Voter-Determined Districts

Ending Gerrymandering and Ensuring Fair Representation

Truly fair U.S. electoral districts would allow the views of elected representatives to better reflect those of the people who elect them.

In this article
A voter takes in the view at her polling place in Agua Dulce, California, November 2018. (Getty/Mario Tama)

Introduction and summary

From 2012 to 2016, the people of Michigan cast more than 50 percent of their ballots for Democratic Party legislative candidates. They voted for Democrats 52 percent of the time for the Michigan House of Representatives; a little more than 50 percent of the time for the Michigan Senate; and 51 percent of the time for the U.S. House of Representatives.1 So one would expect that slightly more than half of Michigan elected officials during this time were Democrats.

Instead, Republicans held a decisive advantage at every level of government. Despite earning a majority of the vote, Democrats received only 44 percent of seats in the Michigan House of Representatives; 31 percent of the seats in the Michigan Senate; and 35 percent of the seats in Michigan’s delegation to the U.S. House of Representatives.2 Although this degree of misalignment is severe, it is not unusual. Currently, districts in most states are drawn in ways that are gerrymandered—meaning the lines are manipulated to favor one group over another—because the process allows elected representatives to choose their voters rather than allowing voters to choose their representatives.

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

The first step in addressing this problem is to create a process for drawing districts that is not controlled by incumbent politicians. But changes to the process are not enough. Independently drawn maps can have the same effect as intentional gerrymanders if they are not drawn according to the right set of criteria. In fact, proposals that have gained widespread acceptance do not directly address the misalignment between voters and their representatives.

Fortunately, there is a solution, and it is surprisingly simple: purposefully drawing districts to reflect the political choices of voters—what this report terms “voter-determined districts.” Voter-determined districts are based on the principle that the makeup of the legislature should reflect the preferences of voters statewide.

This report provides data on the partisan skew of state legislative and congressional districts and explains why a shift is needed in the policy debate about redistricting reform. It also explains some of the sensible reasons why the effort to stop gerrymandering has not, until now, been focused on drawing voter-determined districts. Finally, the report outlines how to implement voter-determined districts while also achieving two other critical goals: maximizing representation for communities of color and ensuring adequate electoral competition.

The problem with existing districts

Although legislators should reflect the voters of their state, they often do not. In Maryland, for example, Republicans received 37 percent of the votes for the U.S. House of Representatives but won only 13 percent of the congressional seats. And in North Carolina, Democrats received 48 percent of the vote for the U.S. House of Representatives but won only 26 percent of the congressional seats.

Figure 1 shows how well legislatures reflect the voting patterns of the population for each state and at each level of government—state House, state Senate, and U.S. House of Representatives. The percentage displayed for each state is the degree to which districts disproportionately favor 1 of the 2 major political parties, calculated by comparing the total percentage of votes cast for Democratic and Republican candidates to the total percentage of elections won by Democratic and Republican candidates, and excluding both votes and wins for nonmajor-party candidates. Biases in favor of Democrats are highlighted in blue, and biases in favor of Republicans are highlighted in red.3 The data reveal substantial biases in favor of each party. Moreover, it shows that biased districts are widespread–about two-thirds of all state House, state Senate, and U.S. House delegations are biased in favor of one party or the other by a rate of at least 5 percent.

Note one caveat to these data: While much of this bias can be eliminated, not all of it can. In some cases, the geographic distribution of voters makes it difficult or impossible to draw better districts. In Hawaii, for example, the 29 percent of voters supporting Republican U.S. House candidates are distributed relatively evenly throughout the state, so it may not be feasible to draw a Republican-leaning district. In two states, Louisiana and Mississippi, switching one U.S. House district from favoring Republicans to favoring Democrats would result in only a slight reduction in partisan bias—a bias in favor of Democrats instead of Republicans.

Figure 2 shows how biased districts for the U.S. House of Representatives translated into actual seats. On average, for each of the three elections from 2012 to 2016, 59 seats would have changed hands to the opposing party if the percentage of seats won matched the percentage of votes cast. On net, Republicans won 19 additional seats each election because of districts biased in their favor.

There are two reasons for these mismatched election outcomes. The first reason is intentional gerrymandering—purposeful manipulation of district lines by incumbent politicians. After each decennial census, when new population numbers are released, and new districts are drawn, incumbent politicians interfere with the map-drawing process in order to create districts that protect them from competition and that favor their political party. In fact, many states explicitly put incumbent legislators in charge of the map-drawing process.4 Intentional gerrymanders have gotten worse with the advent of modern redistricting software, which allows political operatives to pinpoint specific types of voters and more effectively draw districts that will serve their partisan ends.5

Intentional gerrymandering is the redistricting problem that is most well understood and that reform-minded advocates have been almost exclusively focused on addressing. However, it is not the only problem. There is also the issue of what scholars have termed “unintentional gerrymandering.”6

To understand unintentional gerrymandering, consider what would happen if a mapmaker drew roughly rectangular districts at random. In some lucky cases, the random districts might result in legislators that reflect the voting population. Much more often, however, because voters are not evenly spread out, the randomly drawn districts would result in a random gerrymander. The districts are more likely than not to overrepresent one political party, even if that was not the intent.

