Steve LindsayNational Public Radio (which I love) today aired a story headlined "A huge study finds a link between cannabis use in teens and psychosis later." The story argues that by excluding teens with mental health diagnoses at the beginning of the study the research provides causal evidence. I'd agree it strengthens the extent to which the results suggest causation, but at a glance there seem to me to be several weaknesses. Not my research domain. I'm mainly interested in using this as an example in my Research Methods in Psychology course. Any experts care to comment? The study: https://jamanetwork.com/journals/jama-health-forum/fullarticle/2845356
Jan R. BoehnkeI am not sure short communication media are best for this, so I'd rather opt for a virtual or asynchronous journal club.
Teasing with:
- Biggest problem usually exclusions: I cannot gauge who made it into the sample at all (who has such a health plan?), but #STROBE does not look immediately bad.
- Limited control for self-medication and childhood adversity, so yes, there may be confounding pathways of relevant magnitude.
But on surface, much better than most in this area.
@jrboehnke Another consideration is the large N of disorders on which the two groups could have differed. And the low base rates for the two that were emphasized in the report.