Part of the reason why correlation is so hard to make out in the wild is that it’s often disguised in science coverage. Things like:
x is associated with y
x may be connected to y
x may suggest that….
and so on so forth…
It’s a general good rule of thumb that if you see any science reporting that makes you feel an emotion – happiness, sadness, fear, validation – that you take a moment to slow down and actually read how that science was done. You may not be capable of an expert-level dissection of every paper you come across, but what you can avoid is, as my thesis advisor puts it, drinking the author’s Kool-Aid.
Resources
https://tylervigen.com/spurious-correlations – Really fun site where you can see how perfectly correlations line up and uh, decide for yourself if those correlations could mean causation.
https://www.bmj.com/content/bmj/344/bmj.e1454.full.pdf – White rice meta analysis, the first paper discussed
https://www.cambridge.org/core/journals/british-journal-of-nutrition/article/substituting-brown-rice-for-white-rice-on-diabetes-risk-factors-in-india-a-randomized-controlled-trial/A0778FC028F6F25D0E6A73787EECECC4 – Randomized Crossover Trial, the second paper discussed
https://www.tandfonline.com/doi/full/10.1080/10408398.2021.1914541 – Another meta analysis that I didn’t get to, but is your homework if you want to dig deeper into this topic – mentioned briefly at the end of the episode
https://www.sciencedirect.com/science/article/abs/pii/S0002822310005249 – Not mentioned in the episode but I considered covering this one but decided not to due to low sample size
Resources I used to learn about different study designs
https://pubmed.ncbi.nlm.nih.gov/11451349/
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2998589/
https://himmelfarb.gwu.edu/tutorials/studydesign101/cohorts.cfm
https://pubmed.ncbi.nlm.nih.gov/28846237/
https://www.bmj.com/about-bmj/resources-readers/publications/epidemiology-uninitiated/8-case-control-and-cross-sectional