Exploring the temporal stability of global road safety statistics

Accid Anal Prev. 2019 Sep:130:38-53. doi: 10.1016/j.aap.2017.12.015. Epub 2018 Feb 21.

Abstract

Given the importance of rigorous quantitative reasoning in supporting national, regional or global road safety policies, data quality, reliability, and stability are of the upmost importance. This study focuses on macroscopic properties of road safety statistics and the temporal stability of these statistics at a global level. A thorough investigation of two years of measurements was conducted to identify any unexpected gaps that could highlight the existence of inconsistent measurements. The database used in this research includes 121 member countries of the United Nation (UN-121) with a population of at least one million (smaller country data shows higher instability) and includes road safety and socioeconomic variables collected from a number of international databases (e.g. WHO and World Bank) for the years 2010 and 2013. For the fulfillment of the earlier stated goal, a number of data visualization and exploratory analyses (Hierarchical Clustering and Principal Component Analysis) were conducted. Furthermore, in order to provide a richer analysis of the data, we developed and compared the specification of a number of Structural Equation Models for the years 2010 and 2013. Different scenarios have been developed, with different endogenous variables (indicators of mortality rate and fatality risk) and structural forms. The findings of the current research indicate inconsistency phenomena in global statistics of different instances/years. Finally, the results of this research provide evidence on the importance of careful and systematic data collection for developing advanced statistical and econometric techniques and furthermore for developing road safety policies.

Keywords: Data inconsistencies; Data visualization; Hierarchical clustering; Principal component analysis; Road traffic fatalities; Structural equation modeling.

MeSH terms

  • Accidents, Traffic / statistics & numerical data*
  • Data Accuracy*
  • Databases, Factual
  • Humans
  • Policy
  • Reproducibility of Results
  • Safety / statistics & numerical data*
  • Spatial Analysis