Gerrymandering. Stephen K. Medvic

about representation is to consider whether individual legislators are representative of their constituents. This is commonly referred as “dyadic representation,” because it compares one representative to one constituency.⁴⁴ Many would argue that ideological and/or partisan agreement between the representative and his or her constituents is of paramount importance. Indeed, from this perspective, maximizing such agreement ought to be the goal of those who draw district lines. In districts that are gerrymandered to be safe for one party, most voters will share that party affiliation and are likely to be satisfied with their representation.⁴⁵ One might object that representatives from such districts are less likely to be responsive to the voters or are less likely to be held accountable by them. The response to such an objection is that accountability is maintained through primary elections.

      From another perspective, the creation of safe districts amounts to rigging the system. A common refrain from this side of the debate is that when gerrymandering is allowed to occur, politicians are picking their voters before voters can pick their politicians. From this point of view, then, the goal is competitive elections. District lines should be drawn in such a way as to maximize competition in as many districts as possible.⁴⁶ Of course, given that only one representative will be elected per district, a competitive district in which preferences are split roughly 50-50 will mean that about half the constituency will always be unhappy with its representation. Be that as it may, competitive elections are thought to be more effective in holding incumbents accountable, giving voters a meaningful choice over their representation, and generating more excitement and, consequently, more participation than uncompetitive elections. These effects, it is argued, are vital for a healthy democracy.

An alternative to dyadic representation is “collective representation.” In this formulation, what matters is whether legislators, taken as a whole, accurately reflect the aggregate policy preferences of the public.⁴⁷ If a majority in a given state wants to increase the minimum wage but the legislative majority opposes such an increase, the public would lack collective representation on this issue.⁴⁸

      There are those who would argue that the question of whether an institution, like Congress, is representative in a collective sense is largely irrelevant. The American system of government, they argue, is not designed to achieve collective representation at least in majoritarian terms. Features like the separation of powers, bicameralism, and federalism mean that it is purposefully difficult to translate majority preferences into policy. Representation in institutions like the US Senate and the presidency (via the Electoral College) is geographically based, which can easily distort policy preferences. In the House of Representatives, single-member districts with plurality winners mean that the relationship between the seats a party wins and the votes it receives will be distorted, even without conscious gerrymandering.⁴⁹ Thus, perfect correspondence between the policy preferences of the public as a whole and of the legislators collectively (let alone the policy output of the legislature) is, by design, nearly impossible to achieve.

Others believe a significant level of collective representation is necessary if the people are to rule in any meaningful way. Legislatures should produce policies that the electorate desires. When they don’t, they are not providing the responsiveness that democracy requires. There are many reasons why the public’s policy preferences might not be reflected in actual public policy. The many “veto players” in the American system (e.g., the president, majorities in the House and in the Senate, and filibuster-wielding minority in the Senate) are a key reason for the disconnect. But another is that the partisan makeup of the legislature might not match the partisan preferences of the electorate (partisanship being a shorthand indicator of a set of policy preferences). For those who believe collective representation is indispensable for democracy, disproportionate election outcomes, like those mentioned at the outset of this chapter, are prima facie evidence that election processes are undemocratic. And while disproportionate outcomes are always possible in a system of single-member plurality elections, they are made more likely by partisan gerrymandering.

      It is, of course, possible for individual representatives to do a good job representing the views of most of their constituents (dyadic representation) while the legislature as a whole does not represent the views of the public very well in the aggregate (collective representation). The reverse may also be true. Tables 1.1 and 1.2 offer hypothetical illustrations of how representation can go awry. In table 1.1, a clear majority in each district share conservative views with the representative from the district. This is good dyadic representation.

Table 1.1 Hypothetical Example of Good Dyadic (District-Level) Representation but Poor Collective Representation

      Note: Numbers in the last three columns indicate the percentage of the district that holds a given ideological point of view. Check marks indicate the point of view held by the district’s representative.

Table 1.2 Hypothetical Example of Poor Dyadic (District-Level) Representation but Good Collective Representation

      Note: Numbers in the last three columns indicate the percentage of the district that holds a given ideological point of view. Check marks indicate the point of view held by the district’s representative.

      Collectively, however, the population is not well represented as nearly a third of the electorate is liberal and yet not a single representative holds liberal views. In table 1.2, the opposite problem exists. Collective representation is very good as the even split between conservative and liberal representatives (five each) mirrors the breakdown of conservatives and liberals in the population (48.5 percent of the population in each category). District-level representation is not particularly good, however, because the representative from each district represents a view that commands less than a majority.

As contrived as these hypotheticals may seem, results like these do exist. In Connecticut, following the 2018 midterm elections, Republicans held none of the five seats in the US House of Representative even though they received 38 percent of the vote statewide (as in table 1.1). Similarly, Democrats won 35 percent of the House vote in Arkansas but took none of the state’s four seats. A real-world example of table 1.2 is harder to find but the results of Minnesota’s congressional elections in 2018 are close. In five of the eight House districts, the winner garnered less than 56 percent of the vote leaving a very large portion of the electorate (roughly 44 to 49 percent) presumably unhappy with its representation. Statewide, however, the division of seats (five Democratic and three Republican) is reasonably close to the statewide division of votes (55 percent Democratic and 44 percent Republican).

      Regardless of their real-world applicability, the hypotheticals serve to illustrate the fact that dyadic and collective representation do not necessarily occur simultaneously. They can, of course, but even the most neutral district boundaries will not automatically produce robust levels of both types of representation. Importantly for our purposes, these hypotheticals also test the priorities of the two sides of the redistricting debate. If forced to choose one of these two scenarios as the most preferable, those in the realpolitik camp are more likely to prefer table 1.1 to table 1.2 while