Research Methods · Statistics

About the Statistics category

Research Methods › Statistics effect size number needed to treat heterogeneity

Pinned Wiki

s.leclercTL4 Moderator23 Sep 2025#1

Community wiki post. Any member at trust level 3 or above can edit this post; every edit is recorded. Last edited by hana.sato on 26 Feb 2026.

27 Sep 2025 — s.leclerc: Restructured into sections so the outline is navigable.
8 Nov 2025 — mira.patel: Added the limitations paragraph that review asked for.
26 Feb 2026 — hana.sato: Corrected an arithmetic slip in the second example.

Editors: s.leclerc, mira.patel, hana.sato

Staff post. Actions described here are recorded in the public moderation log and may be challenged in Meta.

Effect sizes, intervals, multiplicity, and the difference between absent and undetected.

This post is a community wiki: any member at trust level 3 or above can edit it, and every edit is recorded with its author and a note. If the scope described here has drifted from what actually gets posted, edit it rather than starting a topic about it.

What belongs here

Questions specific to statistics, with the units and the context stated.
Reports and datasets, with the collection method stated.
Disagreement about the substance, conducted in public.

What belongs elsewhere

Anything urgent belongs in Safety, immediately, and not here.
Anything about how this site works belongs in Meta.
Requests for a maintained page belong in Wiki requests.

Before posting, it is worth checking the documentation commons and this category's tag pages. A great deal of what gets asked here has a maintained answer with a review date on it, which is more reliable than a two-year-old reply.

0 likes 10mo

j.mwangiTL4 Moderator10 Oct 2025#2

Staff post. Actions described here are recorded in the public moderation log and may be challenged in Meta.

Moderation notes for this category, so that nothing about how it is run is a surprise.

Topics get moved into and out of here regularly. A move is not a judgement — the category tree is not obvious from outside and putting a question in the wrong place is expected. When I move something I say so in the topic and the action appears in the public log with the guideline cited.

The two guidelines that come up most often in this category are R2, on supplying a source when asked, and R7, on category placement. Neither is used punitively.

1 like 10mo

b.vestergaardTL222 Oct 2025#3

As a member rather than staff: the most useful thing I did when I started reading this category was to work through the tag pages rather than the topic list. The topic list shows you what is recent. The tags show you what is settled.

10 likes 9mo

CSagredoTL3Regular2 Nov 2025#4

Suggestion for the wiki post above: it would be worth linking the two or three most-referenced topics in this category directly, since new members reliably ask for exactly those. Happy to make the edit myself if nobody objects, with a revision note.

23 likes 9mo

a.vestergaardTL212 Nov 2025#5

s.leclerc, post #1: Effect sizes, intervals, multiplicity, and the difference between absent and undetected. This post is a community wiki: any member at trust level 3 or above can edit it, and every edit is recorded with its author and a note. If the scope described here has drifted from what actually gets posted, edit it rather than starting a topic… Go to post

Confounding: a third variable explains an apparent association. In randomised data, randomisation balances confounders. In observational data, confounders can be adjusted for but unknown ones cannot.

Correct me on the arithmetic if it is wrong; I would rather know.

32 likes in reply to #1 8mo

glossary_deskTL3Regular21 Nov 2025#6

Good question, well framed, and I would like to see it answered properly.

0 likes 8mo

l.krastevTL230 Nov 2025#7

Regression to the mean: if you select people with extreme values (very high or very low), their next measurement is often less extreme just by chance. This can look like a treatment effect when it is just statistics.

6 likes 8mo

glossary_checkTL2Member9 Dec 2025#8

l.krastev, post #7: Regression to the mean: if you select people with extreme values (very high or very low), their next measurement is often less extreme just by chance. This can look like a treatment effect when it is just statistics. Go to post

A distribution shown is worth ten summary statistics. Where a paper shows individual data points, read those first.

16 likes in reply to #7 8mo

y.rahimiTL217 Dec 2025#9

Adding a note of thanks rather than an opinion. I did not know most of that.

24 likes in reply to #1 7mo

Suggested topics

Topic	Participants	Replies	Views	Activity
Sample size intuition for a personal experiment — does this still hold? Research Methods › Statistics effect size heterogeneity number needed to treat Solved Sample size intuition for a personal experiment — does this still hold? I have a specific reason for asking rather than idle curiosity, and the context is below. Reading back through what has been written…	+33	38	26k	23d
Second pass at: Multiplicity when you track fifteen variables Research Methods › Statistics number needed to treat confounding effect size worked example Posting this under the heading it deserves: Second pass at: Multiplicity when you track fifteen variables Everything below is what sits behind that. Two things I would like separated before anyone answers on…		2	12k	18mo
Multiplicity when you track fifteen variables Research Methods › Statistics effect size heterogeneity confounding worked example Solved Multiplicity when you track fifteen variables — setting out what I have, and where I think it stops being reliable. Asking about multiplicity directly, because I have read four threads on it and each answered…		2	60k	11mo
What a confidence interval means, from scratch Research Methods › Statistics number needed to treat confounding worked example heterogeneity What a confidence interval means, from scratch — that is the question, and I have not found it answered plainly anywhere I have looked. Asking about confidence interval directly, because I have read four…		2	2.6k	2mo
Second pass at: Absence of evidence and evidence of absence Research Methods › Statistics worked example confounding number needed to treat On the subject in the title: Second pass at: Absence of evidence and evidence of absence Working notes rather than a conclusion. Two things I would like separated before anyone answers on Absence of evidence,…	+50	55	18k	16h

Related topics — sharing the tags effect size, number needed to treat, heterogeneity

Topic	Participants	Replies	Views	Activity
Comparing the STEP programme populations: who was actually enrolled Compounds › Semaglutide SELECT trial STEP programme semaglutide effect size Solved Comparing the STEP programme populations: who was actually enrolled Writing it up because I had to work it out twice and would rather nobody else did. Two things I would like separated before anyone answers…		2	7.5k	2mo
Triple agonism: additive, synergistic, or neither? Compounds › Retatrutide retatrutide effect size preprint Solved The question in the title: Triple agonism: additive, synergistic, or neither? I will give what I have already checked below so nobody repeats it. Asking about triple agonism on behalf of the question I keep…	+71	80	38k	11mo
Second pass at: What pooling buys you and what it destroys Evidence › Meta-analyses risk of bias meta-analysis heterogeneity Posting this under the heading it deserves: Second pass at: What pooling buys you and what it destroys Everything below is what sits behind that. Something about pooling buys does not reconcile and I would…	+32	36	19k	4mo
SURMOUNT-4 and what withdrawal data does and does not tell an individual — one year on Compounds › Tirzepatide SURMOUNT programme randomised trial tirzepatide SURPASS programme Solved Closed Posting this under the heading it deserves: SURMOUNT-4 and what withdrawal data does and does not tell an individual — one year on Everything below is what sits behind that. I would like to disagree carefully…	+122	136	48k	9mo
Journal club: STEP 1 and the treatment-policy estimand Evidence › Journal club journal club randomised trial number needed to treat risk of bias Journal club: STEP 1 and the treatment-policy estimand Writing it up because I had to work it out twice and would rather nobody else did. Session topic: STEP 1 ( N Engl J Med , 2021). Please read it before…		3	11k	9h

Sponsored Sigma-Aldrich — certified reference materialsReference standards, analytical reagents and certified materials for method calibration. If you are validating a method, this is where the standard comes from.Reference standards →