Bad Science: Comments on the paper ‘Quantifying the impact of road lighting on road safety — A New Zealand Study’ by Jackett & Frith (2013)

Abstract

The paper of Jackett & Frith (2013), which purports to show considerable gains for road safety with increasing road luminance, is seriously flawed. It asserts that increasing the luminance on roads causes improvements in road safety. Its cross-sectional design fails to rule out major potential confounders. Using a longitudinal design would be a far superior approach. The paper exhibits poor statistical practice. The selection process for the relatively small sample of urban roads is unclear and the post hoc processing of the data is questionable. The analysis is seriously deficient, as variables which indicate detrimental effects of increased road lighting are removed from the modelling without proper justification and other variables are not included in the first analysis yet appear in the subsequent cosmetic analyses. The latter give an illusion of false certainty. The data collected, which would allow checking, is not published. The practice of the journal in which the paper appeared is seriously deficient in not allowing the publication of critical responses. Although being used to promote increased road lighting, the paper’s claim disagrees with results from better quality research.

The full text is available here.