File talk:% of US Car Crash Fatalities Where Driver Blood Alcohol Level Was .01 and Above 1999 - 2012.gif

What is this trying to represent? It seems to be reporting P(BAC>=0.1 | Car Crash), which seems like a pretty useless statistic by itself, since the conditionality is backwards from what it would seem intended to quantify. If the data shown here were interpreted to represent how alcohol consumption influences the rate of car accidents, it would lead to some pretty strange conclusions, e.g. that in Florida in 2004, not drinking any alcohol at all was twice as dangerous as having a BAC>=0.1.