The federal Voting Rights Act says that members of a minority group should not �have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice� (42 U.S.C. � 1973). The Act does not specify any particular remedy for a proven breach of this standard, but over the years civil rights plaintiffs have asked for and received singe-member districts in most cases where blacks or Hispanics have not been able to elect candidates at-large.

In light of the recent attacks on single-member district remedies¹, those interested in preserving the ability of minority groups to elect candidates of their choice must ask themselves whether there are any other remedies that might meet that goal as well as, or better than, a single-member district plan. This article is about those alternatives.

In this article we compare three forms of modified at-large elections�choice voting, cumulative voting, and member plans to judge the ability of each to meet the goal established by the Voting Rights Act. �Modified at-large elections� is a term used to describe elections with multi-member districts but without the winner-take-all feature usually found in most multi-member elections.

Most democracies use modified at-large systems that result in nearly all votes counting toward representation. Although many of these systems allocate votes among parties rather than candidates, the following three systems are based on voting for candidates and are already used in some local elections in the United States. They require at least some seats to be elected from districts with more than one representative.

WHY SINGLE-MEMBER DISTRICTS ARE USED

In several cases early in the �Reapportionment Revolution�, the U.S. Supreme Court approved plans with multi-member districts and specifically held that multi-member districts are not unconstitutional per se (Fortson v. Dorsey, 379 U.S. 433 (1965), Burns v. Richardson, 384 U.S. 73 (1966)). However, in a Mississippi legislative redistricting case, the district court had mixed single-member districts and multi-member districts in the legislative plan, but the plaintiffs presented strong evidence that multi-member districts were dilutive of black voting strength (Connor v. Johnson, 403 U.S. 670 (1971)). One of the lawyers for the plaintiffs in Connor has explained the argument they made as follows:

The lawyers for the [Connor v. Johnson] plaintiffs had made a strong case that multi-member districts in Hinds County [Mississippi] were racially discriminatory�[T]he Supreme Court could order single-member districts without departing from its repeatedly expressed position that multi-member districts are not per se unconstitutional or from the line of precedent rejecting Fourteenth Amendment challenges to such districts (Frank Parker, Black Votes Count: Political Empowerment in Mississippi after 1965, 1990: 114).

The district court had agreed that �it would be ideal� if single-member districts could be used, but decided that time was too short to order such a plan. On an expedited appeal, the U.S. Supreme Court held, �We agree that when district courts are forced to fashion apportionment plans, single-member districts are preferable to large multi-district as a general matter� (Connor v. Johnson, 402 U.S. 690, 692 (1971)). By adopting single-member districts as the presumptive standard, the Court was following a trend in American politics² and was insulating federal district judges from the charge that, since a multi-member district �allows the majority to defeat the minority on all fronts,�³ it allows the federal judge to pick the eventual majority party or group in the legislature.

When black and Hispanic plaintiffs started bringing vote dilution lawsuits, they generally requested single-member district plans.⁴ In Thornburg v. Gingles, the Supreme Court focused its (and litigants�) thinking on single-member districts even more by defining the proof of voting rights violation in terms that suggested such districts were the standard for comparison. The court held,

These circumstances are necessary preconditions for multimember districts to operate to impair minority voters� ability to elect representatives of their choice for the following reasons. First, the minority group must be able to demonstrate that it is sufficiently large and geographically compact to constitute a majority in a single-member district. Second, the minority group must be able to show that it is politically cohesive. Third, the minority must be able to demonstrate that the white majority votes sufficiently as a bloc to enable it�in the absence of special circumstances such as the minority candidate running unopposed��usually to defeat the minority�s preferred candidate (Thornburg v. Gingles. 478 U.S. 30, 50-51 (1986)).

Once litigants started thinking that proof of a single-member district was necessary to prove a violation, single-member districts were the natural remedy for the violation. As we show below, there are other remedies which meet the statutory standard, if not the Gingles �preconditions.� The three election methods discussed below are all candidate-centered methods of election�that is, they may be used in non-partisan elections as well as partisan elections. In contrast, most European counties use proportional representation systems in which voters vote for a political party (although some systems allow voters to vote directly for candidates).

