Rethinking Redistricting: How Drawing Uncompetitive Districts Eliminates Gerrymanders, Enhances Representation, and Improves Attitudes Toward Congress

Home , Electoral competition, Safe seat

Rethinking Redistricting: How Drawing Uncompetitive Districts Eliminates Gerrymanders, Enhances Representation, and Improves Attitudes toward Congress

Thomas L. Brunell, University of Texas at Dallas

Introduction Winners and Losers ning candidate in a House election will be systematically more likely to have In every contested election there are A developing literature suggests that higher evaluations of said candidate rela- inevitably winners and losers, both voters’ evaluations of government, in- tive to voters who vote for the losing among the candidates and among the cluding overall trust in government, is candidate. Similarly, “winning” voters voters. Some candidates will take their directly related to whether they cast a will have more positive and fewer nega- seats as elected officials, and others will ballot for the winning candidate. Ander- tive things to say about their representa- not. Some voters will be happy with the son and LoTempio ~2002! show that tive than “losing” voters. I test this outcome, others will not. Here I seek to citizens who voted for the winning presi- proposition using the American National better understand the relationship be- dential candidate have significantly Election Study ~ANES! cumulative file tween whether a voter casts a ballot for higher levels of overall trust in govern- with survey data from 1948–2000, al- the winning candidate in U.S. House ment relative to voters whose candidate though the multivariate analysis only elections and that voter’s evaluations of lost the election ~even after controlling includes data from 1980 on, as some of her representative. I build on a burgeon- for other factors that affect trust!. Thus, the variables utilized in the analysis were ing literature on the relationship between evaluations of the government depend, in not part of the survey until that time. voters and their elected governments to part, on election outcomes. However, derive and test a theory about this con- Anderson and LoTempio find no rela- nection. The data will show that voters tionship between votes for congressional Empirical Results whose preferred candidate wins a seat in candidates and overall trust in govern- the House of Representatives are system- ment, although this is likely because vot- Using these data I test to see what atically happier with their representative ers are more likely to think about the effects voting for the winning candidate than those voters whom did not vote for president than Congress when asked how in a U.S. House of Representatives elec- the winning candidate. While this finding much trust they have in the federal gov- tion has on evaluations of that represen- is not especially groundbreaking, the im- ernment.2 Clarke and Acock ~1989! show tative, as well as voter evaluations of plications for the way in which we draw that voting for the winning candidate in Congress as an institution. First, I exam- congressional and state legislative district American elections increases voter effi- ine the relationship between which can- lines are quite provocative. Specifically, cacy as well. didate a respondent has voted for and since district lines in the House are nec- These results are not unique to the that voter’s affect toward the represen- essarily an artificial construct, I argue American case. Anderson and Guillory tative. In the National Election study, that map makers ought to “pack” dis- ~1997! demonstrate that a similar rela- respondents are asked the following tricts with as many like-minded partisans tionship exits in other advanced indus- open-ended question: “Is there anything as possible. Trying to draw “competitive trial democracies. Clarke and Kornberg in particular that you liked about @U.S. districts”1 effectively cracks ideologi- ~1992! show that winning voters in Can- House incumbent candidate# What is cally congruent voters into separate dis- ada have more positive evaluations of that? Anything else?” They are also tricts, which has the effect of increasing their members of parliament in terms of asked if there is anything that they dis- the absolute number of voters who will responsiveness to voters. This “winning like about the incumbent representative. be unhappy with the outcome and dissat- effect” extends not only to voters, but The survey records up to four responses isfied with their representative. I discuss also to the candidates who stand for elec- for both likes and dislikes. Affect for a the benefits of fundamentally rethinking tion. Bowler and Donovan ~2002! dem- candidate is simply measured as the the way in which we draw congressional onstrate that elites’ attitudes toward number of likes minus the number of and state legislative districts, as well as electoral institutions are, in part, depen- dislikes. So if a voter has four positive address likely concerns that might be dent on whether they win the election. and one negative things to say, the result ϩ raised about drawing districts this way. Winning candidates, who have been de- is an affect of 3. This variable ranges ϩ Ϫ livered to parliament by the current elec- from 4to 4. Table 1 presents the toral arrangement, are much more results of a cross tabulation of affect for satisfied and committed to these institu- the incumbent and whether the respon- tions than are losing candidates. dent had voted for this candidate or an opponent ~same party, different party!. Thomas L. Brunell is associate professor of political science in the school of social Clearly, the results indicate that when sciences at the University of Texas at Dallas. Theory and Data people vote for the winner, their affect His research interests include Congress, for their representative is significantly elections, redistricting, political parties, and The theory driving this investigation is more positive. The percentage of people interest groups. simple: citizens who vote for the win- who voted for the losing candidate and

