Accessibility Through Accountability

Application: Accessibility is Comparable and Consistent

MnDOT Performance Measure Dashboard
Figure 4.1: Snapshot from MnDOT Performance Measure Dashboard, displaying statewide average accessibility by auto over time.

A strength of accessibility metrics is that the numbers are strictly comparable across places, times, modes, and people, given the relatively straightforward assumption that time is a universal cost. For instance, calculating accessibility as the number of jobs reachable in 30 minutes results in integer counts that are on the same scale whether traveling by biking, transit, or private auto. A number ten times as high in one mode over another, means ten times the opportunities that are reachable at the same time cost. Examining accessibility for the same mode in the same place over time, the resulting timeseries is easily understood to reflect positive or negative change, with a magnitude of change in integer units. Looking within a geographic region for a single mode, accessibility metrics can highlight where higher or lower origins of access occur. This is a small precursor to a more powerful analysis of who lives in these origins, and thus who benefits from changes (and who does not). Here the advantage of accessibility metrics over traditional metrics of vehicle flow on segments is obvious, as the deleterious effects of congestion are assumed to apply equally to all residents of a region, whether their access is influenced by a particular slow roadway or not.

Accessibility is originally calculated as a location metric, as a property of an origin place. The value of access, however, is only realized when it is experienced by people. To reflect this fact, aggregation metrics of access are averaged across origin blocks, with each block’s contribution weighted by the number of workers living in that residential block. The resulting metric represents the access experienced by an average worker in that region. These summaries form the basis of the spatial and temporal comparisons and allow time series to be built for areas of interest, up to and including an entire state (Figure 4.1).