In a recent New York Times essay, I argued that, among the many different tasks for political reform, one of the most urgent is mitigating the forces of political extremism so dangerous to American democracy. The four areas I singled out in which political reforms could potentially have this effect were the presidential nomination process; the structure of primary elections; regulating money in politics; and designing election districts in ways that create meaningful political competition.
With the Census Bureau having released the 2020 population numbers, which trigger the start of states’ redistricting process, this is the moment to focus on the importance of emphasizing political competition when the states create their new maps of congressional districts. Competitive districts have generally been defined as those in which the winning candidate receives 55% of the vote or less. (Given increasing partisan loyalty among voters, which means fewer voters shift back and forth between the parties, swing districts today might have to include no more than 53% of likely voters for one party to be competitive.)
In “safe seats,” by contrast, the district is so dominated by the voters for one party that its candidate is going to win the general election almost regardless of a particular candidate’s ideology and qualifications. As a recent report from the Unite America Institute notes, in the 2020 election, 83% of candidates to the U.S. House were elected from safe seats. Only 17% of members were elected from seats in which they faced significant competitive pressure from the other party’s candidate – a dramatic decline from the late 1990s, when 38% of districts were competitive.
The dynamics and incentives for candidates running in competitive districts are dramatically different than those candidates face in safe districts. For competitive seats, candidates know they must win over enough voters in the center, who might swing to either party, while also holding on to their base. If they move too far toward the wings of their party, they risk losing the centrist voters needed to win. These dynamics not only weaken the power of the wings of each party but also create incentives for more centrist candidates to run in the first place. Competitive districts also make legislators more responsive to changes in voter preferences; if public opinion on important policies shifts by five points, legislators who want to keep or gain office will need to respond to that change.
In safe seats, by contrast, incumbents’ main threat comes from being “primaried.” Given the make-up of primary voters, that threat mostly comes from the ideological poles of their party. As the authors of a recent book titled “Rejecting Compromise: Legislators’ Fear of Primary Voters” find: “Legislators believe that primary voters are much more likely to punish them for compromising than general election voters or donors.” This point was made in characteristically bald form in Donald Trump’s Jan. 6 speech on the Ellipse when he mainly attacked Republicans who would not vote to object to the counting of electoral votes: “[I]f they don’t fight, we have to primary the hell out of the ones that don’t fight. We primary them.”
When the major competitive threat that safe-seat incumbents face is from the ideological wings of their party, little incentive exists to appeal to the center of the political spectrum. Primary voters in safe-seat districts also have less incentive to compromise and back moderate candidates because the party’s candidate will win the general election regardless of where they fall along the party’s ideological spectrum; In turn, moderates are less likely to run in the first place, because they know they can’t survive the primary electorate. The overwhelming dominance of safe seats in Congress contributes significantly to a Congress that is much more polarized than are citizens themselves.
Over the last decade, gerrymandering has received broader public attention than ever before. And the political system is responding, at least to some extent. Through direct democracy, voters in a number of states have adopted independent commissions to take over redistricting. A centerpiece of the massive voting-reform bill that has passed the House and is now before the Senate are provisions that would require independent commissions to do redistricting and would impose standards of “partisan fairness” on plans. These are moves in the right direction; I have long argued that we should take redistricting out of the hands of the most self-interested actors – partisan legislatures – and use independent commissions.
But the substantive standards in H.R. 1/S. 1 risk giving short shrift to the importance of competitive districts. A plan can be fair in terms of the partisan outcomes it is likely to generate and yet still result in every officeholder being elected from a safe seat. Indeed, when government is divided and both parties have a say in redistricting, it’s not uncommon to see “sweetheart” gerrymanders, in which politicians from both parties agree to give one another safe seats. Incumbents want to be in safe seats – the safer the better (this can be in tension with the party’s goal to maximize its prospects overall). But that’s not good for creating a Congress actually capable of governing, one in which members have incentives to compromise and forge the coalitions needed to enact most major legislation.
To be sure, it’s not possible to make all districts competitive. The Voting Rights Act requires, where voting is racially polarized, the creation of districts likely to elect the preferred candidate of the minority community. Those districts are generally safe seats for Democrats. In addition, given the extent to which voters have sorted themselves geographically along partisan lines, creating competitive districts would in some areas might not be possible or would require unusually shaped districts. But imagine if just one-third of Congress, rather than the current 17%, were elected from competitive districts. The dynamics on Capitol Hill would change considerably.