PSOnline www.apsanet.org 77 cumbent, while fully 21.9% of losing Table 1 voters disapproved of the incumbent rep- Relationship between Voting for Winning Candidate and Likes resentative. Over 82% of winning voters Minus Dislikes for the Incumbent Representative express approval of the elected official. These results are not surprising—people Incumbent Affect who vote for the winning candidate are (likes-dislikes) Losing Voter Winning Voter Total happier than people who voted for the −4 62 7 69 losing candidate—and they fit with much 1.9% 0.1% 0.8% of the other literature on the subject of −3 97 23 120 how attitudes toward government are conditioned by how people cast their 2.9 0.4 1.4 ballots. People are more satisfied with −2 218 79 297 the government ~or its component parts! 6.6 1.5 3.4 when the candidates that they vote for −1 363 175 538 are elected. 10.9 3.3 6.2 The relationship between winning and 0 1,241 1,556 2,797 losing also moves beyond how a voter 37.3 29.3 32.4 feels toward a specific representative. 1 528 1,204 1,732 Election outcomes also influence the 15.9 22.7 20.0 ways in which citizens connect with 2 457 1,104 1,561 Congress as an institution. For instance, 13.7 20.8 18.1 Table 4 is a cross-tabulation of a 3 217 682 899 respondent’s overall rating of Congress 6.5 12.8 10.4 and whether the person they voted for in 4 143 486 629 the House election won. Roughly the same percentage of people rate Congress 4.3 9.1 7.3 as doing a “very poor job” in the win- Total 3,326 5,316 8,642 ning and losing voter columns, so even 100 100 100 among voters whose candidate is victori- Entries represent the number of respondents from the cumulative American Na- ous a substantial number of people re- tional Election Study file 1948–2000 (study no. 8475) who answered questions main highly skeptical of our legislature. about the number of likes and dislikes they have about their incumbent Represen- However, more losing voters rate Con- tative, column percentages below entries. The overall affect is simply the number gress as doing a “poor job” and more of likes (up to 4) minus the number of dislikes (up to 4). Chi-squared = 913.3, p < winning voters rate Congress as doing a .001. Losing voters are those who reported voting for the candidate in the House “good job.” While the differences are not election that lost, and winning voters are those who reported voting for the win- as large as those we saw in previous ning House candidate. tables, they are statistically significant and any increase in satisfaction with our had more negative than positive things to say ~i.e., their affect score is negative! is Table 2 22.3%; that same measure for winning Explanatory Model for Affect toward the Incumbent voters is only 5.3%. Similarly, voters for the winning candidate are far more likely Independent Variable Coeff. SE to have more positive things to say about their incumbent representative than are Constant −1.15 0.218 losing voters. Over 65% of the people Age (higher = older) 0.013 0.002 that voted for the winner had more posi- Gender (0 = female) 0.065 0.05 tive than negative things to say.3 Table 2 South (1 = from Southern state) 0.052 0.058 presents the multivariate analysis with Education (higher = more education) 0.062 0.019 the affect variable as the dependent vari- Income (higher = higher income) −0.061 0.028 able. Even after controlling for other Party Id (1 = strong Dem, 7 = strong Rep) 0.042 0.012 variables that influence how a voter Congress Thermometer (higher = positive evaluations) 0.009 0.002 might feel about his or her incumbent Federal Govt. Thermometer (higher = positive evaluation) 0.0002 0.002 representative, the “Vote for Winner” National Economy (higher = economy doing better) 0.024 0.035 variable is positive and statistically sig- Personal Financial (higher = better financial) −0.010 0.035 nificant at beyond the .001 level. Other Vote for Winner (1 = Resp. voted for winner) 0.89 0.054 variables that influence the number of N 3,429 likes and dislikes include age, education, 2 income, party identification, and re- Adjusted R 0.11 sponses to the congressional thermometer The dependent variable is the number of likes (up to 4) minus the number of dis- question. likes (up to four) with respect to the incumbent Representative. Entries are unstan- Table 3 displays the relationship be- dardized OLS regression estimates. Data are from the American National Election tween voting for the winner and voter Study Cumulative File 1948–2000. Bold entries are statistically significant at p < approval of the incumbent. Only 4.8% of .05 or better. winning voters disapproved of the in-

78 PS January 2006 The greater the distance between the rep- Table 3 resentative and the voter, the more likely Relationship between Voting for the Winning Candidate and the voter will be dissatisfied.5 Approval of Incumbent

