Edit - I managed to mis-read the press release and read through the wrong paper in detail! Disregard the below… <s> Quite. If you read the paper -- https://www

Edit - I managed to mis-read the press release and read through the wrong paper in detail! Disregard the below… <s> Quite. If you read the paper -- https://www.nature.com/articles/s41598-020-59643-7.pdf https://www.nature.com/articles/s41598-020-59643-7.pdf -- the word 'sardine' doesn't appear once; 'fish' doesn't either. The trouble with observational cohort studies is that causality is very hard, if not impossible, to identify. Their main result is based on multivariate logistic regression. I like logistic regression as a technique, and it is very powerful, but if, for example, there is an unmeasured cofounding or causal variable that manifests itself as a linear combination of a number of measured but perhaps more acausal variables, it _will_ confidently give you the wrong result. </s>