To make matters worse, unintentional gerrymandering does not affect both parties equally. Democrats are heavily concentrated in cities, while Republicans are more spread out across suburban and rural areas. Therefore, randomly drawn, compact districts will tend to pack Democrats in a small number of urban districts, while giving Republicans more modest majorities in a larger number of rural and suburban districts—not unlike an intentionally designed pro-Republican gerrymander.

Political scientists Jowei Chen and Jonathan Rodden analyzed this phenomenon at length in their 2013 article, “Unintentional Gerrymandering: Political Geography and Electoral Bias in Legislatures.”7 As an example, they looked at the state of Florida during the 2000 elections, when the state was so evenly divided between Democrats and Republicans that it resulted in a recount, which was ultimately halted by a U.S. Supreme Court decision.8 Chen and Rodden ran a simulation in which they generated 25 different state maps for Florida’s federal House districts. Each map was randomly drawn, except that all the districts contained close to the same total population and were subject to a formula to ensure that they were somewhat compact. The authors then calculated how many seats Republicans and Democrats would be likely to win based on each map.

Sure enough, the average map would have awarded Republicans 61 percent of the legislative seats.9 And even the least-biased map would have awarded Republicans 56 percent of the seats.10 In other words, the maps uniformly produced pro-Republican gerrymanders—unintentionally. Chen and Rodden then repeated a version of their process without the compactness criteria, drawing 250 random maps. Again, all the maps were biased in favor of Republicans, producing a congressional delegation between 56 percent Republican and 76 percent Republican.11

States vary in terms of how voters are distributed and how much unintentional bias is likely to result. However, the pattern seen in Florida appeared to be common nationwide. Chen and Rodden were able to obtain data to perform a similar analysis for 20 other states and found that “the simulations consistently produced similar or even higher levels of pro-Republican bias than in Florida.”12 The worst bias tended to be in states that had experienced the most urbanization13—a trend that has increased since 2000 and is likely to continue.14 Given this context, it seems unlikely that maps drawn randomly or using some set of politics-blind criteria would produce legislators that accurately reflect the political choices of their constituents.

Unfortunately, such politics-blind redistricting is currently the leading policy solution. There is an undeniable appeal to the idea that politics should be taken out of the redistricting process. And there has long been a focus on stopping intentional gerrymandering for a variety of practical and historical reasons that are explained further in the next section.

Yet, solving gerrymandering without considering voters’ political views is sort of like trying to address systemic racism without considering race; ignoring the crux of the issue actually perpetuates the problem. In the final section, this report proposes an alternative approach to fixing gerrymandering: affirmatively harnessing political data and using that data to draw districts so that legislators’ views reflect the views of the voters they represent, while maximizing representation for communities of color and ensuring competitiveness in elections.

Even under the approach offered below, however, there will occasionally be differences between votes and outcomes; this is an inevitable result of a political system that requires districts to be drawn. Perfect alignment between representatives and votes would require a very different political system—something that is also discussed below. However, most significant mismatches between voters and their legislators are the result of gerrymandering—intentional or otherwise—that can and should be eliminated.

How the current gerrymandering landscape was formed

Gerrymandering has been a reality in the United States for a long time; the term dates back to 1812, and the practice itself predates the Constitution.15 The fight against gerrymandering, therefore, has been shaped in part by a justifiable pessimism about the prospects of getting self-interested politicians to fix the system through legislation. Rather, advocates for reform have often been focused on curbing the most egregious abuses of the redistricting process—gerrymanders aimed at disenfranchising entire communities on the basis of race—and extreme, intentional gerrymanders by partisan incumbents. These fights have most often taken place in the courts, where incumbent politicians have less sway.

The long fight against gerrymanders that exclude communities of color

The United States has a long history of depriving nonwhite citizens of their political rights. Only as the result of a long and hard-fought struggle have African Americans been able to secure representation in Congress and in their state and local governments. Although African Americans and other underrepresented communities have made substantial gains, there is still a long way to go for communities of color to have political power commensurate with their numbers.16

The enactment of the Voting Rights Act (VRA)of 1965 was a historic victory for the civil rights movement and had a major impact on redistricting. The law prohibited any jurisdiction from imposing barriers that denied the right to vote on the basis of race; one of those barriers, the courts determined,17 was district-drawing that diluted the political power of communities of color. This is not a problem of the distant past: A 2006 review of litigation under the VRA found that between 1982 and 2006 there were 111 cases challenging postcensus redistricting plans under Section 2 of the VRA, with favorable outcomes for the plaintiffs in 46 of the cases.18

