The reason why I’m so skeptical towards statistics:

Statistics can be gamed to say or suggest whatever you want it to say.

For example, some problems of statistics:

1. Hidden agendas and motives for whoever funds the studies.
2. Cherry picking of data, and sweeping “outliers” under the rug.
3. People assume numbers and data is ultimate truth (data deification in Silicon Valley)
4. Statistics cannot “prove” anything, but can falsify (Karl Popper) false ideas.
5. Statistics comes from the notion of “the State”. Thus all statistics has a political or governmental or state-based angle.
6. Collection methods are often dubious. Either those administering the surveys are either incompetent, or people who fill out these forms lie or under-represent (or over-represent the truth).
7. Many studies are observational studies, not experimental studies. Observation doesn’t tell the truth. Experimental (double blind studies) are far more expensive yet accurate and does uncover some hidden truth. For example it is unethical to do a study which forces women to drink 2 glasses a wine a day while pregnant vs a group of pregnant women who don’t drink alcohol to answer the question: “Is drinking alcohol for pregnant women bad for the future child?” An observational study can suggest (not prove) that women who typically drink more alcohol lead to more child births with complications. But also what the observational study will show is that women who typically drink a lot of alcohol while pregnant ALSO smoke, ALSO do heroin or other weird drugs. Or women who typically drink a lot also are part of a lower socio-economic class, thus has less access to good health care. Therefore we cannot trust observational studies. Observational studies ain’t hard science.
8. Small study sizes. Most studies are sooooo maps all (involve fewer than 20 individuals) which cannot be generalized to the vast population.
9. Issue of replicability: Many old school Harvard or Stanford psychology studies have not been able to be replicated today. If a study or experiment cannot be successfully replicated from 30 years prior to today, the findings from the study probably cannot be trusted.