CUMULATIVE VOTING

In cumulative voting, voters cast as many votes as there are seats. The most common form of cumulative voting allows voters to allocate their votes however they wish, including giving all their votes to their favorite candidate.

Because voters in a minority can assign all of their votes to one candidate, that candidate can win with support from fewer voters than in a traditional at-large election. The �threshold of exclusion� is the lowest percentage of support that ensures a candidate will win no matter what other voters do. With cumulative voting, this threshold is equal to 1+(1/(Seats +1)), which is 20% in a four-seat race and 10% in a nine-seat race.

In practice, a cohesive block of voters can elect a candidate with less than the threshold of exclusion, as majority community voters are unlikely to distribute their votes evenly among a number of candidates equal in number to the number of seats (with cumulative voting) or not to rank any minority-supported voters before some majority-supported candidates (with choice voting).

The equal allocation form of cumulative voting may promote more coalition-building and provide easier voting instructions. Voters simply vote for as many candidates as they like, and votes are allocated equally among these candidates. In a five seat race, for example, a voter choosing two candidates would give each candidate 2.5 votes, while a voter choosing four candidates would give each candidate 1.25 votes. A locality presently using at-large voting in which the voters are voluntarily limiting their choices to less than the full number for whom they may vote could easily change to this system. Most voters would not have to change their voting behavior, but could do so if they wished.

Cumulative voting was used for more than a century to elect the house of representatives in Illinois. In recent years cumulative voting has been adopted to resolve voting rights cases in over fifty localities, including Amarillo (TX), Peoria (IL), and Chilton County (AL). A member of the racial or ethnic minority won office in nearly every subsequent election in which a minority candidate participated.⁵

LIMITED VOTING

Limited voting is similar to traditional at-large elections, except voters may cast fewer votes than the number of seats available. The greater the difference between the number of seats and the number of votes, the greater the opportunities for fair representation of those in the minority. The threshold of exclusion for limited voting is 1 + (Votes/(Votes+Seats)) where �Votes� is the maximum number of votes cast by each voter and �Seats� is the number of seats to be filled in the election. In a race to elect five candidates in which voters had two votes, any cohesive group of voters comprising more than 2/7 of the electorate would be guaranteed to win one seat.

(Why would this group not win two seats, you ask. If the other 5/7 of the voting population divided their votes equally among four candidates, each of the majority�s candidates would win. For instance, assume there are 350 voters. The threshold would be one more than 2/7 of 350, or 101. One hundred cast their votes for minority candidates X and Y, 125 cast votes for A and B, and 125 cast their for C and D. The winners would be A, B, C, D and either X or Y. On the other hand, if the majority group tried to win all five seats, the most votes each of its five candidates could receive would be 100 each.)

Limited voting has been or is used in Rome, New York; Hartford and other Connecticut cities and towns; in Pennsylvania counties; and in several counties in North Carolina and small towns in Alabama.⁶ Limited voting was also used in 12 three-member districts and one for-member district in the British Parliament between 1867 and 1885 (when the three-member districts were abolished); in each case, the voter could cast one vote less than the number of seats. It has also been used to elect the former City Council in Gibraltar (four votes for seven councilors) and has been used since 1968 for the Gibraltar House of Assembly (eight votes for 15 seats). Japan used limited voting (one vote in three-, four-, and five-member constituencies) for electing the Diet, its house of representatives, until 1996 when it converted to another semi-proportional system. Limited voting is used for the election of the Spanish upper house (mostly three votes for four seats in each district).⁷

CHOICE VOTING

Choice voting (also called �preference voting� and �single transferable vote�) is a form of limited voting in which voters have one vote, but can maximize the strength of this vote by ranking candidates in order of preference. Voting is easy with choice voting, but determining the results is more complicated.

To vote, electors simply indicate their first choice candidate, and ar second choice and so one. Voters can rank as many candidates as they wish, knowing that supporting a lower choice may help that lower choice, but will never help defeat a higher choice.

Candidates win by gaining enough votes to guarantee victory. The most commonly used �winning threshold� is the same number as the �threshold of exclusion� in a cumulative voting race; e.g., the fewest number of votes that only can be obtained by winning candidates. Thus, in a race to elect three seats with 100,000 voters, the winning threshold is 25,001, as the most votes that a fourth-place candidate could obtain would be 24,997.

