Plurality and runoff systems and numbers of candidates

Although political scientists have in the last few decades learned much about the abstract properties of multicandidate electoral systems, we have accumulated little theoretical knowledge on the ways in which strategic behavior of voters and candidates changes when we move from one voting system to another. This essay has illustrated how the relative desirability of voting systems can change when the number of candidates — usually considered fixed in such comparisons — is taken as endogenous to the voting system in force at the time. Our main conclusion is, therefore, that the method of voting must be understood in relation to the total electoral system. It is not enough to analyze properties of voting systems in the abstract, but also in terms of the interactions that voting systems have with political groups and the constitutional setting. A complete comparison should, consequently, consider also the way in which these systems tend to reduce or inflate the number of candidates. Indeed, the process by which actual candidacies are eliminated may be as important as the fact that the numbers are reduced. In a sense, both plurality and runoff reduce the relevant candidates in different ways: runoff through the mechanical elimination of all but the top two candidates (assuming no candidate receives a majority of the vote), and plurality, through the pre-election negotiations and calculations that tend to discourage candidates who fear they may not come in first. One particularly difficult problem is the possibility that plurality, in the process of discouraging candidates who do not expect to come in first, may also discourage potential Condorcet winners from running. Under the runoff system a Condorcet candidate, even though squeezed on both left and right, may choose to run, hoping for a second place finish that may translate to a majority at the second election. Likewise, a dark horse candidate, who is potentially a Condorcet winner given appropriate publicity following a second place finish initially, also has an incentive to run. In this second example, the identity of the Condorcet candidate is dynamic. By contrast, under plurality, such present or potential Condorcet candidates may fail to announce. These concerns seem to be echoed in the Blacks' (1987) study of the current southern experience with the runoff system, which, in the absence of strong party organizations, ‘provid[es] a mechanism for eliminating many weak frontrunners — candidates who might be able to win a small plurality of the vote in a crowded first primary field but who would likely be ineffective campaigners in the general election’ (Black and Black, 1986: 6). They also find that strong frontrunners in runoff gubernatorial primaries — candidates who win at least 40 percent of the initial primary vote and have at least a 5 percentage point lead over their closest rival — have not lost an election in the last two decades. Political novices or dark horses, meanwhile, who manage to qualify for the runoff against an incumbent, generally perform quite well. Of course, whether these candidates are Condorcet winners and whether, more generally, the runoff system does in fact favor ‘potential’ Condorcet candidates is not known. In fact, there is little direct empirical evidence on the existence or emergence of Condorcet winners among given candidate fields competing under either the runoff or plurality system. And one important investigation for future research is, for example, to discover how and why a Condorcet winner in a 14-candidate field becomes non-Condorcet when there are 8–13 candidates, and then Condorcet again when there are 3–7 candidates. A less puzzling problem concerns the reluctance of southern Democrats to abandon the runoff system, which continues to serve the electoral interests of that party (Stanley, 1985). Likewise, the vast majority of states continue to use the plurality method, perhaps because it maintains a political status quo, or perhaps because politicians recognize, if only dimly, the attractive features of simple majority decision promoted by the use of this rule. Similarly, that the dynamic movement toward only a few candidates might still involve certain unfairness was probably perceived by those adopting the runoff primary in the South. Thus, while the historical origins of the runoff system have not been thoroughly investigated, it seems likely that with the disappearance of the two-party system (and the accompanying weakening of the southern Democratic party organization) in the 1890s, southern Democrats wished to avoid the bruising intra-party contests that might have resulted under plurality rule. Alternatively, where the two-party system still existed, the runoff provided a means to incorporate the many factions of the Democratic party with less risk that these groups might ‘bolt’ that party (Kousser, 1984). Then, as Key (1949) suggests, they might simply have hoped — in the event that many factions or candidates were not initially deterred from entering primary races — to avoid nominating ‘minority’ candidates. Unfortunately, while we know little on the historical origins of the runoff primary, we know even less about the dynamics of the operation of electoral systems such as plurality and runoff, in terms of their effects on the status of potential (as well as eventual) Condorcet candidates. The tentative conclusion from this essay thus reaffirms that of Arrow (1951), as elaborated elsewhere (Riker, 1982b): speaking descriptively, voting systems operate in ways that are not fully understood; speaking normatively, all voting systems are unfair in certain ways. Still, we have learned that the plurality system is much more efficient in selecting Condorcet winners than previous research suggests and is in fact more efficient than the runoff system in this regard. Copyright Kluwer Academic Publishers 1989