I would proffer that the recruitment bias for this survey renders its results dangerous. I would proffer further that programmers at large do no use cannabis a

I would proffer that the recruitment bias for this survey renders its results dangerous. I would proffer further that programmers at large do no use cannabis and the "true" conclusion for such a study might well be the exact opposite of the authors. The recuitment basis for the survey is not random and is only amongst a very small subset of candidates the authors considered "ethically permitted" (specifically from GitHub, the University of Michigan and social media). So probably few enterprise programmers and few non-US participants, etc.. In other words not the majority of programmers. "We now present indirect evidence that our participants, while not a random sample, are similar in many ways to previous random samples or studies. A true random sample would not have been ethically permitted, but we gain confidence in our results’ generalizability by contextualizing participants’ gender, age, and employment. ... our study population by recruitment pool (e.g., how they were contacted to participate in this study)... from GitHub ... from the university ... and social media" - page 4, https://arxiv.org/pdf/2112.09365.pdf https://arxiv.org/pdf/2112.09365.pdf A more meaningful lead-in survey would have been to ask programmers from all sources if they are subject to any form of drug-testing. Perhaps they should have selected a less clickbaity title for their study: A Survey of [Ethically Permitted USA] Programmers' Cannabis Usage, Perception, and Motivation