracking to test the expectation that negative performance feedback positively affects problem-definition and solution-generation search, but that its positive effect on solution-generation search is stronger. We collaborate with the Dutch waterboard Rijnland – a local authority that addresses urgent challenges concerning water – to sample 30 participants who are presented 20 vignettes in random order. Using equivalence framing, we randomly assign participants to a positive or negative framing of factual performance information on the level of individual vignettes. The vignettes contain problem-related and solution-related information in randomized sequence, for which eye movements are registered (dwell time per area of interest).

The pre-registered study will be implemented in June 2024 and preliminary results will be presented at the EGPA conference. By embedding our study to the challenges of a public authority on the front-lines of climate change adaptation, we aim to generate Behavioral Public Administration insights that contribute directly to solving real-world problems.

References

Cyert, R. M., & March, J. G. (1963). A Behavioral Theory of the Firm. Englewood Cliffs, NJ: Prentice-Hall.

Posen, H. E., Keil, T., Kim, S., & Meissner, F. D. (2018). Renewing Research on Problemistic Search—A Review and Research Agenda. Academy of Management Annals, 12(1), 208–251.

Van der Voet, J. (2023). Search in response to negative performance feedback: problem-definition and solution-generation. Journal of Public Administration Research and Theory, 33(1), 43-62.

It is well established that men and women are treated differently when accessing public services, leading to biased administrative outcomes. We lack however empirical insights on the underlying micro-mechanisms causing inequality in service delivery patterns. This paper seeks to contribute towards closing this gap by investigating gender biases in the verbal communication between street-level bureaucrats and citizens during public service encounters. We focus on the complexity and emotional orientation of bureaucratic speech acts, as two features affecting citizen satisfaction. We draw on original data comprising 156 recorded and transcribed dialogs between street-level bureaucrats and their clients across different service domains. Using a combination of linguistic measures and machine learning methods to measure complexity and emotionality across approximately 20,000 utterances, we find no statistically significant effects for bureaucrats’ gender. Conversely, clients’ gender exhibits a significant impact on administrative language, indicating that bureaucrats communicate more complex and emotional when interacting with male clients. Further, the analysis yields that gender-matching decreases complexity and emotionality in frontline communication. Offering an initial empirical test of the impact of gender stereotypes on frontline communication the paper provides profound novel insights on the micro-foundations of biased administrative practices towards citizens. It thereby majorly contributes to literature on administrative communication, bureaucratic encounters and inequality in public administration.

Rankings and performance information on public services have become increasingly prevalent in the public sector. However, the practical ramifications of performance information remain unclear. It is therefore of interest to investigate how citizens use performance information and whether their use is impacted by personal experiences relating to public performance.

This study examined the effects of performance information on citizens’ trust in public services and voice behaviour in a simulated policymaking environment using a budget game experiment. The experiment utilised a demographically representative sample of United Kingdom (UK) residents recruited using Prolific (N = 2001).

Participants were first presented with three different types of performance information on two fictional schools, after which some participants participated in an online simulation in which they made budget decisions for the schools. All participants were then asked questions measuring their trust in each school and voice behaviour.

Participants were significantly more likely to raise their voice if they were given more detailed performance information. Receiving more detailed performance information also significantly increased the differences in the level of trust in each school depending on the type of performance information provided. Participants who played the budget game expressed significantly lower trust in the schools in the game than the control group and exhibited significantly lower actual voice behaviour but higher potential voice behaviour.

Negative performance information was more impactful than positive performance information. The participants exhibited negativity bias, and reacted more strongly to worse than expected performance than better than expected performance.