To ensure that as many voters as possible help elect someone and that support cannot be split between similar candidates, ballots are transferred during the ballot-count to the next candidate ranked on the ballot when a higher choice is already elected or has no chance to win. Ballots count for a voter�s highest-ranked candidate who can win with the vote. A sample count in a choice voting election is included in the appendix to this article.

Choice voting is used for national elections in Ireland and Australia and for local elections since 1941 in Cambridge (MA),Blacks make up less than 20%of Cambridge�s population, but have been represented on council since the 1950s and as if 2001 hold two seats on the council and school board. It was used for New York city council elections in the LaGuardia era, Cincinnati from 1925 through 1955, and 21 other American cities, but fell victim anti-reform forces when people of color and leftists won seats (Kathleen Barber, Proportional Representation and Electoral Reform in Ohio 11 ff (1995)).

Choice voting is also used in New York City community school board elections. Unfortunately, many observers have dismissed the system as complicated without trying to understand the fairness of the representation that is has produced�a problem magnified by the fact that most eligible voters have little interest in these elections and that the ballot-count administration is done with very primitive, slow methods. In 1998, the state legislature voted to convert these elections to limited voting, but the Department of Justice refused to preclear the change due to the fair results generated by the choice voting system.

The Asian American Legal Defense and Education Fund in New York City reports that in 1996 11 of 15 Asian American candidates were elected, despite the Asian population being below 20% (usually much lower) in every school district in the city. The report found that �a majority of the Asian American candidates who ran in those electionsn seem in agreement that the proportional representation system of voting works to the advantage of minority candidates and voters�The majority of candidates did not support the notion of abandoning the entire system when many of its weaknesses are clearly due to mistakes of elections officials at specific stages of the election process� (AALDF, Asian American Participation in the 1996 New York City Community School Board Elections).

WINNER-TAKE-ALL ELECTIONS DILUTE MINORITY REPRESENTATION

Multi-member districts typically cause a severe underrepresentation of racial minority groups-perhaps even a complete exclusion of the minority group (Richard L. Engstrom and Michael D. McDonald, �The Effect of At-Large versus District Elections on Racial Representation in U.S. Municipalities,� in Bernard Grofman and Arend Lijphart, eds., Electoral Laws and Their Political Consequences, 1986).

It is an unfortunate fact of life that racially polarized voting exists. Racially polarized voting simply means that two racial groups vote cohesively within each racial group and differently from the other group. The fact that the two groups vote differently and cohesively does not assign blame or even tell us a reason for the difference. Racial polarization may be based on the different political interests of the two groups, or it may be based on their prejudices about the other group. It is really fruitless for election system designers to try to decide why polarization occurs; instead, they should try to alleviate its pernicious effects.

Any �winner-take-all� election will allow racially polarized voting to have an effect. Assume, for instance, that a city had two polarized groups, the Blues and the Greens. If the Blues win 51% of the vote in an at-large election, they will win all the seats. Many cities have adopted single-member district plans as a method of ameliorating the effect of racially polarized voting. If the Blues and Greens were residentially concentrated, it would be a simple matter to draw districts in which one group or the other had a clear majority. For instance, the Blues with 55% of the city�s population could be the majority in three districts, and the Greens in two. This example of the Blues and Greens is not far-fetched. In 1832, the Democrats swept New Jersey�s at-large election for members of the U.S. House by winning with a margin of 24 votes (the lowest Democrat over the highest National Republican); overall the range of votes was 23,257 to 24,383 (The Guide to U.S. Elections, 714, 726, 1985).

In the Blue-Green example, single-member districts provide an adequate remedy because the city has a geographically concentrated minority that votes cohesively. On the other hand, the city may have a racial minority for which one district cannot be drawn. In many cities, for instance, Hispanics live in several areas of the city, but still have a distinctive and cohesive political agenda. Or the minority group may be cohesive in wishing to have its own representation, but different factions would like to have the upper hand in choosing the representative. Let�s assume the city council has agreed to draw one black majority district, but not all blacks can be fitted into the district. Should the district start at the northern end of the back residential area leaving out the southern end�or vice versa? What if the Booker T. Washington faction wants different district boundaries than the W.E.B. DuBois faction? How does the city council make a decision between those plans?