At the time of the VRA, one way that white communities prevented African American communities from having a political voice was by spreading their votes out over multiple districts so that they would not be able to muster majority support for an African American candidate in any one district. For example, in the 1970s, the city of New Orleans was 45 percent African American.19 However, that population had been intentionally divided among city council districts so that no single district was made up of a majority of African American voters. Between 1960 and 1970, not a single African American was elected to the New Orleans City Council.20

In the wake of the 1982 amendments strengthening the VRA, the U.S. Supreme Court established a set of factors to determine whether a redistricting plan or other electoral change unlawfully infringed upon the voting rights of “any citizen of the United States… on account of race or color.”21 One important factor was the presence of racially polarized voting—patterns of voting where white communities supported different candidates than their nonwhite neighbors, with little crossover. In the civil rights era, voting was heavily polarized; candidates supported by communities of color could expect to receive little or no support from white voters. Additionally, registration and turnout among African American voters was low, substantially due to campaigns of violence, intimidation, and limits on the franchise imposed by white majorities. In combination, that meant that some districts needed greater than 65 percent of the population to be African American in order for the African American community to elect a candidate of their choice.22

Districts were redrawn in the early 1990s, for the first time following the Supreme Court’s decision, and included many more of these supermajority-African American districts. The effect was dramatic: The number of African American members of Congress increased from 25 at the end of the 101st Congress in 1991 to 41 at the beginning of the 103rd Congress in 199323—increasing the size of the Congressional Black Caucus by 65 percent.24 Prior to the enactment of the VRA, the number of African Americans in Congress had never reached double digits.25

Even in the 1990s, however, the gains in representation for African Americans were not as substantial as they could have been. Part of the reason for this is that, even by then, voting had become substantially less racially polarized. Studies have shown that over time, many more white voters have supported nonwhite candidates; this means that, generally, African Americans do not have to make up a supermajority of the population of a district in order for their community to elect a representative of their choice.26 By drawing districts with more modest African American majorities, a map can include more districts in which African American communities are able to elect their chosen representatives.

Ironically, however, pro-Republican map-drawers often try to use the VRA as an excuse to pack large African American populations into a small number of districts in order to decrease overall Democratic representation. In recent times, the most frequent way that gerrymanders have reduced the political power of these communities is by packing them into fewer districts than they could otherwise win—not by spreading them too thin.27

Ensuring political representation for communities of color remains a top priority for reform advocates and should be a central consideration in the redistricting process. Despite some recent progress,28 Congress, and particularly state legislatures, are nowhere near as diverse as the U.S. population29—a harm that has been shown to have substantive effects on how well the concerns of communities of color are addressed.30 Part of the solution is to draw more districts that give these communities a seat at the table.

Sometimes debates about redistricting can get hung up on the potential for conflict between securing representation for underrepresented communities and other goals. However, the proposal advocated in this report affirms that maximizing representation for communities of color should be a primary goal when drawing districts. With the advent of modern map-drawing software, it is likely that most district maps can both more accurately represent voter preferences and increase representation for communities of color.

The court-focused fight against partisan gerrymandering

For most of American history, there has not been much prospect of ending partisan gerrymandering through legislation—and it’s not hard to see why. Incumbent politicians are the only beneficiaries of a successful gerrymander, and even legislators from the minority party may be resistant to changing the districts because minority-party members often negotiate with majority-party map-drawers to secure safe districts for themselves.31

However, the political logic of gerrymandering has shifted dramatically in recent years. Multiple polls have found that the public is overwhelmingly opposed to gerrymandering,32 which aligns with the growing public perception that legislatures are corrupt and unresponsive.33 Elected officials at all levels of government have taken note and are taking up the banner of reform.34 On March 8, 2019, every Democrat in the U.S. House of Representatives voted in favor of the For the People Act, which contained a provision championed by Rep. Zoe Lofgren (D-CA) that would take redistricting out of the hands of incumbent politicians and require maps to be drawn by independent commissions instead.35 Meanwhile, former President Barack Obama, with the help of his former attorney general, Eric Holder, has reportedly gone “all in” on the fight against gerrymandering.36 A few prominent Republicans have spoken out on the issue as well, including former California Gov. Arnold Schwarzenegger37 and former Ohio Gov. John Kasich.38 And a number of states have reformed their redistricting processes—some by ballot measure—so that incumbent politicians no longer draw their own districts.39

Not long ago, all of this progress would have been unthinkable. And for that reason, much of the momentum for change has been focused on the courts, where advocates continue to hope that judges will force legislators to do the right thing: redraw heavily gerrymandered maps.

