How Do You Measure Distance in Spatial Models? An Example Using Open-Space Valuation

Abstract

Spatial distance is a critical component of theories across the social, natural, and information sciences, but too often the methods and metrics used to describe spatial distance are implicit or underspecified. How distance is measured may influence model results in unanticipated ways. We examined the differences among distances calculated in three ways: Euclidean distances, vector-based road-network distances, and raster-based cost-weighted distances. We applied these different measures to the case of the economic value of open space, which is frequently derived using hedonic pricing (HP) models. In HP models, distance to open space is used to quantify access for residential properties. Under the assumption that vector-based road distances better match actual travel distance between homes and open spaces, we compared these distances with Euclidean and raster-based cost-weighted distances, finding that the distance values themselves differed significantly. Open-space values estimated using these distances in hedonic models differed greatly and values for Euclidean and cost-weighted distances to open space were much lower than those for road-network distances. We also highlight computational issues that can lead to counterintuitive effects in distance calculations. We recommend the use of road-network distances in valuing open space using HP models and caution against the use of Euclidean and cost-weighted distances unless there are compelling theoretical reasons to do so.

Perception vs Reality: Is Perceived or Objective Proximity to Physical Activity Opportunities in the Environment More Associated With Recent Use Among Adolescent Girls?

“…Straight line and road network distances (in meters) were calculated for each of the nearest physical activity opportunities from the girl's home address. Only road network distance is presented in these analyses because it is the most feasible route most girls would travel to get from her home to the closest physical activity opportunity (Sander, Ghosh, van Riper, & Manson, 2010).…”

Measuring geographic access to health care: raster and network-based methods

“…In the raster data model, the strict topology that governs real world travel along roads is replaced by predefined directional movement among cells. Thus, in routing applications, the raster data model has the potential to produce unexpected results [33,34]. Furthermore, travel time estimates may be either overestimated or underestimated depending upon the geometric complexity of the road network and the cell resolution.…”

Cycling willingness: Investigating distance as a dependent variable in cycling behavior among college students

“…Table 1 shows further descriptive statistics for the distances from a student's home to the closest bike-friendly network segment (DIS2PATH), closest distance along the network to the major bike path network along the river (DIS2GREEN), and the total distance from named intersection to the campus library (DIS2CAM). These measures are distinct from most of those in the reviewed literature, which uses a straight line, or categories of distances (Krizek & Johnson, 2006;Millward et al, 2013;Sander et al, 2010). Using a network gives the benefit of topographic and built environment barriers (i.e., near-by interstate or large river) being represented in this continuous measure.…”