An alternative election system would avoid most of these problems. Choice voting, limited voting, and cumulative voting do as well as, if not better than, single-member districts in ameliorating the effects of racial bloc voting. Rather than drawing districts for each group, these alternative systems allow voters to group themselves into �voluntary constituencies.� Such non-geographic groupings will allow the formation of bi-racial coalitions and racial crossover voting, both of which might have been frustrated by the rigid district lines imposed under single-member districts. Cumulative voting and limited voting give minority groups a chance to have some representation on the council, while choice voting gives each group a chance at fair representation. Because a minority group will want to undertake a cautious strategy of minimizing the votes they receive, cumulative voting and limited voting may only result in only one minority representative. On the other hand, because choice voting allows votes to be transferred according to the wishes of each voter, the minority group does not need to limit the number of its candidates and may be able to win several seats on the council.

Finally, these alternative elections systems may provide female candidates with a greater electoral opportunity than single-member districts do. Studies of various types of electoral systems have shown that women do better in multi-member plans than in single-member district plans.⁸ Generally, the larger the number of members elected from a particular district, the more likely it is that women will be elected. Thus, any system electing several council members in the same election or district will probably elect more women than single-member districts.

MODIFIED AT-LARGE SYSTEMS ARE FAIR TO THE MINORITY

While modified at-large elections give minority groups representation, the majority of the electorate is not short-changed by their use. Winner-take-all inevitably leads to �wasted votes,� which occur whenever a voter gets nothing back for his or her vote. While the number of wasted votes may be diminished by districting, all the people in the minority political group in each district will still be wasting their votes-the election results will not be affected by their participation. Under the example of the Blues and Greens, discussed above, the line drawers usually put some of each group in each district, but make sure that there are enough of one group to make that district �safe� for its candidates. When the city creates a district with a voting majority of some minority group, it generally has to add enough voters from some other group to make the requisite number of people to meet the one-person-one-vote standard. Two law professors have called these added voters �filler people�: ^

These additional individuals must not be of the relevant demographic group (in order to avoid claims of packing [the district to reduce the minority group�s influence in other districts]), and in the interest of minority representation, they should not be expected to compete in any genuine sense for electoral representation in the district to which they are assigned lest they undo the preference given to the specified minority group (T. Alexander Aleinikoff and Samuel Isaacharoff, �Race and Redistricting: Drawing Constitutional Lines after Shaw v. Reno,� 92 Mich. L. Rev., 588, 631, 1993).

By avoiding both districts and winner-take-all elections, alternative election plans eliminate �filler people� and reduce waste votes greatly. For instance, in a choice voting election for five members of a city council, we know that at least 5/6 of the ballots will be counted for winning candidates. Many of the final 1/6 of the voters might have expressed a preference for one or more of the winners, so that the number of wasted votes will be even less than 1/6. Contrast that with the wasted votes in five single-member districts; as many as 49% of the voters in each district could have voted for a losing candidate.

Because of the way limited voting and cumulative voting work, it is much harder to predict the number of potentially wasted votes. An empirical study of the limited voting and cumulative voting elections in Alabama has shown that at least 73% of the voters in limited voting jurisdictions and at least 61% of the voters in cumulative voting jurisdictions have voted for winning candidates.⁹ The number of wasted votes will depend, of course, upon the particular conditions in the election. If there are many candidates with fairly even support in the electorate, the number of wasted votes could fall closer to or even exceed 50% of the electorate.

The will of the majority can also be frustrated by a candidate winning a seat with only a plurality of the vote. While most medium sized and large cities have runoffs, they are less common in county and state elections outside the South.¹⁰ If several such candidates in a multi-member district election are all supported by the same group of voters, that majority of the voters can win a majority of the council or other legislative body. Similarly, in single-member district elections, one group could win a majority of the seats with only a majority of the vote in a majority of the districts and no other votes at all-again, a plurality of the voters would have elected a majority of the council. For instance, consider the results in the 1993 Canadian election:

In the election, Canadians clearly wanted to show the ruling Progressive Conservatives that they had lost confidence in them, and the party won only 16% of the popular vote. However, the workings of the voting system turned a show of non-confidence into a massacre. Rather than electing 46 of the 295 members that a proportional system would have provided, the Tories elected only two. By contrast, two regionally-based parties, the Bloc Quebecois and Reform, with 13% and 19% of the popular vote respectively, elected 54 and 52 Members. The voting system also turned the victorious Liberals� 41% of the vote into a very solid majority of 177 seats (Henry Milner, �Prospects for Electoral Reform in Canada,� Voting and Democracy Report 1995, 159, Washington: Center for Voting a Democracy, 1995).

