Talk:Controlling for a variable
 | This article is rated Start-class on Wikipedia's content assessment scale. It is of interest to the following WikiProjects: | ||||||||||||||||||||
| |||||||||||||||||||||
 |
This article links to one or more target anchors that no longer exist.
Please help fix the broken anchors. You can remove this template after fixing the problems. | Reporting errors |
Difference between statistical control and controlled experiment
editThe title of this article seems like it is referring more to statistical control (which is used more in observational studies) and less to controlled experimental designs. However, most of the text seems to refer to the latter, even though we already have another article on control in experimental designs. Unless anyone objects in the next few days, I'm going to do a major overhaul on this article to make it more specific to observational studies. Danni Ruthvan (talk) 20:17, 13 June 2014 (UTC)
controlling for vs counterfactuals
editI don't have time to fix it, but this article repeatedly makes the error that controlling for mitigates confounders. Unfortunately, it opens back doors to unknown confounders. It should be written more conservatively. It should also illustrate the causal diagram showing a confounder along with the interpretation of controlling for as observation.
data binning link
editOne controls for a variable to investigate causation, but data binning says its purpose is to remove noise through rounding, so linking to it in the tone of a definition is misleading. — Preceding unsigned comment added by 91.64.181.162 (talk) 09:35, 5 February 2023 (UTC)