E-Democracy: Can We Model Adequately the Close National Presidential Elections?

This research shows that the Regression Discontinuity Design (RDD) can be applied to close presidential elections worldwide to obtain samples of elections whose outcome - -leadership change or leadership continuation - -is assigned quasi-experimentally. It is shown that the number of close presidential elections is sufficiently numerous to fulfil regression discontinuity demands. Furthermore, a robustness checks reveals that the sample indeed behaves quasi-experimentally, as the assignment of electoral outcomes is independent of pre-election characteristics. Using this sample, researchers can estimate the causal effect of leadership changes on the substantive outcome of interest - -for instance, sovereign debt defaults - -without worrying too much for selection effects or omitted variable biases. The study contributes to the modelling of e-Democracy processes, including the elections, the mechanisms and their significance.