Approve of House Implications Incumbent Losing Voter Winning Voter Total Based on these results, and since con- Approve 2,033 4,447 6,480 gressional districts are necessarily artifi- 59.5% 82.4% 73.5% cial constructs, there are clearly reasons Disapprove 750 260 1,010 to draw districts in such a manner as to 21.9 4.8 11.5 increase efficacy and happiness with our Don’t Know 635 688 1,323 government, particularly since Congress 18.6 12.8 15.0 almost always has significantly lower Total 3,418 5,395 8,813 approval ratings than the president, the 100 100 100 Supreme Court, and state governments ~see Hibbing and Smith 2001; Hibbing Entries represent the number of respondents from the cumulative American Na- and Theiss-Morse 1995!. Packing dis- tional Election Study file 1948–2000 (study no. 8475) who either approve or disap- tricts with like-minded partisans makes a prove of the incumbent Representative, column percentages below entries. Chi- great deal of sense, as long as both squared = 665.8, p < .001. Losing voters are those who reported voting for the major political parties are packed to sim- candidate in the House election that lost, and winning voters are those who re- ilar degrees. Drawing competitive dis- ported voting for the winning House candidate. tricts or systematically “cracking”6 one party or the other is not beneficial and ought not be present in redistricting plans. Thus, what we think of typically representative institutions is surely a pos- didate won chose Congress as the most as competitive districts ~those with itive development. trusted branch of the federal government. roughly equal numbers of Democrats and The relationship between voting for Thus, winning voters are much more Republicans! provide negligible benefits the winning candidate and which branch likely than losing voters to choose Con- and come with significant costs. One of of government the respondent trusts most gress as their most trusted branch of the 4 the most significant benefits from draw- ~Congress, the Supreme Court, the presi- federal government. ing a legislative map with packed dis- dent, or political parties! is presented in People who vote for the winning can- tricts is that it makes it significantly Table 5. The biggest difference between didate are systematically more satisfied more difficult to effect a map that consti- winners and losers is related to voting with their representative and with Con- tutes a partisan gerrymander. Districting for the winning House candidate. More gress as a whole. But the underlying plans that dilute one party’s vote must than 23% of the voters on the losing side story is not merely one of winning or use a combination of packing and crack- versus nearly 32% of voters whose can- losing, but rather ideological distance. ing to create a map that contains significant levels of partisan bias ~i.e., where one party might win the statewide vote for the House, but still end up with Table 4 fewer seats!. Relationship between Voting for the Winning Candidate and Preserving communities of interest is Approval of Congress one of the main principles guiding map makers. Among these principles, pre- Performance of serving communities of interest is cer- Congress Rating Losing Voter Winning Voter Total tainly the most ephemeral. What really constitutes a community of interest? Is Very poor job 389 532 921 any American city or county, really a 31.4% 30.3% 30.8% community of interest? Sometimes— Poor Job 289 333 622 depending on the issue. A city is a uni- 23.3 19.0 20.8 fied community of interest when the Fair Job 443 657 1,100 issue is non-partisan, for example if the 35.8 37.5 36.8 issue is obtainment of federal funds to Good Job 108 213 321 repair bridges and roads, but it is not if 8.7 12.2 10.7 the issue involves anything with ideo- Very Good Job 9 19 28 logical disagreement. Properly con- 0.7 1.1 0.9 ceived, I argue, communities of interest Total 1,238 1,754 2,992 should be composed entirely of either 100 100 100 Democrats or Republicans ~liberals or Entries represent the number of respondents from the cumulative American Na- conservatives! in reflection of the pri- tional Election Study file 1948–2000 (study no. 8475) who indicated how they rate mary cleavage in American politics the job that Congress is doing, column percentages below entries. Chi-squared = ~Poole and Rosenthal 1991!. On contro- 16.4, p < .003. In the original dataset there are nine valid responses to the de- versial issues, ideological communities pendent variable, here they have been collapsed. Losing voters are those who re- of interest will typically face issues in ported voting for the candidate in the House election that lost, and winning voters unison, increasing the likelihood that are those who reported voting for the winning House candidate. their representative will vote on legislation in Congress in congruence with the

PSOnline www.apsanet.org 79 voters through the threat of a primary Table 5 challenge. Minimizing the ideological Relationship between Voting for the Winning Candidate and distance between the average voter and Which Branch of Government a Citizen Trusts the Most the likely representative fosters more effective representation and makes it eas- Branch Respondent ier for elected officials to discern our Trusts the Most Losing Voter Winning Voter Total preferences, thus strengthening the bonds Congress 137 245 382 between the represented and representa- 23.2% 31.7% 28.0% tives. Buchler ~2005! develops this idea Supreme Court 267 330 597 more formally and finds that homo- 45.3 42.6 43.8 geneous non-competitive districts “do a President 170 187 357 better job than competitive districts in 28.8 24.2 26.2 achieving representative outcomes.” Fur- Political Parties 16 12 28 ther, these districts “produce legislators 2.7 1.6 2.1 that are closer to their district medians Total 590 774 1,364 and more representative of everyone in 100 100 100 their district” ~457!. The House of Representatives ought Entries represent the number of respondents from the cumulative American Na- to be the closest of our national institu- tional Election Study file 1948–2000 (study no. 8475) who chose each branch of tions to the people. We want the connec- the federal government that “they trust the most,” column percentages below en- tion between the representative and the tries. Chi-squared (3 df) = 13.99, p < .003. Losing voters are those who reported represented to be a close and faithful voting for the candidate in the House election that lost, and winning voters are connection. The delegate theory of repre- those who reported voting for the winning House candidate. sentation “posits that the representative ought to reflect purposively the preferences of his constituents” ~McCrone and Kuklinski 1979, 278!. Research on the vast majority of their constituents. Citi- Some states, like Arizona,8 have connection between elected officials and zens living in knife-edged districts can- passed laws or referenda specifying that the people that they represent has a long not expect this kind of representation. a districting plan ought to maximize the tradition in our discipline. Miller and Drawing districts to increase competi- number of competitive districts. This is Stokes’s ~1963! seminal piece on this tiveness in the general election only op- not particularly surprising since the com- subject called into question the ability of timizes the number of voters that will mon wisdom among most voters and the representative to know what her con- be upset with their representation. Thus, certainly among the media is that the stituents want, as well as the ability of not only do competitive districts not House of Representatives does not have the constituents to know and understand provide a social “good,” they actually enough competitive districts currently, how the elected official is voting in increase dissatisfaction and make it less and that an increase in the number of Washington. likely that voters’ preferences are repre- competitive elections or in the amount of For example, Miller and Stokes ~56! sented in government. turnover in Congress will somehow en- conclude: “The Representative has very Drawing competitive districts also in- hance representation. There is absolutely imperfect information about the issue creases the volatility of the electoral sys- no evidence that this is the case.9 In fact, preferences of his constituency, and the tem. If every district is a “50–50” maximizing competitive districts is harm- constituency’s awareness of the policy district, then any small change in the ful rather than beneficial in many re- stands of the Representative ordinarily is voting behavior of the electorate could spects. The most obvious effect of slight.” While other scholars have produce enormous changes in the parti- drawing cracked or competitive districts pointed out some methodological con- san distribution of seats. While it is true, is to maximizes the number of voters cerns with the original Miller and Stokes and oft-cited, that the Founders intended who are dissatisfied with their represen- study and have questioned the validity of for the House to most closely mirror the tation.10 Second, a state that draws all or some of their conclusions ~see Erikson, wishes of the public, it is not reasonable most of its districts in this knife-edged Luttbeg, and Holloway 1975; Erikson to assume that they expected the institu- fashion increases the likelihood that 1978!, there is very little doubt that these tion to be hyper-sensitive to relatively small swings in voting behavior translate connections between the representative small changes in the partisan leanings of into large swings in the percentage of and the constituency could be much the people. Districting this way also de- seats that one party can win in a single stronger. Creating packed ideological livers a disproportionate share of the vot- election. This can result in statewide rep- districts will necessarily strengthen these ing power to independents and moderate resentation that is widely incongruent bonds. If a Democrat is sent to Congress swing voters.7 A district that packs voters with statewide partisanship, which, in from a district heavily populated by like- from both major parties equally into dis- turn, leads to voting behavior in the leg- minded partisans, she will have little tricts ~i.e., the average district is either islature that does not accurately represent doubt as to how her constituents prefer 80% Republican or 80% Democratic! has the views of the constituents. her to vote on policy proposals. Simi- the added benefit of sending congressio- From a utilitarian perspective, the larly, creating more “winners” out of nal delegations to the House that closely ideal congressional district is one popu- voters alleviates some of the uncertainty mirror the overall distribution of parti- lated entirely by people with the same with respect to how the representative sans in that state. A plan with many political preferences. Such a constituency votes on their behalf. This will also serve competitive districts can easily send a would see its wishes effectively trans- to encourage representatives to be more delegation to the House that is truly un- lated into votes by their representative responsive to constituent needs. reflective of the underlying partisan divi- and this type of district assures that the Elections serve as the main catalyst sions in a state. representative will remain faithful to the for keeping representatives faithful to

