Samuel Kernell and Bernard Grofman
Gerrymandering is intended to have the effect of creating unequal opportunities for racial or political groups to elect candidates of choice. 1 For example, Morrill (chapter 10) defines gerrymandering as "the intentional manipulation of territory toward some desired electoral outcome," while McDonald and Engstrom (chapter 8) define it as "the drawing of electoral districts so as to assign unequal voting weights to cognizable political groups." These definitions highlight two key aspects of the gerrymandering issue: the need to identify relevant cognizable groups and the need to identify the probable consequences for these groups of particular manipulations of district boundaries.
In this paper we shall focus on one important type of political group, the political party, and confine ourselves to one state, California. Moreover, we shall look only at the major parties in California, Democrats and Republicans. Before we can decide if there are manageable standards to detect and control partisan gerrymandering, we believe it is important to determine if the concept of partisan voting strength can be meaningfully defined for the units of political geography (ranging in size from census blocks or census tracts to whole cities or counties) that are the building blocks of state legislative or congressional districts. If this cannot be done, then the search for a measure of gerrymandering may be futile (see however, Wells, 1981; chapters 13 and 16, this volume, for other points of view).
It has been argued that the concept of partisan voting strength is