The 'Grand Rapids dip': could it be real?

The "Grand Rapids dip" refers to a finding in the Grand Rapids study of drinking-and-driving, indicating that drivers with a low blood alcohol concentration had a lower risk of accident involvement than drivers with a zero-blood alcohol concentration. In the Grand Rapids study, it was found that this was a statistical artefact only and did not indicate a real decline in accident risk. This paper asks if a decline in accident risk at a low blood alcohol concentration could be real. It is concluded that it could but is likely to be uncommon. It may occur if: (1) There is a comparatively large a group of drivers who are experienced both in driving and in drinking alcohol; (2) This group of drivers drive regularly with a low blood alcohol concentration, and (3) The drivers can predict how a low blood alcohol concentration will influence them and adapt behaviour so as to avoid an increase in accident risk.