Timeline for What to conclude when you fail to find an association in an epidemiological study?
Current License: CC BY-SA 3.0
13 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jul 1, 2012 at 19:50 | vote | accept | Mr Alpha | ||
| Jun 22, 2012 at 8:58 | answer | added | Jeromy Anglim | timeline score: 1 | |
| Jun 22, 2012 at 8:33 | history | edited | Jeromy Anglim | CC BY-SA 3.0 |
added 9 characters in body; edited title; edited title
|
| Jun 21, 2012 at 18:59 | comment | added | Fomite | @Macro The "Yes, of course" is something a remarkable amount of the people who trot out that particular statement seem to miss. | |
| Jun 21, 2012 at 18:57 | comment | added | Macro | @EpiGrad, re: and technically, anyone using 'correlation doesn't imply causation' ... is also wrong by the same logic. - I probably don't disagree but I'm not sure what you're saying. Using the usual statistical definition of correlation, it clearly doesn't imply causation. Re: "But causation does imply association." - yes, of course. Otherwise, I'm not quite sure what "causation" would even mean :) | |
| Jun 21, 2012 at 18:26 | comment | added | Fomite | @Macro CV keeps going back and forth on this - and technically, anyone using 'correlation doesn't imply causation' in regards to an epidemiological effect estimate is also wrong by the same logic. But oddly, it doesn't seem to get pointed out as much then. But causation does imply association. | |
| Jun 21, 2012 at 18:25 | answer | added | Fomite | timeline score: 3 | |
| Jun 21, 2012 at 13:34 | history | tweeted | twitter.com/#!/StackStats/status/215799828289036288 | ||
| Jun 21, 2012 at 12:50 | answer | added | Néstor | timeline score: 0 | |
| Jun 21, 2012 at 12:23 | history | edited | onestop |
edited tags
|
|
| Jun 21, 2012 at 12:12 | comment | added | Macro | @ChrisTaylor, assuming you're using the word "correlation" in the usual statistical sense, causation does not need to imply correlation if the relationship is non-linear. If $X \sim N(0,1)$, then $X$ is causally linked to $|X|$ but not correlated. | |
| Jun 21, 2012 at 11:46 | comment | added | Chris Taylor | Correlation doesn't imply causation, but causation does imply correlation. Taking the contrapositive, lack of correlation implies lack of causation. So this removes one of the criticisms commonly levelled at epidemiological studies - other criticisms may still apply, however. | |
| Jun 21, 2012 at 11:43 | history | asked | Mr Alpha | CC BY-SA 3.0 |