The story is that drivers with a BAC of 0.01 or 0.02 cause fewer crashes and fatalities than drivers with a BAC of 0.00, or a BAC above 0.03. The legal limit in many states is BAC= 0.08. This statistic seems to convey the message that one glass of wine will turn you into a safer driver, while two glasses will make you impaired. Source: Traffic Accidents, by R.F. Borkenstein at al., Indiana University Department of Police Administration.
Actually, the reality is very different. The above conclusions - despite being based on hard facts - is a typical example of how statistics can be wrongly interpreted. The explanation about lower car accidents rate at the 0.01 and 0.02 BAC (compared with 0.00 BAC) is caused by heavy drinkers who drive much more safely, when their BAC is 0.01 or 0.02 rather than 0.15 or 0.20. Indeed, older alcoholics in the rare instances when they drive with a BAC as low as 0.02 (as opposed to their typical 0.20 BAC) are much safer than 16 year old drivers who don't drink at all (BAC = 0.00).
In other words, if you create a 2-group segmentation (heavy drinkers vs. moderate and non-drinkers), you will find that within each group, the higher the BAC, the higher the risk for car accidents.