Some academics, however, dissent from the view that competitive districts marginalize ideological extremism and foster moderation in Congress. This disagreement exposes a remarkable disjuncture between these political scientists and journalists who actually cover Congress. Stories with headlines such as “House Democrats in Swing Districts Are Torn Over Impeachment” are common. Beyond the headlines, stories covering Congress and state legislatures routinely describe the more moderate positions of members from swing districts compared to those in other seats. That there is often a difference between the ideological views of members from swing districts and others is a simple matter of fact among those who cover Congress most closely. Indeed, legislators themselves strongly believe this, as John Boehner’s recent memoir describes in detail.
Yet the view among this dissenting group of political scientists in recent years is exactly the opposite: that it does not matter whether members are elected from competitive seats or safe seats, because all members of each party vote similarly, regardless of the type of district from which they are elected (see here and here for two important studies). How can there be such a disconnect between those who cover Congress up close and certain social scientists who survey Congress from a greater distance?
One possible answer: These political scientists define the ideology of members of Congress or state legislatures by aggregating all (and only) the roll-call votes taken on bills. This is the easiest data point to measure, but as with quantification in general, basing analysis on the dimensions easiest to quantify can distort reality. For many years now, most legislation has not been put together at the committee level but in leadership-led centralized processes. Behind-the-scenes negotiations among party members take place at this level, when party leaders broker a package that best accommodates the distinct interests within the party. This centralized process also enables party leaders to logroll across bills, giving resistant members on one bill what they want in another bill, thus bringing them along to support both measures. By the time a bill is put on the floor for a vote, the party has largely united behind that package. And if party leaders can’t get sufficient party consensus on a bill, they never put it to a floor vote.
Counting up only the final roll-call votes obscures the differences in policy preferences between centrists and the wings of the party. As congressional scholars Frances Lee and James Curry put it, “the roll-call vote is censored”; the reality is that the parties “contend with much more intraparty conflict than one might expect from roll-call votes.”
In addition, much of roll-call voting in the modern Congress is designed as party messaging. These are votes taken to sharpen and highlight major party differences, rather than to support bills that have a realistic prospect of being enacted. As Lee documents, “Very little actual legislation becomes law by narrow or partisan majorities, but the Congress nevertheless takes many roll call votes that pit one party against the other.” Not surprisingly, these messaging roll-call votes display a high degree of party unity – that is part of their point, after all. Yet when political scientists combine all roll-call votes into a single number – without distinguishing bills on minor versus major issues or bills that are purely messaging legislation – they inevitably obscure genuine differences within the parties on significant legislation.
Indeed, sophisticated new work in political science is already starting to undercut the view of those analysts who doubt that safe seats foster extremism. Rather than treat roll-call voting on all issues as the same, Professors Brandice Canes-Wrone and Kenneth Miller focus on only the most significant bills a given Congress votes on, which in their study range from two to six bills a year. On these bills, they find that members in safe seats respond far more to their most polarized donors than do members in competitive districts. This is true for a broad set of issues, including capital gains taxes, partial-birth abortion, the Affordable Care Act, or other highly salient issues.
More specifically, Canes-Wrone and Miller find, first, that national donors are much more polarized than donors from within a member’s district. They then find that when seats are competitive, representatives respond much more to the preferences of their constituents, but that when seats are safe, representatives are more responsive to the preferences of this highly polarized national donor class. In other words, in safe seats, members can defect more from their district’s preferences and endorse the more extreme positions of their national donors. This is not surprising: If you’re going to win a safe seat with 75% of the vote, you have a lot of slack to satisfy your national donors with positions that your constituents don’t support, even if your victory margin winds up dropping by 10 points next time. But if your district is competitive, you can’t afford to stray much from the preferences of your constituents. This is why I have also argued that to reduce extremism, a public financing system that provides $6 for every $1 a candidate raises from small donors – another aspect of H.R. 1 – should limit those matching funds to small donors from within a member’s district.
In sum, those who cover Congress most closely, congressional members themselves, and some cutting-edge work in political science all confirm that members from safe seats tend to be more extreme than those elected from competitive districts. One way to counter political extremism, as new districts for Congress and state legislatures get drawn across the country, is to insist on the importance of creating competitive districts.
Richard H. Pildes is the Sudler Family Professor of Constitutional Law at NYU School of Law.