useR! 2024

Swiss voters head to the polls around four times a year. One of the responsibilities of the cantons is to collect the results from the municipalities, process and transmit them to the federal level. The Canton of Zurich used to check the data manually and gave feedback in case of suspicious looking results, however this non-systematic approach on an individual data point level was error-prone, labour-intense and often lacked the context of historical trends and issue specific dynamics. To remedy these shortcomings, we developed an R- and Shiny-based application called “PlausiApp”, which conducts outlier detection in real-time on a variety of results, such as shares of voting channels used, intracommunal differences in participation levels, or the deviation of the results from our prediction. It is in use since 2020. The application works as follows: Prior to the voting Sunday, we load historical data and store it in the project repository. On the voting Sunday, the application transforms the live data and runs the checks. Finally, the results are presented in a shiny dashboard.