Multi-member Districts
In some states, legislative districts elect multiple members. In Washington State, for example, each district elects 2 representatives to the lower house, but since each is elected separately, it’s essentially a single-member system, where each voter elects 2 members to the lower house. The Maryland lower house, however, has 71 districts, each of which elects between 1 and 3 members, for a total of 141 seats. Each district’s population target is commensurate with the number of seats, roughly 44,000 per seat.
For a while, DRA has allowed users to build these kind of multi-member districts, but most of our analysis doesn’t apply to these kinds of redistricting plans. In addition, our statistics show partisanship and minority representation based on districts, regardless of how many seats they have. One of the problems that prevents giving better analysis and statistics is that they all depend on the voting system used to elect the members in multi-members districts. If each member is elected separately by the same voters, chances are they will all be Democrats or all be Republicans and a minority will have as much opportunity to elect a representative of its choice as in any single-member system. But a different voting system could result in a split between the two parties, and can give a minority a greater opportunity.
FairVote, a national non-profit working toward fairer representation and better government, advocates for a system call Proportional Ranked Choice Voting (PRCV). And now in DRA you can view statistics assuming this system is used! Let’s explore how.
I created a theoretical map for the Washington State lower house. The map has 20 districts, 18 of which have 5 seats and 2 of which have 4 seats, for a total of 98, the current number of seats in the lower house. I was totally winging it, so I make no claims about whether it’s a good map or not, but this gives us an example of how to create a multi-member district map and understand its potential partisanship and minority representation using PRCV.
To create a map like this, start with New Map and select a state and plan as you normally would. Then select the District Count and check Allow MMD.
This shows you a # Members column that you can edit to set the number of members in each district. If Equal Population Per Member is checked, the target population will be adjusted as you modify the members.
Once you Apply, you’re ready to paint your map. You’ll notice that when you hover over the district ID in the District Selector, the tooltip shows you the number of members. The district label shows you that as well.
Now, as you work on your map, note that the District Partisan Lean and District Demographics shown the same thing as for single-member districts — district statistics, not member statistics.
When you are ready, click the Statistics tab. By default, you’ll still see district statistics. But, click the checkbox in the upper left to Show statistics using proportional ranked choice voting and the table will change.
Proportional ranked choice voting is a system where each voter simply ranks the candidates in their district. If there are 5 seats, then the top 5 candidates, using ranked choice voting, win the 5 seats. That means the threshold for winning a seat is 1/(n + 1), where n is the number of seats. To estimate the likely partisan seats, we use the formula:
#seats = floor((pct - 0.02)/thresh), where thresh = 1/(n + 1) + 0.001 and pct = fraction of party votes
So, for this map, we would estimate a 54D–37R state house, with 7 tossups (or others). Given that the state voted 57% Dem and 41% Rep, using our default election composite, that’s pretty darn proportional! We also calculate minority opportunities using the formula:
#opportunities = floor((pct - 0.03)/thresh), where pct = fraction of minority VAP
These calculations of partisan seats and minority opportunities are very much untested at this point. We expect to get more information in the years ahead to refine these, and to provide additional analytics around multi-member districts.
You may have also noticed that most of our analytics are not shown for multi-member districts. This is because those measurements were developed for single-member districts, so they are not valid for multi-member, at least as far as we know. In the future, we expect to introduce new measurements appropriate for multi-member districts. But for now, we hope this lets you explore this exciting democracy reform!