When drawing boundaries of electoral districts, officials often try to respect communities of interest (COIs) by keeping them intact. COIs can be defined socioeconomically, such as an area where a particular ethnic group concentrates, or cognitively, such as a neighborhood or region that people commonly agree upon. My research focuses on the nature and scale of the latter type, called cognitive COIs. I investigate whether people conceive of different scales of cognitive COIs when they are exposed to different map extents of their local area. If they are indeed different, that would point to the existence of different scales of cognitive COIs. I seek to answer this question through two studies. The first splits subjects into three groups and exposes each group to a different map extent; they then draw on the map where they think their COI is located. I find that the size of the COI that they draw depends greatly on map extent. The second study exposes subjects to two different map extents, but all of them receive the same two; they then rank places on the map according to how confident they are that they are in their COI. I likewise find that the size of the COI that they define by their rankings depends on the map extent. These findings indicate that the map induces people to externalize their COI at different scales, confirming that multiple scales of cognitive COIs exist and officials must be aware of that fact when they are redistricting at different levels of government.

When deciding where to draw the boundaries for electoral districts, officials often strive to ensure that communities of interest are not split up but kept together within a single district. What exactly constitutes a community of interest is somewhat vague, with legal and academic sources describing either a thematic region with shared demographic and land use traits or a cognitive region that is meaningful to people and commonly agreed upon. This research seeks to identify communities of interest at the sub-city level in both the thematic sense—by clustering Census tracts and land parcels according to classes of important variables—and the cognitive sense—by surveying residents about the location and extent of their community and finding areas of highest agreement. Then the degree to which the two senses of communities overlap is assessed; the more overlap, the more evidence there is that the two definitions correspond closely. Finally, the amount of overlap between the different communities and existing electoral districts is determined, to see which of the two types receives more attention from boundary drawers. The study finds that the two types of communities of interest correspond relatively well, and that the thematic type corresponds with the electoral districts better than the cognitive type.