Pavement Marking Retroreflectivity and Crash Frequency: Segmentation, Line Type, and Imputation Effects

This paper investigates the effects of segment length, line types, and imputed retroreflectivity values on pavement-marking retroreflectivity and crash frequency analyses. Road data, crash data, pavement condition data, and pavement-marking retroreflectivity data from Iowa Department of Transportation (DOT) databases were acquired and spatially integrated. Data sets for 1-, 3-, and 5-mi segments were prepared to investigate the effect of segmentation. Additional data sets with imputed and measured retroreflectivity data were prepared for comparison. A series of negative binomial regression analyses were run to estimate the expected number of crashes on varying segment lengths and data subsets based on the two retroreflectivity collection methods. The findings show that using smaller segments and data sets with measured retroreflectivity rather than imputed retroreflectivity leads to a more significant relationship between the retroreflectivity of longitudinal pavement markings and crash frequency. The findings also suggest that keeping longitudinal pavement markings in good condition has significant positive effects on safety. The results further suggested that the expected number of annual crashes significantly decreased with the increasing retroreflectivity of white-edge lines (WELs) and yellow-edge lines (YELs) for four-lane road segments. In addition, a significant relationship between pavement condition, measured with the International Roughness Index (IRI) and the expected number of crashes was found for all data sets