This is an ongoing effort—and an important one. The U.S. Supreme Court has had prior opportunities to decide whether claims about partisan gerrymanders presented justiciable constitutional harms, and so far, it has chosen to defer.40 However, there are two cases currently pending before the Supreme Court that could decide the issue;41 although given the conservative makeup of the court, it appears to be an uphill battle. Yet, depending on the outcome, the Supreme Court could undo some of the worst gerrymanders in the country with the stroke of a pen.

It is important to note, however, the limits of what any victory at the Supreme Court would mean. At best, the court will impose a new standard that limits extreme gerrymanders. The court is not being asked to determine the best way to draw a state’s districts; it is being asked to determine, in the case of districts already subject to partisan manipulation, how egregious a gerrymander must be to have violated a constitutional right. Even the League of Women Voters—one of the parties arguing that the Supreme Court should strike down a North Carolina gerrymander—says that the district map must produce a “large and durable advantage for the line-drawing party” in order to be unconstitutional.42

If a majority of justices decide that partisan gerrymandering is justiciable, then the effect of that decision will depend on the relative leniency or stringency of the court’s definition of an unconstitutional partisan gerrymander. In case there was any doubt, former Supreme Court Justice Anthony Kennedy made clear in a 2006 decision that the court would not require a state’s congressional delegation to match its statewide share of the vote, writing that “there is no constitutional requirement of proportional representation.” 43

Thus, even the most positive court decision will not eliminate intentional partisan gerrymandering—and it may do very little about unintentional gerrymandering. One of the plaintiffs in the pending cases has asked the court to prohibit gerrymandering based strictly on intent,44 which would mean that even a map that in every way resembles an extreme partisan gerrymander would be constitutional so long as it could not be shown that this outcome was intentional. Even though the Supreme Court once said that the “basic aim” of redistricting is “fair and effective representation for all citizens,”45 litigation is unlikely to impose redistricting criteria that fully assure such fairness.

The challenge of educating the public about gerrymandering

Efforts to end gerrymandering have been shaped by the history of the fights against racial gerrymandering as well as the effort to fight partisan gerrymandering in the courts. But they have also been shaped by the advocates and activists who have been tirelessly working to educate the public about the negative effects of gerrymandering and to persuade them to join the fight against it.

Because of these efforts, the prospects for taking action against gerrymandering are better than ever. Ironically, however, some strategies that have proved necessary and useful in informing the public about the evils of gerrymandering have had a negative effect on how the public may understand the solutions. For example, in the larger public imagination, gerrymandering is the result of a bad actor—incumbent politicians—drawing maps for their self-benefit. That’s the way the story is often told, and it’s the simplest, most appealing way to explain the harms of gerrymandering.

As this report shows, however, bad maps are sometimes the result of geography or of a map-drawing process that is essentially random. Yet, to many people, the idea of simply taking politicians out of the equation and adopting seemingly neutral criteria such as compactness and respect for existing boundaries sounds like a solution. It’s not obvious that there is more to it.

Moreover, many people associate gerrymandering with oddly shaped districts. This is for good reason; redistricting is a complicated subject, and showing people bizarre, gerrymandered districts is an effective way to demonstrate the lengths to which partisans are willing to go to secure an unfair advantage. Even the term “gerrymandering” likely refers to a district that a cartoonist portrayed as a monstrous salamander, from a districting plan signed into law by Elbridge Gerry when he served as Governor of Massachusetts.46

Yet, although bizarrely shaped districts can sometimes indicate a partisan gerrymander, extreme gerrymanders can also result from normal-looking districts;47 packing Democratic voters into an urban district can be as simple as drawing an inoffensive-looking square. Conversely, an odd-looking district might help better represent the population as a whole.

The solution: Voter-determined districts

Conventional thinking on redistricting reform holds that political considerations have no place in the map-drawing process. But the problem is not that voting involves politics—of course it does. The problem is that partisans seek to undermine fair representation of voters’ political views. In fact, consideration of voters’ party preferences is the only way to ensure that district maps are democratically fair and that elected bodies reflect the people’s will.

In the scholarly literature on gerrymandering, a frequently discussed method for comparing voter preferences with outcomes is a concept called “partisan symmetry,” which is sometimes described as a way to effectuate “partisan fairness.”48 First introduced by Professors Gary King and Robert Browning,49 partisan symmetry is achieved “when reversing the outcome of the election—flipping each party’s average district vote totals—would also reverse the number of seats won.” 50 For example, “a map would be perfectly symmetrical if a party that wins 60% of the average district vote receives 80% of the seats, provided that the opposing party could do the same.”51 In the words of Professor King, “any translation of votes to seats is fair—as long as it is symmetric.”52

In other words, partisan symmetry encapsulates the most minimal way in a two-party system that both parties ought to be treated. The results don’t have to be fair; they must only at least be equally unfair, if the tables were turned. It’s a simple but powerful concept for identifying extreme partisan gerrymanders and is often referenced in litigation.

