Introduction to the Special Issue

Starting as a metaphor common in discussions of mass elections, the model’s basics were developed by Hotelling (1929) and Downs (1957) for use in the electoral arena and by Black (1948a,b) for use in legislative studies.> In both its electoral and its legislative versions, the spatial model was originally purely deterministic. Voters were presumed to be faced by w issues, each issue represented by a dimension in a policy space, W (a subset of Euclidian w-space).

Voters’ preferences were represented by utility functions defined overW, with each voter possessing a unique ideal or most-preferred point. When presented with a choice between rejecting a bill, and thereby retaining the status quo s ∈ W, or passing that bill, and thereby instating a new policy b ∈ W, a voter voted “aye” (for passage) if she preferred b to s, voting “nay” (against passage) if she preferred s to b. It took some time before the somewhat parallel literature on scaling legislators’ roll call votes—which assumed probabilistic voting and focused on the empirical task of recovering legislators’ scale positions—merged with the spatial model.