The impact of ambient light on road traffic crashes: A study of analysis robustness

Six analyses of road traffic crashes (RTCs) were carried out to investigate how these change in daylight and darkness. The analyses used purposeful changes in the researcher who conducted the analysis, the analysis software they used and the source of solar altitude data. These decisions are amongst those that may be considered arbitrary in such an analysis, and the analyses were therefore conducted to assess the robustness of the analysis across these decisions. It was also done to compare the outcomes with two previous studies using the same data set (STATS19), the same data period (2005 to 2015) and the same approach to filtering data (the clock change approach). There tended to be good agreement between the analyses except when the analysis required the use of solar altitude to identify the daylight or dark status of an RTC. It was found that one solar altitude calculator (SunCalc) produced different estimates of solar altitude than did the other three used in this work (Astral, PysunNOAA and NOAA Excel). While the difference is small, it was sufficient to bring changes in the outcomes of the analysis.