When it comes to making policy, however, legislators should do better than equally unfair. For starters, the focus should be on the voters, not the parties. If 60 percent of the voters are supporting one party, and that party wins 80 percent of the seats, that’s not fair to the 40 percent of the voters who supported the other party. Holding only 20 percent of the seats means their votes are effectively counting for half of what they should. It is no consolation to the voter that the unfairness would point in the other direction if the tables were turned.

The Center for American Progress’ solution is to have districts determined by voters, to the greatest extent possible, through the exercise of the right to vote. If voters in a state favor their Democratic candidates by a margin of 60 to 40, then districts should be drawn so that the same ratio of districts favor each of the respective parties; in this case, 60 percent Democratic-leaning districts and 40 percent Republican-leaning districts. Conversely, in a state where voters favor Republicans, districts should favor Republicans in the same way. That’s basic fairness, and it is the most likely way to get representation that is democratic.

In addition to fairness, this proposal has another significant benefit: It makes every vote count in a whole different way. Voters in noncompetitive districts would now have a stronger incentive to go to the polls and cast a ballot because, even if their chosen candidate is unlikely to win, their votes will be instrumental in how the next set of districts are drawn. Political parties, similarly, would have an incentive to recruit competitive candidates for every district in order to boost their total votes. And all voters, including those who already regularly participate in elections, would have a stronger incentive to vote in down-ballot state legislative contests, which are often left blank.

Of course, to effectuate voter-determined districts, the redistricting process must be free of manipulation. And district-drawers should also ensure that final maps achieve critical goals such as providing representation to communities of color and ensuring adequate levels of electoral competition.

Policy recommendations

This section provides more detail on how policymakers at the federal and state levels can implement voter-determined districts: a set of criteria that results in districts that accurately represent the public, while also maximizing representation for communities of color and ensuring reasonable levels of electoral competition.

Establish independent commissions to oversee the redistricting process

One thing that most redistricting reform advocates agree on is the need for independent commissions insulated from political interference to oversee the redistricting process. Such commissions should exclude all incumbent politicians or political party officials, and they should generally be structured to avoid allowing registered members of any one political party from controlling commission decision-making. Often this is achieved by requiring even numbers of members from both parties, and at least one independent tie-breaker, as was proposed in the House-passed For the People Act.53 Members can also be chosen through processes designed to reduce the possibility of political influence such as random selection from a pool of qualified applicants.54

Finally, the criteria for redistricting, specified below, should be specific enough that even a politically motivated bloc of commissioners would have difficultly manipulating the process for partisan ends or using the redistricting process to protect incumbent politicians.55 Clear, specific criteria have the added advantage of helping avoid costly litigation after a plan is passed.

Draw voter-determined districts

Commissions should be required at the outset to generate a large number of computer-drawn maps with districts that are voter-determined and comply with the requirements of the Constitution and the VRA.

Benchmark for determining voter preferences

Voter preferences should be determined by totaling the votes cast for candidates of each party in each of the general elections since the prior redistricting cycle. Maps should then be drawn so that the percentage of districts that favor each party should—to the extent reasonably possible—match the percentage of the total votes received by candidates of each party in prior elections.

These numbers should be calculated separately for each set of districts drawn. In other words, state Senate districts should be calculated based on votes for state Senate candidates; state House districts should be calculated based on votes for state House candidates; and federal congressional districts should be calculated based on votes for U.S. House candidates.

As noted above, there will be some circumstances in which it is not possible to draw maps that align the percentage of votes with the percentage of seats—either because of the distribution of voters or the limited number of representatives. In such cases, districts should be drawn as close to alignment as possible, and map-drawers should also attempt to include more competitive districts, to give the under-represented party a fighting chance to win additional seats.

How to account for uncontested districts and incumbency advantage

One issue that may arise with the method above is that the current, gerrymandered maps can affect the vote totals. Many districts are so uncompetitive or have had such long-entrenched incumbents that they are not contested—and because voters don’t have an opposition candidate to select, that drives down vote totals for the opposition party. However, the simplest and best solution may be to use the raw vote totals anyway, understanding that: 1) The new maps will still be much better than the current ones; and 2) Any residual advantage from prior gerrymandering should disappear after a couple rounds of redistricting because the new maps will reduce the uncontested districts, which will improve the results of the next round of redistricting. Alternatively, policymakers could account for uncontested districts by using widely accepted statistical methods to estimate how many votes an opponent would have received.56

Produce a large number of maps

Modern redistricting software allows map-makers to relatively easily produce a large number of maps with a given set of criteria. The advantage of doing this is that it makes it easier to achieve multiple goals. Many different maps can usually be drawn to fit one goal such as accurately representing voters, and then, among those maps, the commission can select maps that achieve additional goals such as maximizing representation for communities of color and providing reasonable levels of competition, as detailed below.

