Does the inclusion of mapped estimates of attribute uncertainty for demographic estimates change spatial decisions made by urban planners?
1Griffin, A.; 2Spielman, S.; 3Jurjevich, J.; 4Merrick, M.; 5Nagle, N.N.; 6Folch, D.C.
Recent changes to the US Census have led to more timely updates of demographic statistics that are used in the delivery and planning of many social and environmental programs. However, this timeliness has a tradeoff: increased uncertainty in the estimates for small area geographies such as census blocks and tracts. Although the Census Bureau publishes information about the uncertainty of the estimates, few end users engage with and utilize this information, perhaps because it comes in a difficult to use form; another column in a table with many columns. Many techniques for visualising uncertainty in attribute data have been proposed, but few have been empirically tested, and fewer still with real end users using an ecologically valid task. We undertook a context-of use study with urban planners to find out how they used demographic estimates from the American Community Survey (ACS). These findings were then used to design an experiment undertaken with 55 urban planners in which they had to make spatial decisions using uncertain demographic estimates. We compared visualisation methods derived from two metaphors for communicating uncertainty: the stoplight and sketchiness. The experimental task required planners to define an area of contiguous census tracts that meets a particular threshold with respect to the attribute in question: percentage of households in poverty. We found that the addition of uncertainty information did alter the spatial decisions planners made. Moreover, planners were more likely to change their decision when the did not have knowledge of the place than when they did have knowledge of the place.