Imagine you wanted to know which phone was faster: the iPhone 11 or the Samsung Galaxy S20.
But instead of setting up an experiment with tests, you decided to find the answer by surveying a couple thousand people, asking them to rate their phone's performance on a scale from 1-5; and base your conclusion on those results.
Sounds pretty stupid, doesn't it?
Yet, this is what almost all studies on health and nutrition are like!
They're epidemiological, i.e. not based on experimentation, but on surveys – people filling out questionnaires about what they ate over the last X number of months and rating how they feel.
Remember that the next time somebody claims that a certain food is "scientifically proven" to be good/bad for you.
Experimentation > questionnaires