Simulating election results and party systems

Simulated data sets are widely used in sciences, natural and social alike, because of their ability to overcome constraints inherent in the real-world data. In political science, particularly in such important subfields as electoral systems and their effects upon party systems, districting and gerrymandering and coalition politics, simulated data sets are often employed to address specific research questions. This study introduces a new method for generating simulated election results/party system configurations. Starting with formulating general requirements to this kind of data, the study develops an algorithm that is suitable for the purpose, and then validates the new method theoretically, by using several tools of quantitative validation developed in political science and empirically. Empirical validation relies on the data from 1391 legislative elections held in 103 countries of the world in 1792 to 2010. The analysis confirms the theoretical and empirical validity of the proposed method.