The insurance lobby says states with legal marijuana have had a 14 percent increase in accidents. But The Oregonian says traffic deaths are down 50 percent this year. Did last year's drunken 100-mph maniacs morph into spacey, 15-mph fender-bending stoners? —Darrell T.

The study in question was released June 22 by the Insurance Institute for Highway Safety, an industry-funded group. I'm not saying such groups have an agenda, but don't wait for the American Beef Council to remind you to get your cholesterol checked.

From the press release: "After retail marijuana sales began in Colorado, the increase in collision claim frequency was 14 percent higher than in nearby Nebraska, Utah and Wyoming."

Of course, that's a tricky way of putting it: If your investments are up 7 percent from last year and mine are up 8 percent, my increase is 14 percent bigger than yours—but I'm not much richer. (Later in the document, IIHS acknowledges that the overall increase is more like 2.7 percent.)

What if we looked at the numbers ourselves? Not at insurance claims, but at highway fatalities per 100 million miles traveled—after all, that's what's really important.

In 2013, Colorado's first year with recreational weed, the state fatality factor increased by 2 percent—but in 2014 it dropped 3 percent. Washington's first year brought a 1 percent drop in deaths, but they jumped 3 percent in 2014.

Oregon's coming-out year, however, brought a 20 percent spike in the fatality rate—pretty deadly! Even sharing a border with us was dangerous—Nevada's rate increased by 10 percent and Idaho's by 13 percent in that same year. It's almost like the numbers fluctuate year to year based on random chance.

Weed may be making the roads more dangerous, but it's going to take many more years of data to glean a signal from the noise. I've said it before, and I'm going to keep saying it: You can make numbers confess to anything if you torture them long enough.

QUESTIONS? Send them to dr.know@wweek.com.

Related: