Evidence · Journal club · continued

Second pass at: Journal club: LEADER as the historical anchor posts 31–60

Evidence › Journal club randomised trial journal club effect size estimand

This is a continuation of a long topic, addressed by post number rather than by page. Start at post 1.

n.okwuosaTL216 Jul 2024#31

blank_injection, post #29: Post #26 is the version of this I will quote in future. One addition. A methods point on LEADER rather than a substantive one: if the comparison is not like for like, the difference you are measuring is the difference in method. Go to post

Reading order that works for these sessions: registry entry, methods, baseline table, primary result, then abstract last. Reading the abstract first anchors everything that follows.

It is worth stating the boring hypothesis before the interesting one.

0 likes in reply to #29 2y

hana.satoTL4 Moderator16 Jul 2024#32

v.bhattacharya, post #23: This follows post #21 rather than contradicting it. One more thing on LEADER that took me far too long to see: the two figures people quote are not measuring the same quantity. Once you notice that, the apparent contradiction disappears. Go to post

A request rather than an answer: could whoever has the primary source for LEADER post it? I have seen the claim three times this month and each version had lost a qualifier.

28 likes in reply to #23 2y

z.okonkwoTL216 Jul 2024#33

Where the group cannot agree, record the disagreement rather than resolving it by seniority. The recorded disagreement is more honest and more useful later.

9 likes 2y

p.iyer_pharmdTL3Pharmacist17 Jul 2024#34

Taking post #31 at face value and following it one step further.

The useful distinction on LEADER is between what was measured and what was inferred from it. Both end up in the same sentence and only one of them has error bars.

2 likes 2y

z.yildizTL217 Jul 2024 · edited#35

n.villalobos, post #30: I have no financial interest in anything named in this thread and I want to say so before I comment on LEADER, because it is the sort of subject where it matters. Go to post

I read post #31 twice before replying, because I had assumed the opposite.

What would change my mind on LEADER is a second dataset collected by someone with no stake in the first. Until then I hold it loosely and I would rather say so than pretend to more.

0 likes in reply to #30 2y

f.wojcikTL217 Jul 2024#36

Reading rather than answering, but this is the post I would point somebody at.

21 likes 2y

a.aguirreTL217 Jul 2024#37

Collapsed as off-topic by two members at trust level 3 or above

The baseline characteristics table is the single most useful page for the questions that get asked here, because it tells you who the result applies to.

It is a small point and it changes the answer, which is an awkward combination.

v.fontaineTL218 Jul 2024#38

The most common failure mode is spending fifty minutes on the effect size and ten on the population. Reversing that ratio would improve most sessions.

0 likes 2y

r.vukovicTL218 Jul 2024#39

v.szabo, post #7: If someone has run LEADER properly I would rather read that than my own reconstruction of it. Posting mine only because the thread has gone quiet. Go to post

I would rather this thread reach "we do not know" about LEADER than reach a confident answer that nobody can support when asked.

29 likes in reply to #7 2y

tracked_parcelTL2Regular18 Jul 2024#40

Post #39 is right about the mechanism and I think understates the practical bit.

The most useful thing anyone has posted about LEADER in this category was a table of what had been measured and by whom. That is what I would want again.

14 likes 2y

h.kimaniTL219 Jul 2024#41

The figures usually contain the finding and the text usually contains the interpretation. Separating them for the first twenty minutes is a discipline worth keeping.

2 likes 2y

a.schaefferTL2Member19 Jul 2024 · edited#42

LEADER is worth one more sentence than it usually gets, and the sentence is the one about how the number was arrived at.

9 likes 2y

n.kirchnerTL219 Jul 2024#43

d.vukovic, post #20: The baseline characteristics table is the single most useful page for the questions that get asked here, because it tells you who the result applies to. If anyone has run this properly I would rather read that than my own guess. Go to post

This follows post #40 rather than contradicting it.

Having read the whole LEADER thread before replying: the question in the first post has not actually been answered yet, and three of us have answered a nearby one instead.

21 likes in reply to #20 2y

integrator_traceTL2Member20 Jul 2024#44

Right — I had this wrong and I am glad to have read it before it mattered.

0 likes in reply to #30 2y

ak.kravchenkoTL220 Jul 2024#45

Post #43 is right about the mechanism and I think understates the practical bit.

