2
1. Introduction
Over the past three decades, political scientists have advanced numerous models to
account for the empirical observation that the two major American political parties (and
their candidates) consistently present divergent policy positions.
1
Formal theorists have
incorporated into their spatial models such factors as politicians’ policy motivations
(Wittman, 1973, 1983; Roemer, 2001; Cox, 1984; Groseclose, 2001; Aldrich, 1995), the
threat of entry by third parties (Calvert, 1985; Callander, 2000), the possibility of acquiring
scarce campaign resources by appealing to special interests or party activists (Baron, 1994;
Miller and Schofield, 2003; Moon, forthcoming), restrictions on candidate positioning
(Kollman et al., 1992), and politicians’ expressive motivations (Roemer, 2001). Empirical
researchers emphasize many of these same factors, and have also investigated the potential
polarizing effects of primary elections (Burden, 2000; Fiorina, 1973, 1974; Polsby, 1980).
Recently, Adams and Merrill (2003; see also Adams, 2001; Adams, Merrill, and
Grofman, forthcoming.) have advanced an alternative motivation for candidate policy di-
vergence: namely, that Democratic and Republican candidates can use their policy appeals
to mobilize members of their partisan constituencies to turn out to vote, but that – provided
that voter abstention is driven by alienation from the candidates – candidates’ policy posi-
tions have limited impact on the turnout decision of the members of the rival candidate’s
partisan constituency. This argument – which relies on the behavioral researcher’s voting
model that was first explored in the spatial modeling context by Erikson and Romero
(1990) – implies that each candidate has electoral incentives to reflect her own constitu-
ency’s policy preferences, and, given that Democratic partisans tend to hold more liberal
beliefs than Republican loyalists, it implies that candidates from both parties have electoral
motivations to diverge from the center. Thus unlike the alternative explanations summa-
rized above, the Adams-Merrill partisan constituencies explanation implies that, ceterus
paribus, candidates enhance their vote margins in general elections by shifting away from
1
For empirical evidence on the policy divergence of the Democratic and Republican parties (and their candi-
dates) see Poole and Rosenthal, 1997; Ansolabehere et al., 2001; Burden, 2001; Erikson and Wright, 1993,
1997; Page 1978.
Convention