Causation, correlational analysis and prospective studies

#1
I am writing a systematic review of prospective studies, where two studies with two measurement points say that they cannot infer causation because of their correlational analysis. One report covariate analysis and path model, the other multiple regression analysis. But they don't explain why this is so. Why is it not possible to look at causation if they measure the exposure variable before the outcome? And what is it with the correlational analysis that makes it impossible to infer causation?
 

hlsmith

Less is more. Stay pure. Stay poor.
#3
A key limitation is that without randomization of the treatment, one can not ensure exchangeability of the two treatment groups. They may differ in background variables that may impact the treatment and outcome, which could misrepresent the estimated relationship between the treatment and outcome.

Of note, even in randomized control trials there could still be factors that impact the results, such as selection bias in recruitment and informative censoring/loss to follow-up.