If you read the actual paper, and it's very easy to skim through, the authors offer a lot of reasons to doubt the results, from strong bias in the data to repeated samples across studies. That being said, it is nice to start getting some better data about marijuana use.