Fabrice Lehoucq and Kiril Kolev, Varying the Un-Variable: Structure, Electoral Formulae, and Election Quality (PRQ). Here is the abstract:
This paper assesses the hypothesis that election quality is worse under plurality voting systems than under proportional representation (PR). We use a two-pronged research design that permits us to harness the advantages of most similar and most different approaches to limit problems of endogeneity that afflict hypothesis testing in comparative politics. We use a subnational database of more than 1,300 accusations of electoral fraud from Costa Rica (1901–1948) that uniquely varies formulae among (provincial) electoral districts. Our statistical models reveal that plurality leads to more ballot rigging than proportional systems. We also demonstrate that plurality voting systems are associated with inferior election quality in the Quality of Elections Database (QED), which covers 170 countries between 1975 and 2004. Our findings suggest that electoral formulae, a basic feature of institutional design, have as much impact as social structure on whether elections are free and fair.