Summary: They looked at people aged 14-24 ("time T1", which they actually call "baseline"), and followed them up 3.5 years later ("time T2") and 5 years after

Summary: They looked at people aged 14-24 ("time T1", which they actually call "baseline"), and followed them up 3.5 years later ("time T2") and 5 years after that ("time T3"). They looked at whether they'd used cannabis five times or more by T1, or between T1 and T2. And they looked at whether they'd had any "psychotic symptoms" -- quite broadly defined, so that ~10% had such symptoms between T2 and T3 -- between between T2 and T3. They excluded from their sample anyone who'd had psychotic symptoms by time T2; that was 23% of their sample! (They wanted to avoid finding people who had, e.g., already had psychological problems and taken cannabis in the hope that that would help.) They also excluded anyone who'd taken cannabis before T1. (The results without those exclusions were similar to the results with them.) They found a substantial increase -- about a factor of 2 -- in incidence of "psychotic symptoms" between T2 and T3 among people who used cannabis between T1 and T2. Remember that "psychotic symptoms" doesn't mean actually being psychotic. They also looked for evidence of that "self-medication" scenario, by looking to see whether people with psychotic symptoms by T2 were more likely to use cannabis between T2 and T3. Apparently not. They found some evidence that using cannabis over a longer period is correlated with (and probably causes, given their findings about self-medication, but that's harder to be sure of) larger risk increases for psychotic symptoms. There are a bunch of caveats; for instance, they're relying on self-reported data for both cannabis use and psychotic symptoms. (They give some reasons for thinking that this doesn't hurt their analysis too much.) I think their investigation would have been improved by introducing an extra delay, and looking at whether cannabis use between T1 and T2 is associated with more psychotic symptoms between T3 and T4. (Compare the two propositions "Taking cannabis now will make your brain a bit weird for a year or two" and "Taking cannabis now will make your brain a bit weird for the rest of your life".)