80 PS January 2006 their constituents. The threat of being Figure 1 replaced in a popular election incentiv- The Relationship between Margin of Victory and Ideology izes elected officials to respond to what in the House their constituents want. A districting plan that maximizes the number of winners would clearly affect election outcomes. Most obvious is the likelihood that general election outcomes would not be particularly close. Districts packed with 80% or more of one party will deliver general election victories to the majority party in the district. Redistricting is al- ready the oft-accused suspect of the de- cline of competition and turnover in the House of Representatives ~Cox and Katz 2002; Hirsch 2003! and this type of plan may not help the current public image of this process. However, it is critical to keep in mind that general election competitiveness is neither a necessary nor sufficient condition for ideal representation. First, we utilize a two-stage process for electing members of Congress, where citizens are given the ability to choose, via a primary election, which candidate will represent their party in the general election.11 Therefore, even in an 80–20 district, the incumbent can easily be re- moved if he fails his constituents. This type of districting plan encourages intra- rather than inter-party competition. By no means do homogeneous districts spell an end to electoral competition or responsiveness on the part of our elected officials. Increasing the relevancy of pri- maries may have other beneficial effects: nationalized and election outcomes de- overwhelmingly Republican ~Demo- turnout in primary elections might in- pended heavily upon what the voters in cratic! districts that are more likely to crease, the organized political parties each district thought of their representa- elect very conservative ~liberal! mem- may be more apt to get involved at the tive. “Candidate saliency is a double- bers. This is an empirical question that is primary election stage, and more quali- edged sword for incumbents; while it can easily addressed. The question is whether fied candidates may choose to run mean an enormous advantage in visibil- members who win by large margins are against an incumbent in a primary. An- ity over challengers, it can also spell di- more extreme than their colleagues who other reason that incumbents, even in the saster if the voters come to believe that win by relatively small margins. Figure 1 absence of tough general election compe- their representative has some personal plots member ideology by vote percent- tition, will remain faithful to their con- failing” ~103!. age for the years 1952–2000. The x-axis stituents is that the specter of losing the The lack of turnover in the U.S. Con- represents the Democratic proportion of next election is always prevalent in the gress should not be used as an indication the vote, so very small values indicate minds of incumbents; this implied threat that our federal institutions are failing as Republican victors by very large mar- is what keeps our elected officials components of a democracy. While I am gins. The closer one moves toward the responsive. not arguing that stagnation is the signa- vertical line at .5, the more competitive There exists, I argue, sufficient uncer- ture of a perfectly health democracy, I the elections. The data on the far right tainty among elected officials with re- do not see rapid replacement of incum- side of the graph ~those data points near spect to their ability to win the next bents as a necessary part of a truly repre- 1.0! are districts in which the Democrat election which keeps them from behav- sentative institution. Indeed, if we want won by very large margins. The solid ing in ways that might contribute to a rapid turnover in our legislature, then line is the predicted values from a qua- loss. My argument is not novel. David term limits are a vastly more effective dratic regression and the shaded regions Mayhew wrote in 1974 that: “It is possi- method of producing this outcome than represent the 95% confidence region. If ble to conceive of an assembly in which drawing competitive districts. Many so-called safe seats elected more ideolog- no member ever comes close to losing 50–50 districts end up with an incum- ically extreme members we would expect a seat but in which the need to be bent who can use the perks of the office the predicted lines to slope downward reelected is what inspires members’ be- to leverage what might have been a com- for the Democrats as we move along the havior. It would be an assembly with no petitive district in one election into a x-axis toward 1. The line is more or less saints or fools in it, an assembly packed very safe seat for many successive straight with a slight upward slope with skilled politicians going about their elections. ~driven mainly by uncontested elections!. business” ~37!.InUnsafe at any Margin One common objection to this method For the Republicans we would expect the Thomas Mann ~1978! argued that con- of districting is that it would add to the line to be much higher ~i.e., more congressional elections were becoming less polarization in Congress by creating servative NOMINATE scores! for values

