Association for Public Policy Analysis & Management

Many organizations, particularly local governments, offer performance information to the public using "dashboards" or similar performance reports. These dashboards are intended to inform the public about how local governments are performing on various metrics. However, previous research has shown that individuals motivate their reasoning through ideological frameworks, making errors in judgment that correspond to their ideological beliefs (Kahan, Peters, Cantrell Dawson, & Slovic, 2017). Nevertheless, local governments spend considerable resources in constructing dashboards and navigable user interfaces.

This paper will extend previous research on numeric reasoning with graphical performance information using a survey experiment and nine separate treatment groups and one control group. This survey experiment will also extend previous research by testing liberal, moderate, and conservative (in the American context) cues from public officials as well as liberal, moderate, and conservative organizations. Importantly this research will be able to test whether people or organizations dominate errors in interpreting visual performance information (or whether the effects are multiplicative). It will also examine whether “moderate” officials have a similar influence on individual’s perception of performance and whether moderate arbiters have great or less influence over ideological cues.

A pretest survey will validate the perception of ideologies for the treatment prompts (individuals as well organizations). 2,250 respondents will be recruited using MTurk. Power analysis suggests that given the sample sizes and a difference in responses of 15 percentage points, the main results will have a power of .90, while secondary results will have a power of .67, acceptable for such tests of proportions.