The impact of weekday, weekend, and holiday crashes on motorcyclist injury severities: Accounting for temporal influence with unobserved effect and insights from out-of-sample prediction

• Determinants of motorcyclist injury severities are studied. • Random parameters logit models with heterogeneity in means and variances are estimated. • Significant differences between weekday, weekend and holiday motorcyclist injury severities are observed. • Factors affecting motorcyclist injury severities have changed over time. • Temporal shift and nontransferability of theses crash types are confirmed by out-of-sample prediction. This paper examines the differences between weekday, weekend, and holiday crashes on the severity of motorcyclist injury using four-year motorcycle crash data in Thailand from 2016-2019. While also considering the temporal stability assessment of significant factors, this study adopted a random parameters logit model with possible heterogeneity in means and variances to account for unobserved heterogeneity. Three levels of motorcyclist injury severity were considered including minor injury, severe injury, and fatal injury. Two series of likelihood ratio tests clearly indicated nontransferability between weekday, weekend, and holiday crashes and substantial temporal instability over the four-year study period. Findings also revealed many statistically significant factors that affect motorcyclist injury severity probabilities in various time-of-year and yearly models. In addition, the prediction comparison results (using out-of-sample prediction simulation) clearly illustrated substantial differences between weekday, weekend, and holiday motorcyclist injury severity probabilities, and substantial changes in each injury predicted probabilities over time. This paper highlights the importance of accounting for day-of-week and holiday transferability and temporal instability with unobserved effects in determinants that affect motorcyclist injury severity. Through nontransferability and temporal instability, the findings provide valuable knowledge for practitioners, researchers, institutions, and decision-makers to enhance highway safety, specifically motorcyclist safety, and facilitate the development of more effective motorcycle crash injury mitigation policies.