Association for Public Policy Analysis & Management

Performance of policy implementation is an important source of information for citizens to evaluate the policy outcome and to hold government accountable. However, given increasing amounts of information disclosure and exchange by social networks, people’s evaluation and attitudes of public policies might be influenced by a social contagion effect rather than the performance information. In this study, we design a survey experiment to investigate whether social contagion effect will influence people’s attitude toward a certain policy, given the performance information available.

The social contagion model in public policy study suggests “adoptions of policies in neighboring states increases the public support of those policies in the home state” (Pacheco, p.189). People learn information about policy implementations in other states through multiple social networks, and information transmitted by such channels has strong influences on people’s political perceptions and preferences. In addition, adaptions of a policy in other states might also signal the level of legitimacy that the policy gains from the public. Therefore, we hypothesize (H₁) that people will evaluate the same policy performance more positively when knowing the policy has been adapted by multiple states.

This model holds an assumption that people use policy adoption information to complement the deficits of policy information. Therefore, it is expected that providing additional information about the policy might reduce people’s reliance on other sources of information. In our scenario, articulated information about underperformance might weaken the supportive attitude toward the policy triggered by policy adoption information. In this study, we use expectancy disconfirmation to indicate an underperformed policy, and test the hypothesis (H₂) that articulated information of policy underperformance will moderate the positive attitude that resulting from the social contagion effect.

In this study, we use a 2 by 2 online survey experiment on Amazon MTurk. Subjects will be asked to read a policy memo reporting the performance (recidivism rate) of a state wide criminal justice program for education and reentry called Alternatives-to-Incarceration (AOI) in recent three years. We randomly assign subjects into four groups: 1) control group; 2) treatment group reporting policy adoptions in other 20 states; 3) treatment group reporting underperformance; 4) treatment group reporting both policy adoptions in other 20 states and underperformance. Citizens’ attitudes of continuing the fiscal investment on this policy will be measured after the vignette on a group of Likert scales.

The results of the experiment will expand the theoretical discussion on performance information delivery. Our study extends the conversation by inviting additional information that might not be directly relevant to the policy performance and explores how such information might interact with the characteristics of performance data. Thus, experimental results of this study will contribute to our systematic understanding of performance information in a more complex context.