Simpson's Paradox Skepticism

Commenters debate data visualizations and trends, frequently attributing counterintuitive results to Simpson's paradox, regression to the mean, single data points, small samples, outliers, and other statistical artifacts rather than genuine effects.

📉 Falling 0.4x Science
3,443
Comments
20
Years Active
5
Top Authors
#9405
Topic ID

Activity Over Time

2007
4
2008
15
2009
64
2010
42
2011
91
2012
122
2013
146
2014
164
2015
195
2016
210
2017
233
2018
215
2019
205
2020
281
2021
284
2022
333
2023
313
2024
274
2025
219
2026
33

Keywords

MSE WHY github.com modeanalytics.com VPN wikipedia.org regression data outliers bob data point alice variance paradox single flights

Sample Comments

Tomte Apr 6, 2025 View on HN

Could be Simpson's Paradox: https://en.wikipedia.org/wiki/Simpson%27s_paradox

Natsu Feb 10, 2016 View on HN

Perhaps https://en.wikipedia.org/wiki/Simpson's_paradox ?

ellisv Feb 12, 2025 View on HN

Really makes you wish the chart showed some measure of variance

TechBro8615 Apr 22, 2021 View on HN

Could be sampling discrepancies? I’m no statistician but I would assume there’s a good chance it’s due to the modeling rather than an actual reality.

Terr_ Jun 6, 2018 View on HN

Sounds like a case of https://en.wikipedia.org/wiki/Simpson%27s_paradox

gohrt Nov 4, 2016 View on HN

Simpson's paradox!https://en.wikipedia.org/wiki/Simpson%27s_paradox

drsopp Jun 13, 2018 View on HN

Perhaps regression toward the mean counteracts this?

hvasilev Nov 12, 2021 View on HN

you are misunderstanding what this data shows.

jasonvorhe Jul 18, 2018 View on HN

Because of one data point? Seriously?

lupire Aug 5, 2023 View on HN

Statistics says it's unlikely for small differencees to have large effects.