PSOnline www.apsanet.org 81 close to 0 relative to values close to .5. they like and are more likely to highly versity is not a community of interest. The predicted line here is completely value the job that their Representative A county with a significant population flat. Thus, as the margin of victory in- does in the House. These districts will cannot really be thought of as a commu- creases for members to the U.S. House, elect people neither less responsive to nity of interest on any issue other than we seen no real change in the ideology the wishes of the electorate, nor more one that affects the community directly of the members who are elected.12 Thus, extreme in their partisan leanings than in a clearly positive or negative way there is no reason to expect that packing any other configuration of districts. ~i.e., putting a nuclear storage facility districts on the basis of ideology will in the county, or a federal grant to the have any influence on the ideology of Impact on Traditional county to improve roads!. Typically in the House as a whole. Redistricting Principles court, competing maps try to quantify the preservation of communities of Example The decennial task of redrawing dis- interest by counting the number of city trict lines for U.S. House elections is and county splits or segments ~Plan A Imagine a state with sufficient popula- always a bitterly partisan affair. Since splits four of the 26 counties into sepa- tion for two congressional districts. The redrawing the maps is done at the state rate districts, Plan B splits six of the state is perfectly segregated from east to level, district lines depend in large part 26 counties, etc.!. In court, protecting west in terms of partisanship with all upon which party controls the state legis- communities of interest generally de- Democrats living in the western half of lature and the governorship. However, volves into a beauty contest to see the state and all Republicans living in the many districting plans, regardless of the which map can keep as many counties eastern half. Further assume that the partisan composition of the state govern- and municipalities whole. As I argued number of Democrats and Republicans is ment, end up being litigated over a vari- above, communities of interest are better exactly equal. Given the restrictions of ety of issues. thought of in terms of overall ideologi- equal population, contiguity, and com- The first principle used to guide map cal congruence among citizens, rather pactness there are still many different makers is equal population. Courts have than in geographic considerations. Thus, ways in which to cut the state into two often tossed out maps that have even the the overall compactness of districts, at districts. Consider three different slightest deviation in population across times, will be sacrificed if this method scenarios. districts within a state ~the original fed- were used, but again, geography should The first is a line bisecting the state eral district court decision in the Vieth v. not be a sacred cow in the redistricting from east to west creating two 50–50 Pennsylvania is a good example!. This process.13 districts. If a state preferred “competitive stems from the one person-one vote prin- Lani Guinier is a vocal critic of the districts” this would be the most efficient ciple outlined in two 1964 Supreme single-member district system widely method. From the outset, it is non- Court decisions ~Reynolds v. Sims and used in American elections. Guinier and controversial to say that this state should Wesberry v. Sanders!. Packing districts I agree that using geography as one of elect, and is best represented by, one with partisans would not violate this the central principles for drawing dis- Democrat and one Republican. But by guideline. tricts is often harmful rather than benefi- creating two competitive districts this Second, districts must be contiguous, cial. She writes: significantly increases the odds that one which is to say that every point in the party can sweep the state even with just district must touch an adjacent point. Winner-take-all territorial districting im- a very small shift in voting behavior. If Districts cannot be made up piecemeal perfectly distributes representation based the Democrats run a particularly effec- across the state. Again, contiguity could on group attributes and disproportion- tive campaign and swing 1% of the vote easily be preserved with the method out- ately rewards those who win the repre- statewide, this leaves nearly half the state lined in this paper. sentational lottery. Territorial districting with no ideological representation in the Third, districts are supposed to be uses an aggregating rule that inevitably House. Furthermore, even if the election compact. Compactness is, at least in groups people by virtue of some set of outcome delivered one seat to each party, part, in the eye of the beholder. A com- externally observed characteristics such this still leaves roughly half the state pact district tends to be one whose as geographic proximity or racial iden- unsatisfied with their own representative shape is pleasing to the eye, which is tity. In addition, the winner-take-all prin- because of the way in which the districts to say, a reasonably shaped polygon. ciple inevitably wastes some votes. The were drawn. However, oftentimes a district may look dominant group within the district gets Second, we draw two districts, each of like a bug splat on the windshield of all the power; the votes of supporters of which is 75% one party and 25% the a car. Compactness is rarely an issue nondominant groups or of disaffected other. This would create districts that are in court, although it can be, depending voters within the dominant group are beyond the bounds of “competitiveness” upon the judge or judges involved. wasted. Their votes lose significance insofar as the favored party ought to have Creating packed partisan districts may because they are consistently cast for little trouble electing a candidate of their tend to involve drawing districts that are political losers. ~Guinier 1993! choice. The overall representation will less pleasing to the eye and less com- mirror the statewide breakdown, but this pact, although this is a relatively small Where Guinier and I depart company is districting plan also leaves a quarter of the price to pay for the benefits of having the necessity of wasting votes in a state unsatisfied with their representative districts in which most voters are single-member district system. I advocate and less likely to feel efficacious with winners. that we minimize the number of wasted respect to electing members of Congress. Fourth, map makers strive to “pre- votes, which makes her distinction be- Last, we bisect the state from north to serve communities of interest.” This tween dominant and nondominant groups south, creating two homogeneous parti- may be the single most fleeting guide- less important as the nondominant group san districts. Here too the state’s repre- line in district drawing. What constitutes is redrawn from several districts into sentation will mirror the overall a community of interest? A city? A their own district in which they are dom- partisanship of the state, and all citizens county? Any geographic region marked inant ~and a new nondominant group is end up voting for a representative that by even the slightest bit of partisan di- not created!. Her underlying concern is

