Redistricting

Every decade, American political parties seek electoral advantage through redistricting to win more legislative seats than warranted by the vote while incumbents often cooperate across party lines to preserve their own seats.

Gerrymandering is altogether evil. It effectively disenfranchises millions of voters by dumping them in districts where the results are foregone conclusions. It makes incumbents arrogant and unresponsive because they do not need to fear the voters. And it polarizes American politics because the important elections tend to be primaries, where the party bases rule rather than the less politically motivated middle.

Gerrymandering is not just a device to increase party advantage. The courts have required gerrymanders to ensure representation for preferred but not all minorities. There are amusing exceptions and a few interesting cases of unintended consequences. A state legislator eliminated a precinct from his district just to get rid of a volunteer who cost more votes than she gained. One wanted to retain his mother’s residence in his district, another to keep his vacation home in his. In New Hampshire, a town with four-fifths of the population required for a state senator went unrepresented in one of every five legislative sessions. From 1792 to 1964, the state Senate consisted of districts with representation based on how much each county paid in public taxes (Hardy 1981). John Nance Garner, chairing reapportionment in Texas after the 1900 census, shaped a congressional district for himself, won, and later became Vice President of the United States.

There is no simple way to identify when a gerrymander has, or has not, occurred. Some authorities measure the extent of a gerrymander by the difference between a party’s percentage of elected seats and a party’s percentage of the statewide vote. An alternative is to determine how close a proposed districting plan comes to the “perfect gerrymander,” which is “any set of districts such that no other set, or no possible redistricting, could increase the advantage to the favored party.” Such a perfect gerrymander is not necessarily unique, but the maximum political strength that it can gain is, so it provides a standard of comparison (Musgrove 6-7).

The major impact of the gerrymander for the first sixty years of the twentieth century was over-representation of rural counties at the expense of rapidly growing cities and suburbs. In 1910, counties under 25,000 people represented 30% of the population but held 34% of the seats in Congress. By 1930, counties under 25,000 people had shrunk to 22% of the population but held 28% of the seats in Congress. In 1960, the last census before the Supreme Court issued its one-man one-vote rulings, counties with populations under 25,000, representing only 13% of the population still held 23% of the seats in congress. The greatest extreme was in the case of California’s urban Twentieth District with a population 5.75 times that of Michigan’s rural Twelfth (Schubert 1965).

Gerrymandering uses two basic techniques in the United States. “Cracking” aims to waste the other party’s vote by dividing it among several districts. “Packing” aims to waste the other party’s votes by putting most of their voters into the smallest possible number of districts, so that they win those districts by great margins but lose everywhere else. The basic data used is voter turnout by party at the precinct level, with allowances made for factors such as changing party registration and migration.

Criticism of gerrymandering is as old as the practice itself and has led to seven “good government criteria” of varying levels of acceptance for electoral districts (Cain 1984; Liphart 1982).

Three Supreme Court cases (Baker v. Carr 1962, Reynolds v. Simms 1964 and Wesberry v. Sanders 1964) established the first criterion, “one man, one vote.” If district “A” has twice as many people as district “B,” then in effect, everyone in district “B” has two votes compared to those in district “A.” Thus, districts should be equal in population. This is difficult in practice if for no other reason than Americans are quite mobile, more so in some parts of the country than others. Sensibly enough, the Supreme Court demands only “substantial equality,” which apparently means a variation of less than 10% without further inspection that might justify greater variance.

The second criterion requires all seats to be competitive. In practice, this means that both major parties should have at least a theoretical chance to win every seat. We have a long way to go. The redistricting in California under Governor Reagan that made about a quarter of the districts competitive is the closest ever to meeting this criterion.

Third, districts should be contiguous and compact.

Fourth, electoral districts should coincide with local political boundaries.

Fifth, some suggest that there should be proportionality between the share of seats won by any particular minority group and its vote share, although minorities should not be able to elect majorities.

The sixth criterion is preserving communities with common interests, generally interpreted to mean similar economic and environmental circumstances.

Seventh, a party’s share of the seats should be proportional to its vote share.

Districting that meets all seven criteria is unlikely in the real world. People living along the long narrow California coast north of Santa Barbara have common interests but gathering enough of them into a single congressional district violates the compact and contiguous criterion. In the United States, the one-man one-vote rule of the Supreme Court would largely resolve this conflict in favor of the equal population principle had it not been for the Court also redistricting to create ethnic minority districts (Liphart 1982).

Our geographic system under-represents dispersed minorities such as California’s Armenians while giving undue power to concentrated ones such as blacks in South Central Los Angeles. Drawing district boundaries to achieve representation for one minority may adversely affect the chances for another minority living in the same area (Wells 1978). In the US, the reality is single party domination of many states, which makes it impossible for all electoral districts in a state to be competitive, particularly as about a third of all registered voters are independents or registered in one of the many minor parties. Finally, some raise the question of whether districting should be based on population including illegal aliens, citizens and legal aliens, citizens, or registered voters.

