Is this impressive plot showing "vibration of effects" meaningful?

Sorry to prolong the discussion but I think it would be far better for nutrional epidemiology to

  • be highly aggressive in adjusting for confounding as discussed by @Sander_Greenland here
  • use best statistical practice, also typified by the Greenland article

Best statistical practice does not require entertaining more than one model.

1 Like

Interesting, thanks very much. I have not read that paper.