82 PS January 2006 identical to mine—“votes lose signifi- a goal we ought to be striving for when districts—the critical component is that cance because they are consistently cast we draw legislative maps. If the Ameri- the ratio of the majority party to the mi- for political losers.” can people truly pine for turnover even nority party in each district be roughly The Voting Rights Act implicates all drawing competitive districts is no guar- equal across the state. the states under Section 2 and many of antee, rather we should institute term This method is probably best thought the states under Section 5 in terms of limits.14 While turnover and volatility of as functionally equivalent to state- election law. The Thornburg v. Gingles may decrease using this method, the based proportional representation.15 One decision ruled that districting plans must critical aspect of the plan to keep in of the positive attributes of proportional take care to ensure that minority votes mind is that the overall distribution of representation is the absence of “wasted are not systematically diluted, although seats will closely mirror the distribution votes” ~i.e., people that vote for the los- subsequent decisions have held that race of preferences in each state and nation- ing candidate in the general election!. cannot be used as the predominant fac- wide. The goal of redistricting is not to Single-member district systems, espe- tor in drawing districts ~Shaw v. Reno; maximize the number of seats that cially those with many competitive dis- Bush v. Vera!. Packing districts on the switch from one party to the other every tricts, end up wasting the votes of basis of ideology would not interfere two years; rather the goal of redistrict- millions of people. If we pack districts, with districting plans satisfying the re- ing is for the House to pass legislation the number of wasted votes would be at quirements of the Voting Rights Act. In in such a way that policy preferences a minimum, while still keeping the fact, using ideology rather than race, among the electorate are reflected in single-member district system to which but still preserving the ability of pro- policy outputs. Drawing districts on the Americans have grown accustomed. Lani tected minorities to have a reasonable basis of ideology satisfies this goal, Guinier is a vocal critic of the single- opportunity to elect a candidate of their while drawing competitive districts does member district system used in the choice, would be relatively easy to not. House. One of her most powerful criti- accomplish. Drawing districts packed with either cisms is that the redistricting process will like-minded partisans maximizes the inherently result in a gerrymander be- Discussion numbers of winning voters in legislative cause “in essence @redistricting# is the elections. Packing districts with ideolog- process of distributing wasted votes” Conventional wisdom suggests that ically like-minded individuals will not ~Guinier 1993!. Guinier’s fears are rather drawing competitive legislative districts elect more extreme candidates and ought well founded given the traditional dis- is beneficial to our system of govern- to strengthen the bonds between the tricting process, however much of her ment. In fact, drawing districts with rela- representatives and their constituents. concerns generally about single-member tively equal numbers of Democrats and Inferring how the electorate wants a rep- districts and the redrawing of their lines Republicans maximizes the number of resentative to vote on an issue will be disappear using the method outlined in losing voters ~also known as wasted much simpler when one’s district is sig- this paper. votes!. A voter on the losing side of an nificantly more homogeneous. If this For instance, Guinier alludes to the election is systematically more likely to type of districting plan were put into possibility of creating homogeneous be unhappy with his representative and effect it may also stimulate more com- districts: with Congress as an institution. Further, petition in primary elections. While the assumption that competitive general general elections to the House of Repre- Districts could be made more homo- elections make representatives more re- sentatives would surely be a largely un- geneous to reduce the number of wasted sponsive is also wrong. The implied exciting affair, primary elections could votes. But this alternative demonstrates threat of competition, especially at the rise in importance and become the more the second way that winner-take-all dis- primary level, is sufficient to keep our likely mechanism for replacing mem- tricting wastes votes. When more people elected officials faithful to our opinions. bers. Organized political parties often vote for the winning candidate than is Furthermore, this method of districting is seek to suppress competition at the pri- necessary to carry the district, their not the end of electoral competition; mary election stage largely because votes are technically wasted because rather it refocuses the nexus of competi- they do not want to reduce their party’s they were unnecessary to provide an tion from inter to intra-party. Even if a chances of winning the general election. voter, in a packed district, casts a ballot If the likelihood of winning the general electoral margin within the district and for the candidate that ends up losing in election is overwhelmingly in favor they could have been used to provide the dominant party’s primary, she still of one party or the other, this risk is the necessary electoral margin for a has the opportunity to cast a ballot for alleviated and ought to remove this like-minded partisan in another district. the winning candidate in the general barrier. In other words, packing voters in homo- election. Packing partisans also significantly geneous districts wastes votes because it Maximizing the number of competi- reduces the ability of map-makers to ef- dilutes their overall voting strength tive districts vastly increases the likeli- fect a significant partisan gerrymander. It jurisdiction-wide. hood that very small changes in the is through the combination of packing partisan leanings of voters nationwide and cracking that one party can effec- This is an important point insofar as a get translated into very large swings in tively dilute the votes of their opposition. gerrymander is only possible when the congressional seats. While political ana- As long as districts are packed with combination of packing and cracking lysts oftentimes fall into the trap of try- nearly equal proportions in terms of the districts is utilized. However, if all dis- ing to gauge the relative health of two major parties, the overall representa- tricts are packed then the real problem American democracy through the tion ought to closely mirror the distribu- Guinier addresses above is solved. Votes amount of turnover, or the lack thereof, tion of partisanship statewide. This does in packed districts are not effectively in the U.S. Congress, this is not at all not mean that both parties have equal wasted because all districts are packed, an appropriate method of estimating numbers of districts—if the Democrats and the overvotes could not be used to how well our democratic institutions are outnumbered Republicans 2 to 1 in the help a like-minded partisan in another functioning. Volatility in Congress is not state, there will still be more Democratic district.