Even if the criteria, or at least the ones accepted by a particular state, were compatible, there are problems in how to go about drawing district boundaries. One proposal is a variant of Bram’s suggestion for dividing cake (Chapter 2). One party makes an initial redistricting proposal for the state (or city or water district or whatever is involved). The other party can select and freeze the boundaries of one district, then redraw the remaining districts.

Another idea is to program a computer to implement the good government criteria. Clark (2004) simulated this approach for California using data on party voting by census tract. He reports, “Maximally compact districts produced more districts with the party votes within one or two percent than the actual 1992 districts. At three percent difference, the two sets of districts were equal. For differences up to ten percent, there were more in the 1992 set, but for differences of thirteen, fourteen, or fifteen percent, there were more in the compact set… Another surprise was that maximally compact districts would not have changed the outcome of the election. There were still thirty-three Democrat to nineteen Republican districts. My conclusion is that maximally compact districts are not a threat to either party. We can eliminate gerrymandering without a political upheaval” (Clark, personal correspondence). But, this conclusion is based on a single state at a single point in time, so requires considerably more analysis along the lines carried out by Clark.

When discussing election systems, we tend to assume one representative for each district. It is not the only possibility; it is not even the only system in the US. All states have two at-large senators, and at-large or multimember elections are common for county commissioners, city councilors, judgeships, and school board members. Often, they are at least theoretically non-partisan. In the most common system, candidates choose and run for a particular seat but all voters vote for every seat.

Some have proposed multimember districts as a way to increase the pool of talent from which potential representatives are drawn. Multimember districts are not as efficient as single-member districts when responsibilities other than voting, particularly constituent services, are taken into account (Groffman 1982). The effects on policy are complex, little understood, and potentially more advantageous to politicians than to constituents. As districts become larger, campaigns rely less on personal contact and door-to-door campaigns and more on media, rallies, party organization, and endorsements, raising the costs of election. Name recognition has more importance, increasing incumbent advantages (Jewell 1982). Voters are unlikely to follow large numbers of races closely. At-large statewide elections probably discriminate against minorities and probably would over-represent cities while leaving rural voters poorly represented.

Might all of the difficulties disappear by freeing representation from geographical restraints? In 1960, newly independent Cyprus chose a Parliament that was to be 70% Greek Cypriot and 30% Turkish Cypriot. Six election districts each chose the appropriate proportion of representatives by ethnicity (Nohlen 1978). Lebanon provides an example based on religion, with six Christian and three Muslim sects proportionately represented. New Zealand (to insure some representation to the small Maori population) and Belgium (divided among Flemish and French speakers) have tried electoral systems based on ethnicity. As detailed in Chapter 8, many countries require representation for women. These systems guarantee, rather than merely increase the likelihood, of minority representation—but limit the types of minorities guaranteed representation.

It is difficult to envision non-geographic districts gaining acceptance in the United States. First, ethnicity is not the only important minority status (Liphart 1982). Second, there is the problem of how large a minority has to be to warrant representation. Inevitably, incompatible ethnicities would end up lumped together in the category “other.” Third, a substantial and increasing number of people are multi-ethnic. Finally, many Americans would object to registering by ethnicity or religion, while others claim false ethnic identities to obtain advantage from affirmative action programs.

Proportional representation assigns legislative seats from party lists based on the percent of the total vote received. It is the method in most Western Democracies other than Canada, Great Britain, New Zealand, and the US. It could solve the fundamental incompatibility between plurality rule and geographical districts if multiplying a party’s percentage of the total vote by the number of seats available always resulted in a whole number. In fact, under- or over-representation is inevitable. The usual solution, the largest remainders method, is to give each party the whole number of seats to which it is entitled, then to give the remaining seats to those parties or states with the highest fractions (Liphart). In some systems, the party picks the individuals to fill the seats awarded.

Finally, West Germany during the Cold War devised a system that combined proportional and single-member representation. The Bundestag consisted of 496 representatives from 248 districts. Each voter first voted for an individual, then for a party. The individual votes were counted within the district; the party votes were counted nationwide. The seats won by pluralities in the 248 districts were subtracted from the seat totals assigned to the parties, insuring that popularly elected representatives were seated. The remaining seats were filled from party lists in proportion to the total vote, subject to the requirement that a party had to receive 5% of the vote to earn a single seat (Liphart).

Clearly, there are many ways to be democratic. The strengths of one method often compensate for the weaknesses of others, and vice versa. Perfect systems are likely to remain the stuff of fairy tales.