Multimember Districts: How to Fix the Gerrymandering Problem

Robert A. Levy

One recommended goal in conducting congressional elections is to approximate proportional representation—where the percentage of Republican vs. Democrat vs. “Other” Congress members resembles the mix of registered voters in the state. But under our current structure of single-member districts (SMDs)—one congressional seat in each district—that’s unlikely to happen. If voters have allegiance to their party, Democratic votes won’t matter in a Republican district, and vice versa. Fortunately, the Constitution doesn’t require that structure. Yes, there’s a 1967 federal law that directs states to elect representatives from SMDs, but that’s a statutory requirement, not a constitutional one. Congress could change the rules and prescribe multimember districts (MMDs), which would go a long way toward reducing the partisan advantage of gerrymandering.

The simplest system works like this: Consider Florida, with its 28 congressional representatives—1 per district in 28 districts. First, let’s substitute MMDs for SMDs—for example, 4 representatives per district in 7 districts. The candidates for those seats can be chosen by primaries, caucuses, or other means, as specified by the state (or perhaps by parties with delegated state powers). In some states, voters will simply choose a party; in other states, voters will select individual candidates. Non-affiliated candidates might qualify by petition.

Next, let’s consider a single Florida district with, say, 6 candidates—2 Democrats, 2 Republicans, and 2 Independents—vying for the 4 seats. And let’s assume that voters in the district cast their ballots along party lines. In Florida overall, that would mean 38 percent vote for Republicans, 33 percent for Democrats, and 29 percent for “Other.” In our sample district, Republicans would get 2 seats (0.38 x 4 = 1.52); Democrats would get 1 seat (0.33 x 4 = 1.32); and Independents would get 1 seat (0.29 x 4 = 1.16). In other words, both of the Republican candidates would be elected; the Democratic candidate with the most votes would be elected; and the only Independent candidate would be elected. Those results approach proportionality to voter registration.

Under MMDs, all voters can affect election results by affecting their party’s representation. MMDs also encourage coalition building (e.g., multiparty endorsements), reduce polarization, restrain gerrymandering, and increase competition.

Of course, the fewer the districts (i.e., the more seats per district), the closer the results will be to proportional representation. On the other hand, larger districts generally mean less ability for candidates to represent the interests of their constituents. Somewhere between three and six seats per district sounds to me like a reasonable compromise.

Finally, there are analytically superior (but more complex) MMD systems that use ranked-choice voting with the transfer of surplus votes. For sample legislation, see the Fair Representation Act.