Maryland. State Highway Administration. Office of Policy & Research

PDF

The goal of this study was to identify recurring or common road characteristics of motorcycle crashes

in Maryland from 1998 to 2007. Motorcycle crash data was obtained from the National Highway

Traffic Safety Administration’s Crash Outcome Data Evaluation System (CODES), and road

inventory data was obtained from the Maryland State Highway Administration. Both sets of data were

integrated, and fault tree analysis and variable selection methods were utilized to find the highest

frequency crash cases. A categorical principal component analysis was used to specify the minimum

number of variables that explain most of the observed variance. In addition, ordinal logistic models

were developed to estimate the number of motorcycle crashes for road segments within each road

class. The results of the logistic ordinal regression analysis show that area type, median type, speed

limit, average annual daily traffic (AADT), international roughness index (IRI), and the number of

through lanes affect the number of motorcycle crashes on Maryland road segments. Interestingly,

government control and shoulder type were shown to have no significant impact on the number of

motorcycle crashes. The developed ordinal logistic model can also calculate the number of

motorcycle crashes for each road characteristic. Engineers and safety officials can use this study’s

results to develop solutions for identified safety deficiencies.

US Transportation Collection

SP909B4J