Uncertain geographic context problem

The uncertain geographic context problem or UGCoP is a source of statistical bias that can significantly impact the results of spatial analysis when dealing with aggregate data.^[1][2][3] The UGCoP is closely related to the Modifiable areal unit problem (MAUP), and like the MAUP, arises from how we divide the land into areal units.^[4][5] It is caused by the difficulty of understanding how phenomena under investigation (such as people within a census tract) in different enumeration units interact between enumeration units, and outside of a study area over time.^[1][6] It is particularly important to consider the UGCoP within the discipline of time geography, where phenomena under investigation can move between spatial enumeration units during the study period.^[2] Examples of research that needs to consider the UGCoP include food access and human mobility.^[7][8]

The uncertain geographic context problem, or UGCoP, was first coined by Mei-Po Kwan in 2012.^[1][2] The problem is highly related to the ecological fallacy, edge effect, and Modifiable areal unit problem (MAUP) in that, it relates to aggregate units as they apply to individuals.^[5] The crux of the problem is that the boundaries we use for aggregation are arbitrary and may not represent the actual neighborhood of the individuals within them.^[4][5] While a particular enumeration unit, such as a census tract, contains a person's location, they may cross its boundaries to work, go to school, and shop in completely different areas.^[10][11] Thus, the geographic phenomena under investigation extends beyond the delineated boundary .^[6][12][13] Different individuals, or groups may have completely different activity spaces, making an enumeration unit that is relevant for one person meaningless to another.^[7][14] For example, a map that aggregates people by school districts will be more meaningful when studying a population of students than the general population.^[15] Traditional spatial analysis, by necessity, treats each discrete areal unit as a self-contained neighborhood and does not consider the daily activity of crossing the boundaries.^[1][2]

The UGCoP has further implications when considering the area outside of a study area. Tobler's second law of geography states, "the phenomenon external to a geographic area of interest affects what goes on inside."^[16][12] As a study area is often a subset of the planet, data on the edges of the study area will be excluded.^[17] If the boundary demarcating the study area is permeable to travel, then the phenomena under investigation within it may extend beyond, and be impacted by, forces excluded from the analysis.^[6][18] This uncertainty contributes to the UGCoP.^[1][2]

All maps are wrong, and a cartographer must ensure that their maps' limitations are well documented to avoid misleading the users.^[19] With modern technology, there is an emphasis on individual-level data and understanding how individuals interact with their environment.^[5][8] When making maps with this individual-level data, the UGCoP is one source of bias that can impact the results of an analysis.^[1] When these results inform policy, they can have real world ramifications.^[19]

The UGCoP is particularly important when understanding food access and human mobility.^[6][7]

Geographic information systems, along with technologies that can monitor the position of individuals in real time, are possible methods for addressing the UGCoP.^[2] These technologies allow scientists to analyze and visualize the 3D space-time path of people moving through a study area, and better understand their actual activity space.^[2] Web GIS has also been employed to address the UGCoP by allowing researchers to better contextualize subjects' real and perceived activity space.^[2][15] These technologies have helped to address the problem by moving away from aggregate data and introducing a temporal component to the modeling of subject activity.^[2][15]