Refuted Causal Claims from Observational Studies

It is common to find a correlation between A and B and incorrectly conclude that A causes B or vice-versa.  While correlations are a great source of hypotheses, causal claims based on observations alone should have a lower level of trust than properly run randomized controlled experiments; this is true even when attempts are made to control for several factors.

We review famous examples where causality was claimed as likely in observational studies, but later refuted in studies higher in the hierarchy of evidence, such as randomized controlled experiments.

This summary is made available online in two forms:

  • Refuted causal claims from observational studies in PDF
  • Refuted causal claims from observational studies in Word Read/Write - open for feedback/comments
    The password is the animal mascot we use with the correct capitalization (e.g., lowercase i)

Quick link to this page: