There is no mention that these drivers were ONLY impaired by weed. But I can’t believe a paper would not look at the confounders. I know quite a few who are not regular smokers but will imbibe after a few beers if it’s being passed around. Also weed is popular with consumers of stimulants. Without knowing the possible confounders, this statistic tells you very little.
If you do an adequate job of looking for confounders you'll likely torpedo the study because it will turn out to be something already known.