PSOnline www.apsanet.org 83 District lines are artificial and we to be less concerned with keeping coun- to intra-party. Lastly, it is clear that ought to use this distinction to serve ties or municipalities whole, and con- we cannot rely on the courts to strike our collective advantage. Let us create centrate on keeping ideologically like- down gerrymandered redistricting more winners out of voters and increase minded individuals in the same district. plans.16 By instituting the method the satisfaction with our members of This change would fortify the bonds advocated in this paper, the likelihood Congress and Congress as an institution between voters and elected officials of a significant partisan gerrymander by rethinking the way in which we and transfer the locus of competition by either political party is greatly draw these lines. Map makers ought in congressional elections from inter- reduced.

Notes * I would like to thank Jim Adams, Valerie toward the Democratic candidate—55% Demo- makes it objectively more attractive for challeng- Brunell, Bruce Cain, Geoff Evans, Bill Koetzle, crat, 45% Republican. Although map makers ers from the dominant party to emerge. One only Bernie Grofman, Sam Hirsch, Michael D. Mc- need to be careful not to draw these too com- has to win the primary election in order to take Donald, Iain McLean, Sam Merrill, Glenn petitive as small swings in the vote could then the seat, whereas in a 50–50 district one could Phelps, David Rueda, Alec Stone Sweet, Chris reverse these districts and instead of a Gerry- face the daunting prospect of hard fought elec- Wlezian, and the Politics Group at Nuffield Col- mander you end with what Grofman and Brunell tions in both the primary and general elections. lege for their comments. ~2005! call a Dummymander. 12. This general trend is true using a wide 1. Throughout this paper “competitive dis- 7. For instance if a district is 45% Republi- variety of ideological scores from many different tricts” refers to a district drawn with relatively can, 45% Democratic and 10% Independent, the interest groups, see Lee, Moretti, and Butler equal numbers of voters that favor the two major only votes that matter are those from the small- 2004. parties. This is to say there is some probability est group. These independents will only be able 13. If a city or county does also happen to that the general election will be relatively com- to choose from either a Democrat or a Republi- be an ideological community of interest then it petitive ~although it is certainly not a guarantee can in the election, but nonetheless, they become ought to be kept in a single district. of competitiveness!. the votes that really count. 14. This should not be taken as an endorse- 2. They tested both voting for the winning 8. Proposition 206, passed by Arizona vot- ment of term limits, but rather a much more ef- candidate in an election and voting for the party ers in 2000, requires: “To the extent practicable, fective method of effecting turnover in Congress that wins a majority in Congress; neither model competitive districts should be favored where to than drawing knife-edged districts. yielded statistically significant results. do so would create no significant detriment to 15. It would differ from a proportional rep- 3. Of course this is not all voters, but rather the other goals” ~sec 14, subsection F!. resentation system insofar as we would still have the sub-sample of voters who answered this 9. I highly suggest reading the exchange a two-party system, whereas PR systems tend to question. between Issacharoff ~2002! and Persily ~2002! increase the effective number of parties. 4. This relationship remains positive and on the topic of the utility of competition in con- 16. Recently the Supreme Court ~Vieth v. statistically significant in a multivariate model. gressional elections. As both are also law profes- Jubuleier, 2003! nearly declared partisan ger- 5. See Buchler ~2005! for a formal treat- sors they also touch on legal issues and the rymandering nonjusticiable, which would have ment of this relationship. proper role of courts in regulating elections. made it impossible for courts to declare ger- 6. Cracking refers to the art of drawing 10. In fact competitive districts are optimal rymandered maps unconstitutional in the future. districts that are close to being competitive but in this sense. It remains unclear what standards will be used in give one party the edge in an election. For in- 11. In terms of incentives for challengers, the future by judges to decide gerrymandering stance, if the Democrats control the redistricting the opportunity structure of a packed district cases. process they are likely to draw districts that lean

