Date of Award

5-2015

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Geography and Regional Planning

First Advisor

Robert P. Sechrist, Ph.D.

Second Advisor

Donald W. Buckwalter, Ph.D.

Third Advisor

John E. Benhart Jr., Ph.D.

Abstract

The capacity to identify hazardous roadway segments quantitatively can guide remedial engineering efforts and make smarter use of limited government resources. This research evaluates spatial analytic tools, employing the local network Moran's I indicator, to statistically identify hazardous roadway segments. Hazardous segments are defined as those with crashes occurring in greater proportion to traffic volume and in relation to adjacent roadway segments proportions of crashes. In a traditional non-uniform analysis high incident segments correlate with traffic volume masking. Transformation of the roadway network into a uniform network reveals more high incident segments. The veracity of the findings were tested using the Chi-square test, Spearman's Rho and binary logistic regression. The analysis identified 27 hazardous segments, four were statistically analyzed. Hazardous segments displayed significant explanatory variables and statistical dependencies. Scene visitations revealed evidence of contributing factors. This methodology can be employed to identify crash clustering and understand causal factors.

Share

COinS