Sessions on negative or null results are consistently the most instructive and the hardest to get anyone to attend.

0 likes 2y

ambient_draftTL3Regular20 Jul 2024#46

Where the group cannot agree, record the disagreement rather than resolving it by seniority. The recorded disagreement is more honest and more useful later.

5 likes 2y

s.lundgrenTL220 Jul 2024#47

The practical version of LEADER is three sentences long. The rigorous version is three pages and reaches the same conclusion with the conditions attached.

14 likes 2y

LJankowiakTL3Regular21 Jul 2024#48

f.amankwah, post #28: The most common failure mode is spending fifty minutes on the effect size and ten on the population. Reversing that ratio would improve most sessions. Go to post

On post #46 — agreed on the reasoning, with one qualification.

I would keep LEADER and the decision it usually gets used for separate in this thread. They are related and they are not the same question, and merging them is why the last one went badly.

29 likes in reply to #28 2y

ne.laurentTL221 Jul 2024 · edited#49

tracked_parcel, post #40: Post #39 is right about the mechanism and I think understates the practical bit. The most useful thing anyone has posted about LEADER in this category was a table of what had been measured and by whom. That is what I would want again. Go to post

Bookmarking this. I will come back when I have something worth adding.

8 likes in reply to #40 2y

n.bridgewaterTL2Member21 Jul 2024#50

Answering the question post #46 raises rather than the one it answers.

The most common failure mode is spending fifty minutes on the effect size and ten on the population. Reversing that ratio would improve most sessions.

20 likes 2y

e.ferrariTL221 Jul 2024#51

Answering the question post #50 raises rather than the one it answers.

Adding what did not work for me on LEADER, since the failures never get written up and they are half the useful information.

0 likes 2y

s.silvaTL222 Jul 2024#52

z.yildiz, post #35: I read post #31 twice before replying, because I had assumed the opposite. What would change my mind on LEADER is a second dataset collected by someone with no stake in the first. Until then I hold it loosely and I would rather say so than pretend to more. Go to post

The arithmetic in post #50 is right; the assumption feeding it is the part to check.

The figures usually contain the finding and the text usually contains the interpretation. Separating them for the first twenty minutes is a discipline worth keeping.

0 likes in reply to #35 2y

e.kimaniTL222 Jul 2024#53

My understanding of LEADER is a few years old and may have been superseded. If it has been, I would genuinely like to know rather than keep repeating it.

19 likes 2y

KnowltonTL3Regular22 Jul 2024 · edited#54

That matches what I have seen, for whatever a single anecdote is worth.

8 likes 2y

n.lehtinenTL223 Jul 2024#55

LEADER: I would want to see the raw numbers rather than the summary before agreeing. Summaries lose exactly the information that would settle this.

0 likes 2y

l.dziedzicTL223 Jul 2024#56

hana.sato, post #32: A request rather than an answer: could whoever has the primary source for LEADER post it? I have seen the claim three times this month and each version had lost a qualifier. Go to post

Adding the measurement that post #53 says would settle it.

The baseline characteristics table is the single most useful page for the questions that get asked here, because it tells you who the result applies to.

That holds for the case as described. Change the assumptions and it may not.

27 likes in reply to #32 2y

n.rahimiTL223 Jul 2024#57

Two sentences on LEADER and then I will stop, because the rest is speculation and the thread is better without mine.

What is documented is narrow. What is inferred from it is broad. The gap between them is where every argument here lives.

13 likes 2y

p.amankwahTL223 Jul 2024#58

The reason LEADER is hard to answer is that the obvious measurement and the relevant quantity are not the same thing, and substituting one for the other is silent.

4 likes 2y

d.oyelaranTL3Pharmacist24 Jul 2024#59

t.wojcik, post #21: The figures usually contain the finding and the text usually contains the interpretation. Separating them for the first twenty minutes is a discipline worth keeping. I would hold that lightly until someone with a larger sample weighs in. Go to post

Second this, and I would have said it less carefully.

4 likes in reply to #21 2y

c.tullochTL224 Jul 2024#60

Reading order that works for these sessions: registry entry, methods, baseline table, primary result, then abstract last. Reading the abstract first anchors everything that follows.

The honest answer is that it depends, and here is what it depends on.

0 likes 2y