The results of the 1997 Canadian elections were even more skewed. The Bloc Quebecois won only 37.9% of the vote in Quebec, but captured a comfortable majority of the province�s delegation and will be the third largest party in parliament despite not fielding any candidates outside of Quebec. Meanwhile, the second largest party in parliament, the Reform party, has no patience with Quebec�s separatists and won no seats east of the western province of Manitoba. To add to the distortions, the governing Liberals also won two-thirds of their seats in only one province�Ontario, where their 49.5% of the vote won an astounding 101 of 103 seats.¹¹ The results in Canada show the distortions that can occur when the people spread their votes around among several parties, but the election system is still winner-take-all.

Limited voting and cumulative voting elections could also allow a minority of voters to control a majority of the council. For instance, three of five council seats might be won by three different �single issue� groups if each had the support of about 15% of the electorate, and the remainder of the electorate was fragmented in its choices for the council.

Minority control is extremely unlikely under choice voting. Assume that a particular group with about 52% of the voters endorses five candidates for the five-member city council. If all the voters in the group case their preferences for all five candidates of the group�in any order�the group will win three seats, since it has enough voters to fill three threshold (each threshold would be 1/6 + 1, so three seats would be elected by 3/6 + 3 of all the votes).

MODIFIED AT-LARGE ELECTIONS MAY INCREASE TURNOUT

Several political science and legal commentators have contended that single-member districts depress voter turnout because of �incumbency lock.� In a study of Latino politics, two political scientists said,

Thus, in these [barrio] districts, incumbents have few electoral incentives to mobilize new voters; moreover, they are not indebted to their party for their office, and they have no reason to seek party support to win re-election. These districts are also safe districts for the incumbent�s party. Thus, neither the incumbent nor the party is likely to try to mobilize voters in these districts. The design of these districts, therefore, may effectively eliminate the party�s need to mobilize the grassroots (Rodolpho O. de la Garza and Louis DeSipio, �Overview: The Link Between Individuals and Electoral Institutions in five Latino Neighborhoods,� in de la Garza, et al., eds., Barrio Ballots: Latino Politics in the 1990 Elections, 1994).

Professor Lani Guinier has taken her criticism a step earlier in time�to the creation of the district.

[D]istricting decisions may simply reflect the arbitrary preferences of incumbent politicians who prefer packed, safe districts to ensure their reelection. Indeed, districting battles are often pitched between incumbents fighting to retain their seats, without regard to issues of voter representation. Because the choice of districts is so arbitrary, incumbents enjoy extraordinary leverage in self-perpetuation through gerrymandering.

Thus, districting strategies often promote noncompetitive election contests, which further reduce voter participation and interest (Lani Guinier, �No Two Seats: The Elusive Quest for Political Equality,� 77 Va. L. Rev. 1413, 1451-52, 1454-55, 1991 (footnotes omitted)).

Since alternative election systems do not use districts, they avoid the problems of incumbency lock and depressed voter turnout associated with districts. In limited voting, cumulative voting, or choice voting elections, all candidates�incumbents and challengers alike�are running against each other. No incumbent has a �free ride,� but must instead rally the faithful behind his or her candidacy.

