“There are lies, damned lies, and statistics.” Mark Twain
With science, what could possibly go wrong?
Correlations: The increased diagnosis in the USA and the price of eggs in China have both been rising at the same rate. Coincidence??
This is of course silly, but confusion of correlation with cause is prevalent not only among the conspiracy theorists but also among scientists.
If we get enough correlations that actually make sense, like that between smoking and lung cancer, that can constitute circumstantial evidence, and can be pretty convincing, but are not proof.
Poorly done research: poorly designed, or leaping to subjective conclusions beyond the data, or actual fraud. And then statistics if you have enough variables and keep looking for correlations, you will eventually find some. But what do they mean?
A recent study showed that people who gargle regularly get fewer colds. OK, so we should all gargle. Or, could it be, that people who gargle often are health conscious, and wash their hands more often, eat better diets, take healthy vitamins and supplements, avoid crowds during contagions etc.?
A recent study shocked us by showing that obese people live longer that lean people! Shocking! But then close reading showed that the lean people in their sample were randomly selected, including for example people with cancer and heavy smokers and anorexics, none of whom were much represented in the obese group – and who tended of course not to live so long.
subjective conclusions, not supported by the data
Scientists studied symptoms of patients with schizophrenia over time, and reported their objective results, and then titled their paper, “Schizophrenia has a poor prognosis.” Well, that’s very questionable – compared to what, for example. Or versus studies that followed patients longer and found better outcomes. They leapt from their objective data to their subjective conclusion and title.
So science is obviously imperfect, and needs to be scrutinized with skepticism.
But next time, beliefs.