17
path” (such as dual candidacy and relative seat safety), and coalitional dynamics determining
how candidates are nominated appear to affect a legislator’s voting behavior more than the
electoral rules employed in the component of the election in which she plans to run. Again, these
findings suggest that the interaction of the proportional and majoritarian components of mixed
electoral systems may create incentive structures that differ from those that shape the decisions
of voters, parties, and legislators under SMD or PR.
4. Ruining the Experiment?
Though the research on contamination is rather new, and has in all likelihood only scratched the
surface of what is a complex and multi-faceted research problem, the findings summarized here
do suggest that the assumption of independence is quite tenuous. Moser (1997, 1999) recognizes
this potential complication in the performance of “controlled comparisons” and admits that the
two components of the election should be expected, to some extent, to affect one another.
Nonetheless, Moser (1999: 369) also maintains that “controlled comparisons” remain useful;
contamination “does not ruin the experiment,” as outcomes in each tier should still reflect the
incentives produced by the mechanics of SMD and PR.
More recently, Moser and Scheiner (2004: 2), in an empirical analysis of the sources of
variation in the number of parties conducted on data from fifteen mixed electoral systems, have
concluded that the majoritarian and proportional components “tend to approximate their
expected effects” in spite of contamination. The interaction between the two sets of electoral
institutions, though acknowledged, is found not to produce “entirely new outcomes.” In support
of this claim, Moser and Scheiner (2004) present empirical support for a number of propositions.
Most importantly, disproportionality is greater in SMD than it is in the PR tier and the effective
number of electoral parties/candidates is lower in the majoritarian component. These results lead
them to conclude that mixed electoral systems do, in fact, preserve the independent effects of PR
and SMD.
Convention