In a study of municipal elections in four Massachusetts cities, George Pillsbury found that voter turnout had declined in all four, but less in Cambridge (which uses choice voting) than in the other three (which use plurality voting). Pillsbury concluded, ��the evidence certainly points to choice voting providing more incentive for voters to participate than plurality elections� (George Pillsbury, �Preference Voting and Voter Turnout,� Voting and Democracy Report 1995, 79-80 (Washington: Center for Voting and Democracy, 1995)).¹² Arend Lijphart has explained the connection between the voting system and turnout in this way:

Proportional representation (PR) tends to simulate voter participation by giving the voters more choices and by eliminating the problem of wasted votes�from which systems using single-member districts suffer, this makes it more attractive for individuals to cast their votes and for parties to mobilize voters even in areas of the country in which they are weak. Recent comparative studies have estimated that the turnout boost from PR is somewhere between 9 and 12% (Arend Lijphart, �Unequal Participation: Democracy�s Unresolved Dilemma,� 91, American Political Science Review, 1.7, 1997).

GERRYMANDERING IS NOT A PROBLEM WITH MODIFIED AT-LARGE

Gerrymandering is defined as �political manipulation within the process of drawing district boundaries� (Michael D. McDonald and Richard L. Engstrom, �Detecting Gerrymandering,� in Bernard Grofman, ed., Political Gerrymandering and the Courts 182 (1990)). In essence, it is the effort by one group to force a disfavored group to waste its votes. Methods of gerrymandering include packing, stacking, and cracking. Cracking is splitting a political group�s population concentration to prevent it from having a majority in a district. Stacking occurs when a group numerous enough to elect one representative is combined with an opposing group in a multi-member district. Packing is the over concentration of a group in one or a few districts to prevent its members from having an impact in adjoining districts (Gus Tyler, �Court versus Legislature: The Sociopolitics of Malapportionment,� 27 Law and Contemporary Problems, 390, 400, 1962).

By definition, gerrymandering can only occur if there are districts. Therefore, eliminating districts eliminates gerrymanders. Any of the non-single-member districts election systems may be run from the city at-large or from smaller, multi-member districts. A city electing fewer than a dozen members of a city council can easily run an election in the whole city. The alternative election systems have the advantage of completely eliminating the cost of redistricting.

INTER-CENSUS MALAPPORTIONMENT IS ALSO AVOIDED

A city can spend a lot of money on a districting plan (including perhaps on a court challenge to the plan), but the plan will be malapportioned as soon as there are significant changes in the population of the city. For instance, the annexation of territory with occupied houses, the creation of new subdivisions or residential apartments in the city, the movement of people out of a decaying neighborhood�all these can cause changes in the population of the various districts. Even before the city reaches the next census, its districts can be malapportioned.

Any system without districts will avoid this creeping malapportionment problem.

CHOICE VOTING MAY REDUCE THE COST OF CAMPAIGNS AND ELECTION ADMINISTRATION

There are several factors that may affect the cost of campaigns for the candidates:

� The absolute number of people who are potential voters for a particular office. If the candidate wants to use direct mail or telephone banks to reach 5,000 voters, the cost will be less than reaching 50,000 voters by the same method. Similarly, a smaller district size may allow the candidate to make personal contact with all the voters, while she will have to use media or surrogates to contact everyone in a larger district.

�  The number of votes the candidate expects to need to win. In a winner-take-all election, this will be one-half of all the votes. This is true if there are three or more candidates. Since it is possible that some candidates could receive negligible numbers of votes. On the other hand, in the alternative election systems the usual number of votes needed is 1+(1/(Seats +1)). Thus, if candidates are able to target direct mail, phone bank, or media campaigns, they may be able to spend their money more effectively on voters who are likely to provide them a margin of victory.

� The number of campaigns the candidate must endure to win office. For instance, a candidate will have to spend more money to run in a partisan primary, a possible runoff, and the general election than on a nonpartisan election with no runoff. These three alternative systems have the advantage that they can eliminate the necessity of having (general election or primary) runoffs, and choice voting eliminates the necessity of partisan nominating primaries.

� Joint political advertisements. If the candidate must run a media campaign that will reach many voters who are not in her district, she might be able to share the cost with other candidates by running joint ads�for example, �Vote for the Cost-Cutting Team of Smith (District 1), Katzenbach (District 2), and Paglia (District 3).�These joint ads will be more effective if all the candidates listed are running in the same multi-member district, as they would be under multi-member districts, limited voting, cumulative voting, or choice voting.

