The research for this project is building and testing a general model of political representation in district.based electoral systems. Unlike concepts such as "equality" and "liberty," "representation," though also a key component of democratic theory, can only be defined in the context of a specific constitutional structure. Although the form of representation is controlled by statute in systems with proportional, or at large, election of legislatures, the consequences of the American winner.take.all, district.based electoral system are less understood. The research is organized into four phases. The first phase is directed at resolving a fundamental confusion in the literature between the concepts of representation and bias. In the second phase of analysis, using techniques of probabilistic modeling and statistical analyses, the research will address how the distribution of voter preferences across districts determines the aggregate levels of bias and representation in each state. This stage is aimed at uncovering a potential paradox: redistricting and even intentional gerrymandering may have effects in predicted directions, but these effects are short.lived. Research indicates that a decade of dynamic partisan swings and patterns of voting for incumbents affects representation and bias substantially more than the legerdemain of any political cartographer. The third phase of the research program examines how particular geographically arranged clusters of voters within a state influence the distribution of voter preferences across districts, and hence, the aggregate levels of representation and bias. This phase allows the researcher to assess the effects of the numerous constraints state constitutions and lawmakers impose on gerrymanderers, such as compactness and limitations on splitting counties and municipalities between districts. Racial and other group biases can also be assessed in this stage. The final phase of the research employs a stochastic model and empirical analysis at the level of the individual voter to explore what factors determine geographic voter configurations. These factors take on added importance insofar as they influence the distribution of preferences across districts, and, hence, the aggregate levels of bias and representation. As classical theories of representation usually refer to individuals, this will make it possible to unify these theories with the literature on legislative redistricting.
