The three key elements of the unified theory offered in this book reflect a synthesis of previous work by the present authors, singly and in combination.

The discounting model, in which voters discount the claims of policy change offered by candidates or parties, was first proposed in Grofman (1985), but the idea lay dormant until Merrill (Merrill and Grofman 1997, 1999) showed its connection to a model that incorporates both proximity motivations and directional motivations (in the sense of Rabinowitz and Macdonald 1989).We note that this model is also related to important theoretical and empirical work on divided government by Alesina and Rosenthal (1995), Lacy and Paolino (1998, 2001), Burden and Kimball (2002), and Lewis-Beck and Nadeau (2004).

The potential role of party loyalty as a centrifugal force in spatial models of party competition was identified by Adams (1998, 2001a,b; see also Adams and Merrill 1999a,b). These models generally give rise to moderate divergence of party strategies in multiparty contests with probabilistic voting, and to substantial policy divergence when voting is deterministic. Adams and Merrill also consider a mixed proximity-directional model, which resembles a discounting model, and which implies optimal strategies that fit very well with the actual (mean perceived) strategies of the parties. Using the same model, Adams and Merrill (2000) obtain similar conclusions for the competitive candidates in the 1988 French presidential election. Drawing on Adams' work, Merrill and Grofman (1999: Chapter 10) report the divergence of optimal vote-maximizing strategies in Norway for spatial models that incorporate partisanship.