� Eliminating �swing voter� support. Candidates can win under an alternative system with the support of voters already inclined to support the candidate rather than needing the support of �swing voters� who by definition do not like either candidate. A disproportionate amount of money in single-member district elections must be spent on either winning the support of these swing voters or keeping them from the polls with a negative campaign.

While there will be some tradeoffs in costs, a choice voting election campaign is likely to be more cost-efficient for the candidate. The candidate can target potential supporters wherever they live in the jurisdiction and not have to run multiple election campaigns.

Just as it costs more for the candidates to run in two or three elections (primary, general, runoff), the cost to the jurisdiction is greater also. By holding only one election, the jurisdiction avoids the cost of polling officials and election supplies (and in some places the rental cost of the polling places) for a second or third election.

In the past, one of the complaints about choice voting has been that the counting of the ballots took too long. For instance, in Cambridge, Massachusetts, ballot-counting typically took four days for the city council race However, the City since 1997 has computerized its ballot, using optical scan ballots and software from Voting Solutions, that generate unofficial results on election nights and final results within 24 hours. In terms of simplicity of marking the ballot, the �X� voting (winner-take-all) systems are likely to take the prize. But for whom should individuals cast their ballot? If the voters are Perot supporters in 1996, should they stick with Perot and probably throw away their vote, or make a choice between Clinton and Dole? If voters are pretty sure Perot will lose, they might switch to the least objectionable second choice.

In choice voting elections the voter has only to mark first, second, third,�(and so on) choices. Here an individual could vote for Perot first and Clinton second with the assurance that the vote for Perot would �send those politicians a message� but still allow the second vote to have an effect in choosing the winner. (Such an �instant runoff� system in fact is used in several countries and is under consideration in a growing number of American ciites and states.)

Cumulative voting voters are likely to suffer from the agonies of strategic voting, just as the winner-take-all voters do. If individuals have five votes, but think that their party can probably get only one or two candidates elected, should they call all their votes for one candidate or divide them among two candidates? If they divide their votes, which candidate gets three or four votes and which only one or two? All of these calculations have to be made in trying to figure out what like-minded voters and the opposing voters are going to do. In short, the voter has to play a three-dimensional video game with a constantly moving target.

Limited voting voters may have to make the same sort of strategic choice if there is more than one attractive candidate. Voting for a losing candidate is a wasted vote, and it is especially wasted if there was another candidate who could have been helped by the votes received by the loser.

WHAT SYSTEM IS �THE BEST�?

As Bernard Grofman and Shaun Bowler have recently argued, the �best� system can vary depending on the questions one asks. They compared party list proportional representation (the typical European system), the alternative systems discussed in this article, and winner-take-all by looking at whether they encourage candidate centered or party-centered politics, encourage a parochial or jurisdiction-wide outlook by candidates, make it difficult for voters to make effective votes, and encourage coalition building (Bernard Grofman and Shaun Bowler, �STV�s Place in the Family of Electoral Systems: The Theoretical Comparisons and Contrasts,� 34, Representation, 43, 1997).

Each of the alternatives discussed in this article are candidate-centered, as are traditional multi- and single-member district plans.

While single-member district plans encourage each candidate and office holder to focus on �bringing home the bacon� to constituents, multi-member plans and alternative systems generally encourage a view that encompasses the whole geographic area of the jurisdiction-although alternative systems ensure that any minority group of significant size with pressing geographic interest can elect a candidate of its choice.

Voters will have the easiest time in causing an effective vote in choice voting because they may rank the candidates and are assured that their ballot will be counted for one of their choices. Wasted votes are still a problem in Limited and cumulative voting, although it is mostly self-imposed rather than the result of a gerrymander. In those two systems, voters and candidates organizations may misjudge the potential popularity of candidates so that there are too many candidates vying for the votes of like-minded voters. As already noted, wasted votes are endemic to traditional winner-take-all plans.

Just as voters may have difficulty in casting effective votes under cumulative voting, they may also be discouraged from forming coalitions by fear that they may be diluting their votes for high-preference candidates by sharing them with coalition partners. In limited voting, the fixed number of votes for each elector will obviate the coalition problem, but the effective vote problem remains. Choice voting allows voters the assurance that their ballots will count for their first choices if needed by the first choice, but the second choice if need by that candidate, and so on.