A statistical assessment of temporal instability in the factors determining motorcyclist injury severities

Abstract

Abstract This study explores the temporal instability of factors affecting motorcyclist-injury severities in single-vehicle motorcycle crashes in Florida. Two data sources are used; one covers the 2012 to 2016 crash histories of Florida motorcyclists who were newly licensed in 2012, and the second covers motorcycle crashes that occur on horizontal curves in Florida from 2005 to 2015. In the first dataset (2012 new riders), temporal changes may result from riders gaining experience as well as general temporal shifts. In the second dataset, rider experience is unknown (thus becoming a source of potential unobserved heterogeneity) but the temporal changes will be largely from general temporal shifts. With three possible motorcyclist injury severity outcomes (no visible injury, minor injury, and severe injury), random parameters multinomial logit models, that allow for heterogeneity in means and variances, were estimated for all possible annual time periods in each dataset. Likelihood ratio tests were conducted to examine the overall stability of model estimates across time periods, and marginal effects of each explanatory variable were also considered to investigate the temporal instability of the effect of individual parameter estimates on motorcyclist injury-severity probabilities. A wide range of variables was considered including motorcyclists’ attributes (such as ethnicity and age), roadway and environmental conditions (such as light and road surface conditions), motorcycle characteristics (such as motorcycle make and type of motorcycle), rider actions (such as speeding and improper driving actions), and roadway conditions (such as obstacles on the road and speed limits). The results show significant temporal instability in motorcyclist-injury severity models, which likely result from changes in motorcycle technology and performance, changes in macroeconomic conditions, changes induced by how riders respond to the changing behavior of other road users (whose behavior may be changing as a result of technology changes in their vehicles, evolving use of personal technologies in their vehicle, such as cell phones, etc.), and the changes in riders’ behavior and skills over time.