Anti helmet legislation revisited… and why replication is important

Another anti helmet legislation argument bites the dust

When Ian Walker’s paper appeared in Accid Anal Prev in 2007 purporting to show that cars drove closer to helmeted than unhelmeted cyclists, it was quickly used as another argument against helmet legislation. But for me as a long time cyclist, something did not ring true. Jake Olivier, a statistician from Australia, had the same vibes and reanalysed Walker’s data. The email from him below gives the publication details. Another anti-legislation argument crumbles.  “My re-analysis of Ian Walker’s now famous naturalistic cycling study on motor vehicle passing distance has been published by PLOS ONE.  My conclusion is that while vehicle size, cyclist distance to the kerb and city of occurrence (Bristol or Salisbury, UK) are important factors associated with close passing (<1m), helmet wearing is not a significant factor.  The real differences in passing distance when wearing or not wearing a helmet occur for distances greater than 1.5m. I first submitted this paper to AAP as it was the journal that published Walker’s original work. It was primarily rejected on the grounds that there is no justification for the one metre rule (3 foot rule). My analysis was an assessment of helmet wearing relative to road safety policy that exists or is promoted in many jurisdictions. Still, that is why this paper is in PLOS ONE and not AAP.  This work was actually two studies – one in road safety and the other statistical. They are meant to complement each other. The stats one is pretty dense, but it was needed to demonstrate Walker overpowered his study, thereby misinterpreting his statistical results.;jsessionid=B57108C74B43A32BEF8E9F2A6AB85C9F?articleURI=info%3Adoi%2F10.1371%2Fjournal.pone.0075424

Editors note: A fascinating paper in PLOS ONE highlights the importance of publishing replications in the same journal… and the importance of the original authors acknowledging they were mistaken if that is what is found.  Here is the Abstract. The underlining is mine.  “Social psychology and related disciplines are seeing a resurgence of interest in replication, as well as actual replication efforts. But prior work suggests that even a clear demonstration that a finding is invalid often fails to shake acceptance of the finding. This threatens the full impact of these replication efforts. Here we show that the actions of two key players – journal editors and the authors of original (invalidated) research findings – are critical to the broader public’s continued belief in an invalidated research conclusion. Across three experiments, we show that belief in an invalidated finding falls sharply when a critical failed replication is published in the same – versus different – journal as the original finding, and when the authors of the original finding acknowledge that the new findings invalidate their conclusions. We conclude by discussing policy implications of our key findings.”

Citation: Eriksson K, Simpson B (2013) Editorial Decisions May Perpetuate Belief in Invalid Research Findings. PLoS ONE 8(9): e73364. doi:10.1371/journal.pone.0073364

 PS.. This will appear in the next issue of News and Notes but it seemed important enough to duplicate on the blog. 

(Visited 65 times, 1 visits today)