Incorporating Auto Accessibility into Statewide Project Prioritization: Feasibility, Behavior, and Equity

While accessibility, the number of time-decayed jobs available to each zone within a region, has frequently been proposed as an element in transportation project prioritization, widespread adoption of accessibility has been hindered by two obstacles: computational feasibility in a semi-open source manner and longitudinal transparency. Surmounting the former through computational steps such as automation of turn restrictions, error checking for incorrectly formed GIS-based service areas, and accounting for random perturbations in the formation of such areas has been a necessary, but not sufficient, condition to render accessibility usable at a statewide level. This dissertation shows that design choices (e.g., number of centroid connectors or the catchment radius), which historically have not been examined in detail, do not have a single “correct” value. Rather, such design choices implicitly determine which of three paradigms are followed when using accessibility to prioritize projects.
One theory is that accessibility computations should directly respond to stakeholder concerns. In this context, the study finds that alteration of one particular design choice, the catchment radius, affected project rankings in manner opposite of those expected by stakeholders, where project accessibility benefits decreased, rather than increased, as the radius grew, owing to the fact that the marginal increase in accessibility was less than the marginal decrease in population. Proponents of the project in question would have been satisfied, therefore, with a fairly small catchment radius of 5 miles. A second theory is that accessibility computations should maximize the association between forecast and observed traveler behavior: this dissertation finds that accessibility alone has a statistically significant impact on destination choice and that the catchment radius can be selected to maximize this association (where in this particular case, a value of 35 miles yields the strongest association). In this case, accessibility alone explains between 4% and 10% of the variation in destination choice depending on the radius chosen. While such values will seem low to proponents, note that the socioeconomic factors alone (income, housing prices, and location) only add about a percentage point to this variance explained. A third theory concerns a conflict within the planning process: to what extent should accessibility for low-income populations, as opposed to accessibility for total populations, be part of project prioritization. This dissertation shows how to select a catchment radius that reduces this conflict, such that accessibility for both groups are aligned when one sets the catchment radius to about 25 miles.
The fact that the manner in which some project evaluation criterion (accessibility) is computed can affect the ranking of candidate projects will not surprise veteran observers of agency transportation project prioritization processes. However, this dissertation proves that the three theories for how such computations should be done—address stakeholder concerns, maximize association of forecasts with observed traveler behavior, or reduce conflict in the planning process by aligning equity concerns with the population at large—are all implicit decisions that result from the manner in which computations are performed. By making these choices explicit—that is, by showing that a particular catchment radius, once selected, will tend to favor one of these three theories relative to the other two—this dissertation seeks to advance the state of transparency when using accessibility as an element in project prioritization.