A favorite example

Tim Wilson writes:

For a book I’m writing, I’m looking for good examples in which regression suggested that A caused B, whereas experimental studies showed that there was no causal relationship. Even better (at least for the sake of my example) would be if social policy changes were made based on the regression. Do you have a favorite example or two?

My reply:

Here’s everybody’s favorite example.

8 thoughts on “A favorite example

  1. The financial crisis should give you a wealth of examples in which models that were based on regression produced bad predictions.

  2. Another example from medicine would be Beta carotene and prevention of cancer (specifically lung ca). Observational data seemed to indicate a link and there was a plausible biological mechanism. However, several RCTs provided data, that beta carotene might even increase the risk of lung ca.

    HRT though is definitely one of the most notorious examples.

  3. The bicycle helmet literature might work here, although the story is a bit more complicated.

    Studies in the medical literature suggest wildly improbable effect sizes, e.g. a reduction in head injuries of 85%. These may relate to improper controls in the control groups.

    On the other hand, studies of cases in which helmets were made mandatory in a political jurisdiction suggest either no effect on mortality or an effect mainly due to discouragement of bicycle riding — nothing consistent with the gigantic effects in the "controlled" studies.

    The literature here is voluminous, acrimonious, and off-topic here, so I'll stop after noting that I personally wear a helmet.

  4. Not exactly the same, but the prominent examples of Simpson's paradox (Berkeley admissions, etc) come to mind.

  5. About the cancer R. Francis told that aged cells turn in tumors … "growth arrest in aging cells might directly lead to cell cancer formation explaining this apparent contradiction"

Comments are closed.