2002 Volume 2002 Issue 38 Pages 19-30
The main purpose of this paper is to scrutinize the Duverger's Law. We especially shed light on (1) Duvergerian convergence process as a mechanism that makes the number of prim candidates converges to the stable value, and (2) whereabouts its stable value goes.
In the first section Duverger's law is formalized for applying it to OLS estimation. In the second section, it empirically analyzed it by using single member district (SMD) data of Japan 1996 and 2000. The conclusion can be summarized to the following three points. (I) As long as the number of prime candidate is more than 2, Duvergerian slide down convergence works. However when it was smaller than 2, statistically significant conclusion was not obtained. Therefore, (II) the Duverger's Law in a classic meaning is untenable. In addition, (III) some of results statistically reject the hypothesis assumed that Duvergerian equilibrium is equal to 2 in SMD. According to these results, Duvergerian slide down convergence is tenable while the Duvergerian equilibrium departs from 2.