Select maps that maximize representation for communities of color

The VRA prohibits redistricting plans from diluting representation of certain communities on the basis of race. However, commissions can—and should—go beyond the requirements of the VRA in order to allow as many underrepresented racial, ethnic, and language communities as possible to elect representatives of their choice. Doing so will help offset the range of other barriers faced by nonwhite candidates and begin to reverse the demographic imbalance in Congress and state legislatures. One can expect the resulting, more diverse legislatures to more genuinely represent all the interests of the American people.

Select maps that provide for reasonable levels of competition

One of the most prevalent complaints about gerrymandering is that it sets up incumbents in safe districts made up overwhelmingly of voters from their own party, which insulates them from meaningful competition. This prevents the makeup of the legislature from changing even when there are large swings in popular support. However, it is also possible to go overboard with competition—if too many districts are highly competitive because they are evenly split between the two parties, a small shift in support could create an overwhelming victory for one party, which would mean that the legislature no longer accurately represents the voters. Therefore, redistricting commissions should select maps that create substantial numbers of competitive districts—much higher levels of competition than currently exist—but without creating a situation where small changes in support would lead to unfair results.

Put a thumb on the scale in favor of district compactness

To the extent that it does not significantly interfere with the goals above, commissions should also impose a reasonable limit on the size and shape of districts. In and of itself, compactness is not a particularly important policy goal. However, because the public often equates oddly drawn districts with unfairly drawn districts, commissions should try to minimize, on the margin, excessively bizarre-looking districts.

Make the process transparent

Finally, in order to ensure that commissions follow the steps specified above, the redistricting process should be fully visible to the public. All the substantive decisions that the commission makes should be made in full public view. Furthermore, at each stage of the process, all the maps generated and selected should be made available online so that independent experts can verify that the criteria were followed as specified.

Conclusion: The promise and limits of redistricting reform

The proposal described in this report is designed to promote equal and accurate representation and to eliminate partisan gerrymandering for good. An America with voter-determined districts would be a much more democratic one—where politicians couldn’t hide from accountability; where every vote would make a difference; and, most importantly, where the decisions of elected officials would better reflect the views of the people.

There are some problems that redistricting cannot fix, however. Many of the criticisms that could be leveled at voter-determined districts stem from the simple fact that there are some downsides to having districts at all. Districts inevitably split some communities between two representatives. They inevitably result in some noncompetitive districts. And district-based systems such as that in the United States, where voters select a single candidate, and the candidate with the most votes wins—often referred to as a first-past-the-post system—have other drawbacks as well such as the obstacles faced by third parties and the fact that some votes are cast for candidates who don’t win.

There are other ways to address gerrymandering that are more far-reaching and novel. One idea is to create larger multimember districts where voters can select and rank multiple candidates.57 The candidate receiving the fewest votes would be removed, and their votes would be reallocated to their supporters’ second-ranked candidates. That process then repeats itself until there are only as many candidates as there are representatives assigned to the district—between three and five, under the most-cited proposal.58 It’s a system in which a broader array of candidates could feasibly run and where every voter would be able to have their vote count for a winning candidate. However, it is also an untested, radical departure from the status quo, and it may be a leap that incumbent elected officials would be reluctant to take.59

Fortunately, this report has described a way that elected officials can continue to be committed to districts without being committed to gerrymandered districts. A system of voter-determined districts is a solution, well within reach, that would end gerrymandering without upending U.S. politics. All that is required is for elected officials to heed the advice above and to take literally the redistricting proverb: that politicians shouldn’t choose their voters—voters should choose their politicians.

About the author

Alex Tausanovitch is the director of campaign finance and electoral reform at the Center for American Progress.


The author would like to thank Chris Warshaw for pointing the way on state election data and generously sharing his own data on U.S. House elections; Danyelle Solomon, Connor Maxwell, Sam Berger, and Ben Olinsky for providing thoughtful feedback that much improved the final report; and Danielle Root for her indispensable help on an early version of the report. All errors and opinions are the author’s only.



