Voted! Please get out there and do the same!” by Aaron Gustafson is licensed under CC BY-SA 2.0.

I’m a last-minute voter, so I avail myself of King County’s ballot drop boxes. In the waning days of an election cycle, a ballot dropped in a mailbox might not get postmarked in time to count. Last election day, I was looking for a drop box near my place of work in downtown Seattle. I was surprised to find none in what I would consider the core of downtown. The walk down to South King Street and 2nd Avenue South wasn’t onerous, but it left me with a question.

How well does King County place ballot drop boxes with respect to transit access? There are 24 drop boxes in Seattle—not enough so that every point in the city is within walking distance of one. The elections department recognizes that at least some people will have to use some other mode of transportation. Some boxes are configured for drive-up access. For those who don’t drive, or who find the act of delivering a single piece of paper with a multi-ton private vehicle absurd, transit-friendly placement matters.

To assess the effectiveness of the placement, I conducted an access analysis for a typical weekday in Seattle. This involves dividing the city into a grid of 80 meter by 80 meter sectors. For the combination of every minute of the day and every sector, the analysis calculates which other sectors can be reached within 30 minutes using transit and walking. These calculations produce two sums for every sector: the inbound sum (with this sector as a destination, how many origin sector-time of day pairs are there for which a 30 minute transit trip exists) and the outbound sum (with this sector as an origin, how many destination sector-time of day pairs are there for which a 30 minute transit trip exists). This enables the comparison between sectors containing ballot boxes and all other sectors.

sector box name inbound × outbound percentile fraction of best
20428 University of Washington Campus 103,782,920,730,372 100 0.668
15021 Seattle Central College 78,238,316,077,403 99 0.504
13377 King Street Box 75,402,031,675,023 99 0.485
13282 Uwajimaya 72,954,743,897,133 98 0.470
11328 Beacon Hill Library 71,583,672,817,977 98 0.461
15392 South Lake Union 47,621,906,594,130 92 0.307
28210 North Seattle College 47,216,156,421,431 92 0.304
25037 Green Lake Community Center 45,716,837,283,832 91 0.294
31070 Lake City Library 38,164,457,011,340 84 0.246
23042 Ballard Branch Library 37,180,946,549,144 83 0.239
15777 Metropolitan Market - Queen Anne 35,387,462,349,230 80 0.228
8623 Rainier Community Center 33,288,062,735,543 77 0.214
8536 Alaska Junction 31,670,637,604,320 75 0.204
14100 Garfield Community Center 28,634,833,968,970 69 0.184
4769 NewHolly Neighborhood Campus 27,650,251,459,488 68 0.178
6255 High Point Library 25,093,548,459,318 63 0.162
2599 Rainier Beach Community Center 22,293,899,014,120 57 0.144
19245 Seattle Pacific University Bookstore 21,485,026,822,588 55 0.138
25375 Magnuson Park/The Brig 18,136,946,136,417 48 0.117
31348 Broadview Library 16,524,923,956,240 44 0.106
6140 South Seattle College 14,194,137,603,597 39 0.091
18870 Waterway 19 Park 13,690,627,228,035 38 0.088
2778 South Park Library 9,858,624,229,792 28 0.063
17060 Magnolia Park 7,067,602,495,198 21 0.046

The inbound × outbound column is the product of the two sums mentioned previously, and serves as the measurement that compares the reachability of sectors. A percentile rank provides one perspective as to how the sector compares to all sectors in the city. The fraction of best serves as another, comparing the sector’s inbound × outbound measurement to that of the most reachable one, sector 21259, located near U District Station.

I was immediately drawn to the high percentile rank of several of the drop box locations. This included a 99th percentile ranking for the King Street drop box that struck me as transit-sparse compared to the downtown core. The fraction of best measurement provides a contrast to the percentile rank that helped resolve this disconnect. Even the most reachable ballot box, which falls in the top 1% of sectors, has only around two thirds of the reachability of the most reachable sector. The King Street ballot box isn’t even within 50% of that one. As an additional comparison, the ballot box’s sector has 47.2% of the reach of the best sector for the downtown core, sector 14126, located on 3rd Avenue between Union Street and University Street. Other downtown core sectors are comparably well-reachable, with eight of the top ten located there. The data indicate that some sectors are exceptionally reachable, but the ones containing ballot boxes aren’t among them.

Does this analysis give any insight into a better distribution of ballot boxes? It wouldn’t be useful to just consider exceptional sectors and place all the ballot boxes in them. Most highly-reachable sectors are clustered geographically. A better analysis for evaluating ballot box placement might not focus on the reachability of each ballot box-containing sector, since reaching any ballot box suffices for voting, and instead look exclusively at how many sector-time of day combinations can reach and return from any ballot box within 30 minutes. But that analysis wouldn’t help determine where ballot boxes should be placed; it could only be used to evaluate proposed changes.

That’s where I see value in this analysis. It can highlight missed opportunities in the current placement that may have easy fixes. It allows proceeding from the observation that the lack of a downtown core ballot box is strange, to demonstrating that choosing a location there would provide a drop box that is twice as reachable as the current alternatives. It’s understandable that this would be difficult to see otherwise; a traditional transit map only hints at what areas are easily reached. If reaching ballot boxes by transit is something that King County values, and I suspect and hope that it is, using access analysis could illuminate a path towards a better placement of ballot boxes. I can say for certain at least one last-minute voter would be thankful.