5 ms·
Key conclusion: "observational studies of interventions with small to moderate effect sizes are prone to bias caused by selection and omitted variables." Surpr
by sprague 4y ago
Key conclusion: "observational studies of interventions with small to moderate effect sizes are prone to bias caused by selection and omitted variables."
Surprisingly few interventions are so good that you don't need fancy statistics to justify them. Really good treatments (e.g. antibiotics) are so effective as to be obvious to all.
- colpabar 4y ago> Surprisingly few interventions are so good that you don't need fancy statistics to justify them To me the real story is that the CDC, which is supposed to be the public health authority in the US, decided not to use said "fancy statistics" and instead chose to only include data that supported their policies. It's the CDC, they should have access to tons of data. Why such small sample sizes?
- belltaco 4y ago>Surprisingly few interventions are so good that you don't need fancy statistics to justify them. Really good treatments (e.g. antibiotics) are so effective as to be obvious to all You mean like the HCQ cures covid research and studeis in March 2020 that "showed" a 100% cure rate that was pushed hard by Elon Musk and an entire political wing? And so many people still believe it, even on HN, it's crazy.
- colpabar 4y agoAre any of those people in this thread?