References Anderson, Christopher J., and Christine A. Guil- Cox, Gary W., and Jonathan N. Katz. 2002. El- Hibbing, John R., and James T. Smith. 2001. lory. 1997. “Political Institutions and Satis- bridge Gerry’s Salamander: The Electoral “What the American Public Wants Con- faction with Democracy: A Cross-National Consequences of the Reapportionment Revo- gress to Be.” In Congress Reconsidered, Analysis of Consensus and Majoritarian Sys- lution. Cambridge: Cambridge University 7th edition, eds. Lawrence C. Dodd and tems.” American Political Science Review Press. Bruce I. Oppenheimer. Washington D.C.: 91 ~1!: 66–81. Erikson, Robert. 1978. “Constituency Opinion CQ Press. Anderson, Christopher J. and Andrew J. LoTem- and Congressional Behavior: A Reexamina- Hibbing, John R., and Elizabeth Theiss-Morse. pio. 2002. “Winning, Losing and Political tion of the Miller-Stokes Data.” American 1995. Congress as Public Enemy: Public Trust in America.” British Journal of Politi- Journal of Political Science 22~3!: 511–535. Attitudes towards American Political Institu- cal Science 32: 335–351. Erikson, Robert S., Norman R. Luttbeg, and Wil- tions. Cambridge: Cambridge University Bowler, Shaun, and Todd Donovan. 2002. “De- liam V. Holloway. 1975. “Knowing One’s Press. mocracy, Institutions and Attitudes about Cit- District: How Legislator’s Predict Referen- Hirsch, Sam. 2003. “The United States House izen Influence on Government.” British dum Voting.” American Journal of Political of Unrepresentatives: What Went Wrong Journal of Political Science 32~4!: 371–390. Science 19~2!: 231–246. in the Latest Round of Congressional Redis- Buchler, Justin. 2005. “Competition, Representa- Grofman, Bernard, and Thomas L. Brunell. 2005. tricting.” Election Law Journal 2~2!: tion, and Redistricting: The Case against “The Art of the Dummymander: The Impact 179–216. Competitive Congressional Districts.” Jour- of Recent Redistrictings on the Partisan Issacharoff, Samuel. 2002. “Gerrymandering and nal of Theoretical Politics 17 ~4!: 431–463. Makeup of Southern House Seats.” In Redis- Political Cartels.” Harvard Law Review Clarke, Harold D., and Alan C. Acock. 1989. tricting in the New Millennium, ed. Peter Gal- ~116!: 593–648 “National Elections and Political Attitudes: derisi. Lanham, MD: Lexington Books. Lee, David S., Enrico Moretti, and Matthew J. The Case of Political Efficacy.” British Jour- Guinier, Lani. 1993. “Groups, Representation, Butler. 2004. “Do Voters Affect or Elect Poli- nal of Political Science 19~4!: 551–562. and Race-Conscious Districting: A Case of cies? Evidence from the U.S. House.” Quar- Clarke, Harold D., and Allan Kornberg. 1992. the Emperor’s Clothes.” Texas Law Review terly Journal of Economics 1993~3!: “Do National Elections Affect Perceptions of 71: 1589. 807–859. MP Responsiveness? A Note on the Canadian _. 1994. The Tyranny of the Majority: Mann, Thomas. 1978. Unsafe at any Margin: Case.” Legislative Studies Quarterly 17~2!: Fundamental Fairness in Representative Interpreting Congressional Elections. Wash- 183–204. Democracy. New York: Free Press. ington, D.C.: American Enterprise Institute.

84 PS January 2006 Mayhew, David R. 1974. Congress: The Elec- “Constituency Influence in Congress.” Amer- Sapiro, Virginia, Steven J. Rosenstone, and the toral Connection. New Haven: Yale Univer- ican Political Science Review 57~1!: 45–56. National Election Studies. American National sity Press. Persily, Nathan. 2002. “In Defense of Foxes Election Studies Cumulative Data File, McCrone, Donald J., and James H. Kuklinski. Guarding Henhouses: The Case for Judicial 1948–2000 @Computer file#. 11th ICPSR ver- 1979. “The Delegate Theory of Representa- Acquiescence to Incumbent Protecting Ger- sion. Ann Arbor: University of Michigan, tion.” American Journal of Political Science rymanders.” Harvard Law Review ~116!: Center for Political Studies @producer#, 2002. 23~2!: 278–300. 649–683. Ann Arbor: Inter-university Consortium for Miller, Nicholas R. 1983. “Pluralism and Social Poole, Keith, and Howard Rosenthal. 1991. “On Political and Social Research @distributor#, Choice.” American Political Science Review Dimensionalizing Roll Call Votes in the U.S. 2002. 77: 734–747. Congress.” American Political Science Re- Miller, Warren E., and Donald E. Stokes. 1963. view 85~4!: 955–960.

PSOnline www.apsanet.org 85