Confounding Factors in Studies

Commenters criticize a scientific study's validity, highlighting uncontrolled confounding variables, lack of randomization, and the distinction between correlation and causation in observational research.

📉 Falling 0.3x Health
3,123
Comments
20
Years Active
5
Top Authors
#3322
Topic ID

Activity Over Time

2007
3
2008
12
2009
34
2010
49
2011
40
2012
77
2013
87
2014
106
2015
127
2016
155
2017
172
2018
238
2019
210
2020
291
2021
342
2022
405
2023
292
2024
266
2025
198
2026
19

Keywords

wiley.com e.g ICU MR twitter.com EDIT DHA I.e academic.oup journals.lww study confounding factors correlation variables studies causal randomized control causation

Sample Comments

some_random Dec 9, 2019 View on HN

Have you read the study? How do you know that they didn't control for this? Seems like a pretty spurious assumption.

ikeboy Feb 7, 2019 View on HN

Be cautious of getting causality claims from studies like this that don't try to control for anything

CydeWeys Jun 9, 2024 View on HN

That's not a randomized controlled trial though. You might find a correlation, but you couldn't actually establish causation, or eliminate hidden variables.

_Microft Feb 20, 2022 View on HN

I don't have time to read the study but if you want to have a look yourself, the terms to look for are "confounder", "confounding factor", ...https://en.wikipedia.org/wiki/Confounding

nevinera Jun 20, 2014 View on HN

No controlled study was done, so any unexpected confounding factors would invalidate the results. This type of study is nearly pointless when trying to demonstrate a causal link.

idiotsecant Jan 8, 2026 View on HN

You don't think studies control for this?

blackeyeblitzar Jun 23, 2024 View on HN

This feels like a study with many confounding factors.

kadoban Sep 24, 2023 View on HN

"is associated" means correlation, usually after whatever attempt (or no attempt, you have to read quite closely to tell) to account for confounding variables.Even if they do make an attempt to control for such things, it's often useless. There's only so much you can do with limited data and if you're not of exceedingly high character, there's a lot of ways to cheat intentionally or accidentally.So, this kind of result means very little. It _may_ be a thing th

Arech Jan 28, 2021 View on HN

These's not a word in abstract of the study [0] that they have controlled for anything. (and I don't say that assuming anything about a study is a bad scientific practice, but I do assume it)[0] https://onlinelibrary.wiley.com/doi/abs/10.1111/dar.13239

cjbprime Feb 9, 2018 View on HN

The effect size isn't large, and the potential cofounding correlations are infinite. This doesn't seem like a good study.