I agree with Professor Hiwatari that the causality of electoral systems is more a matter of favoring one strategy over another. For example, the third party in Britain has grown in votes but not in seats. The multi-partization in Japan has happened in bigger districts but less in smaller ones.

By the way, my data shows that there is no correlation between district magnitude and urban-rural. The idea that rural areas were more three-member makes a lot of sense but does not turn out to be the case. Think of the zenken ikku, Tottori (4) and Shimane (5). After 1986 the two-member were all relatively rural and the 6-member was metropolitan, but the 3's, 4's, and 5's are equally distributed.

SReed.