Until about fifteen years ago, spatial modelers focused almost exclusively on elections in which the candidates/parties competed over what Donald Stokes (1963) labeled “positional” dimensions of candidate evaluation – that is, dimensions along which voters (and candidates) could plausibly hold a range of preferred positions. Positional dimensions encompass policy debates relating to such issues as abortion (where voters may prefer pro-life or pro-choice positions), gun control, tax policy, the European Union, immigration policy, as well as overarching ideological dimensions where voters' preferences are typically arrayed along a Left–Right or liberal–conservative scale. A central feature of spatial models of position is that voters are assumed to display a distribution of preferred positions along the dimensions included in the model, and candidates compete by shifting their announced positions in an effort to attract support from policy-motivated voters.

In an influential early review of the spatial modeling literature, Stokes (1963) argued for the need to incorporate valence dimensions of candidate evaluation into spatial models of elections. In Stokes's formulation, valence dimensions were those along which all voters held identical positions (preferring more to less), such as reducing crime, increasing economic growth, or desiring candidates/party leaders who display valued personal characteristics such as integrity, competence, and charisma. The two critical aspects of valence dimensions of evaluation are first, that voters hold identical preferences along these dimensions but may perceive the candidates as being at different locations (i.e., candidates may be perceived as more or less honest, charismatic, or effective at reducing crime or managing the economy); second, that candidates may find it difficult or impossible to substantially alter their reputations along these dimensions (i.e., a candidate cannot easily change his public image with respect to competence and integrity).