A significant problem with using administrative districts for mapping results from the great discrepancies in their physical size and shape. For example, counties in the eastern United States tend to be small and regular, but the further west you get, the larger and more variable counties become. The issue is one of interpretation, as the eyes are naturally drawn to the shading of large areas on the map and can easily lead the reader to give undue weight to them. Even if mapping is done at the block group level, the variance in physical size has significant impacts on the interpretation of a map pattern. In most urban areas, the block groups tend to be larger as one moves away from the urban core, and this can result in the user perceiving a ‘donut’ effect where none really exists.
These visualization issues can be overcome in many cases by using geographic areas of a fixed size – such as regular square grids or hexagons. Square grids are the easiest to implement and can often result in greatly improved map interpretation.
AGS has created a grid cartography level using one one-hundredth of a degree latitude/longitude intervals which are available for any variable and have been implemented as an output geography in our Snapshot API. These are excellent for displaying data at a metropolitan or regional scale especially for the continental United States. While they are not square or even equal area, the distortions at a regional level are minimal. Grids are named by the southwest coordinates of the cell, such as ‘3437N-11884W’.
Data are estimated from the block group level using the AGS block allocation table, which differentiates between three key patterns – historical census, current year demographic estimates, and daytime/workplace estimates. The use of a separate allocation table for business establishments is critical since the spatial patterns of residential versus business demographics vary considerably within block groups.
Grid geography is also particularly useful for quickly aggregating data, as grid naming and numbering conventions make it simple to directly compute the addresses of adjacent cells. For applications like site filtering and selection, the use of grid cells can greatly improve performance by allowing approximate trade area demographics to be calculated without using distance calculations.
The practical application of grids is readily demonstrated. The first map shows a block group distribution of median household income for Salt Lake City, UT. Note that the first place where the eyes are drawn is to the large relatively low-income areas towards the edges of the map, or, equally dangerously, to the large high income area in the Park City areas. Both have the effect of visually obscuring the great amount of detail in the densely populated areas of the city.
As a contrast, the same shading pattern is displayed using the grid cells, resulting in a significantly different visual impression and draws the eyes towards the populated area from Ogden south to Payson. It also clearly has an effect of demonstrating the extent of the area where people actually live, since grid cells with no value are suppressed.