  1. Data are on file with author.
  2. Data are on file with author.
  3. Some states are displayed in gray in Figure 1A, even though the partisan bias is greater than 5 percent, because the state cannot do anything more to reduce partisan bias with the number of seats available. For example, almost 43 percent of the votes cast in Montana U.S. House elections were for Democrats, but no Democrats won because Montana only has one seat in the U.S. House. In other states, such as New Mexico, which has three House seats, the partisan bias is low enough that changing a district—in this case, from a Democratic district to a Republican district—would result in a worse partisan bias in the opposite direction. In other words, these states have drawn the most representative House districts they can given the number of seats apportioned to that state. The 10 states displayed in gray for this reason are: Alaska, Colorado, Delaware, Maine, Montana, North Dakota, New Mexico, South Dakota, Vermont, and Wyoming. For more information, see the table in the Appendix.
  4. All About Redistricting, “Who draws the lines?”, available at (last accessed March 2019)
  5. Vann R. Newkirk II, “How Redistricting Became a Technological Arms Race,” The Atlantic, October 28, 2017, available at
  6. Jowei Chen and Jonathan Rodden, “Unintentional Gerrymandering: Political Geography and Electoral Bias in Legislatures,” Quarterly Journal of Political Science 8 (3) (2013): 239–269, available at
  7. Ibid.
  8. Bush v. Gore, 531 U.S. 98 (December 10, 2000), available at
  9. Chen and Rodden, “Unintentional Gerrymandering,” p. 253.
  10. Ibid.
  11. Ibid, p. 254.
  12. Ibid, p. 262.
  13. Ibid, p. 262.
  14. Beau Dure, “Millennials Continue Urbanization Of America, Leaving Small Towns,” NPR, October 21, 2014, available at
  15. Emily Barasch, “The Twisted History of Gerrymandering in American Politics,” The Atlantic, September 19, 2012, available at
  16. Amber Phillips, “The striking lack of diversity in state legislatures,” The Washington Post, January 26, 2016, available at
  17. See, for example, Allen v. State Board of Elections, 393 U.S. 544 (March 3, 1969) (holding that the right to vote protected by Section 5 of the Voting Rights Act includes “all action necessary to make a vote effective,” including a change from districts to at-large voting), available at; Beer v. United States, 425 U.S. 130 (March 30, 1976) (setting out the standard for review of districting plans under Section 5 of the Voting Rights Act), available at; Thornburg v. Gingles, 478 U.S. 30 (June 30, 1986), available at (setting out the factors for determining whether a legislative redistricting plan violates Section 2 of the Voting Rights Act, as amended in 1982).
  18. Ellen D. Katz, “Not Like the South: Regional Variation and Political Participation Through the Lens of Section 2,” in Ana Henderson, ed., Voting Rights Act Reauthorization of 2006: Perspectives on Democracy, Participation, and Power (Berkeley, CA: Berkeley Public Policy Press, 2007), p. 192, available at
  19. Beer v. United States.
  20. Ibid.
  21. Thornburg v. Gingles.
  22. Richard H. Pildes, “Is Voting-Rights Law Now at War with Itself—Social Science and Voting Rights in the 2000s,” North Carolina Law Review 80 (5) (2002): 1517–1574, p. 1525, available at
  23. U.S. House of Representatives, “Black-American Representatives and Senators by Congress, 1870–Present,” available at (last accessed March 2019).
  24. John Bresnahan, “Congressional Black Caucus: A bastion of seniority,” Congressional Black Caucus, March 2, 2014, available at
  25. United States House of Representatives, “Black-American Representatives and Senators by Congress, 1870–Present.”
  26. Pildes, “Is Voting-Rights Law Now at War with Itself,” p. 1529–1539.
  27. Kim Soffen, “How racial gerrymandering deprives black people of political power,” The Washington Post, June 9, 2016, available at
  28. Cristina Marcos, “115th Congress will be most racially diverse in history,” The Hill, November 17, 2016, available at
  29. Amber Phillips, “The striking lack of diversity in state legislatures,” The Washington Post, January 26, 2016, available at
  30. Kenneth Lowande, Melinda Ritchie, and Erinn Lauterbach, “Descriptive and Substantive Representation in Congress: Evidence from 80,000 Congressional Inquiries,” June 8, 2018, available at
  31. Campaign Legal Center, “Designing Independent Redistricting Commissions” (2018), p. 21, p. 27, available at
  32. The Harris Poll, “Americans Across Party Line Oppose Common Gerrymandering Practices,” November 7, 2013, available at; and Kylee Groft, “The Results are in: Most Americans Want Limits on Gerrymandering,” Campaign Legal Center, September 11, 2017, available at
  33. Ella Nilsen, “New polling shows voters—including independents—want Congress to pass an anti-corruption bill,” Vox, January 3, 2019, available at
  34. Brennan Center for Justice, “State Legislative Redistricting Bills – 2018,” available at (last accessed March 2019); Brennan Center for Justice, “Congressional Redistricting Bills – 115th Congress,” December 18, 2018, available at
  35. For the People Act of 2019, H.R. 1, 116th Cong., 1st sess. (January 3, 2019) available at
