Regression to mean

The phenomenon of regression to the mean needs to be considered as part of this analysis. The theory of regression to mean implies that because locations are chosen for investigation when the accident total is high, accident levels will come down regardless of treatment, due to random fluctuation. 

As far as this book is concerned, the potential for regression to mean is acknowledged, along with the acceptance of the possibility that some of the claimed accident reduction values will overstate the real change that has occurred. 

Moreover, a possible regression to mean effect (Barnett et al., 2005) is expected because most of the enterprises participating are chosen because of their high injury rates in the year before intervention. The association between relative injury rates before the intervention and the reduction in injury rates could indicate that part of the effect of the program a regression to mean effect thus further research is needed to estimate the real effect of the program in terms of injury rates reduction. 

Linkedin (http //www.linkedin.com) provides some useful discussion topics of relevance to Road Safety Auditors. For example, a discussion group in early 2014 had nearly 30 pages of debate concerning the effectiveness of speed cameras and whether the regression to mean effect was adequately accounted for in the published reports. 

---