  36. Edward-Isaac Dovere, “Barack Obama Goes All In Politically to Fight Gerrymandering,” The Atlantic, December 20, 2018, available at
  37. Edward-Isaac Dovere, “Arnold Schwarzenegger’s War on Gerrymandering Is Just Beginning,” The Atlantic, November 9, 2018, available at
  38. Lynn Hulsey, “Kasich backs redistricting reform,” Dayton Daily News, January 3, 2016, available at–politics/kasich-backs-redistricting-reform/MZWrwndiJR1n8O3KzdxJkO/.
  39. Emily Moon, “How Did Citizen-led Redistricting Initiatives Fare in the Mid-Terms?”, Pacific Standard, November 7, 2018, available at
  40. See Justice Anthony Kennedy’s concurring opinion in Vieth v. Jubelirer, 541 U.S. 267 (April 28, 2004) (stating that no workable standard for political gerrymandering had been identified, but that “(i)f workable standards do emerge to measure… courts should be prepared to order relief”), available at; Gill v.  Whitford, 138 S. Ct. 1916 (June 18, 2018) (remanded on procedural grounds without addressing the partisan gerrymandering claim), available at
  41. SCOTUSblog, “Rucho v. Common Cause,” available at (last accessed March 2019); Brennan Center for Justice, “Lamone v. Benisek,” available at (last accessed March 2019).
  42. Robert A. Rucho, et al., v. Common Cause, et al., brief for appellees League of Women Voters of North Carolina et al., U.S. Supreme Court, No. 18-422 (March 4, 2019), p. 55, available at
  43. League of United Latin American Citizens v. Perry, 548 U.S. 399 (June 28, 2006), available at
  44. Robert A. Rucho, et al., v. Common Cause, et al., brief for common cause appellees, U.S. Supreme Court, No. 18-422 (March 4, 2019), p. 55, available at
  45. Reynolds v. Sims, 377 U.S. 533, 565–568 (June 14, 1964), available at
  46. Erick Trickey, “Where Did the Term ‘Gerrymander’ Come From?”,, July 20, 2017, available at
  47. Beverly R. Gill, et al., v. William Whitford, et al., brief for appellees, U.S. Supreme Court, No. 16-1161 (August 28, 2017) (“Gerrymanders, that is, may exist even when they do not announce themselves with strange shapes or carved communities.”), p. 27, available at
  48. See, for example, League of United Latin American Citizens v. Perry (Stevens concurring in part and dissenting in part, stating that partisan symmetry is “widely accepted by scholars as providing a measure of partisan fairness in electoral systems.”), p. 466; American Academy of Arts and Sciences, “Redistricting and Representation,” available at (last accessed March 2019).
  49. Gary King and Robert X. Browning, “Democratic Representation and Partisan Bias in Congressional Elections,” American Political Science Review 81 (4) (1987): 1252–1273, available at
  50. Beverly R. Gill et al. v. William Whitford et al., brief of amici curiae Heather K. Gerken and others, U.S. Supreme Court, No. 16-1161 (August 30, 2017), p. 13, available at
  51. Ibid, p. 21.
  52. American Academy of Arts and Sciences, “Redistricting and Representation.”
  53. For the People Act of 2019.
  54. The For the People Act would require such a system, as does California’s Citizens Redistricting Commission. All About Redistricting, “California,” available at (last accessed March 2019).
  55. It is conceivable, and perhaps even desirable, to make the legally required criteria so specific that the redistricting process could be nearly automated. However, even with a fully automated process, an independent commission could serve a useful purpose in protecting the process from tampering and providing cross-partisan legitimacy to the results. For more discussion of the possibility of automated redistricting, see Samuel Issacharoff, “Judging Politics: The Elusive Quest for Judicial Review of Political Fairness,” Texas Law Review 71 (1993): 1643–1703, available at
  56. For example, one recent article imputed vote shares for unconstested districts “using a model based on presidential election results, lagged legislative results, and incumbency status.” Christopher Warshaw and Nicholas Stephanopoulos, “The Impact of Partisan Gerrymandering on Political Parties” (2019), p. 11, available at
  57. Lee Drutman, “This voting reform solves 2 of America’s biggest political problems,” Vox, July 26, 2017, available at; Matthew Yglesias, “Proportional representation could save America,” Vox, October 15, 2018, available at; Madeline Brown, “A Strong Argument for Multi-Member Districts & Ranked Choice Voting in Georgia,” FairVote, July 25, 2017, available at
  58. FairVote, “The Fair Representation Act,” available at (last accessed March 2019).
  59. Notably, six current members of the U.S. House of Representatives who would be willing to take this leap are Reps. Don Beyers (D-VA), Jamie Raskin (D-MD), Ro Khanna (D-CA), Jim Cooper (D-TN), James McGovern (D-MA), and Scott Peters (D-CA), who sponsored or cosponsored the Fair Representation Act, which would create such a system for U.S. House elections. Fair Representation Act, H.R. 3057, 115th Congress, 1st sess. (June 26, 2017), available at

The positions of American Progress, and our policy experts, are independent, and the findings and conclusions presented are those of American Progress alone. A full list of supporters is available here. American Progress would like to acknowledge the many generous supporters who make our work possible.


Alex Tausanovitch

Former Senior Fellow