【2022/7/15 開催レポート】 統計学的な有意差が必ずしも生物学的意味をもつわけではない。という点について、p値の意味やp-hacking、疑似反復、有意水準といった単語と絡めて話しました。また、最近の動向(添付リンク)や統計モデルによる推論といった提案も行いました。 https://t.co/DbVch2KHT6
If the conventional level of statistical significance for new discoveries were changed to .005 from .05 (which was suggested*), in my opinion ... (1/3) * https://t.co/YOhVDJzfdt
The three most cited articles in NHB are actually two perspectives and one comment. I think all of them are non-citable and therefore inflate the impact factor (but I'm not 100% sure about these). https://t.co/4MKnq9u4K6 https://t.co/ktohdt8BI8 https://t.c
RT @mathiaswullum: The funny thing is that programmatic Open Science papers boosted NHB's IF from the start: https://t.co/n84M3Gxf7D https…
The funny thing is that programmatic Open Science papers boosted NHB's IF from the start: https://t.co/n84M3Gxf7D https://t.co/yccpFAF4ZD https://t.co/XTkObJnzed
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
@Ilanim67 @vamrhein @FredBarraquand @Trends_Ecol_Evo there was the widely discussed paper that suggested to lower significance to 0.005 https://t.co/cDeT3r8t2k and of course they consider that there are cost (lowering power / increasing sample size) for th
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
The debate goes on. Part of the problem is that the false positive probability (resulting in failed replication) depends on other things in addition to the P-value and so different P-values required for different studies with different powers and differen
RT @leticiakawano: What about that #epitwitter? ;-) https://t.co/ud2raGDL3X
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
What about that #epitwitter? ;-)
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
RT @GuyattGH: Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of…
Very high proportion of articles reporting p < 0.05 and associated replication crisis have prompted calls for #threshold p of 0.005, ask researchers to justify threshold p. Better is abandon threshold p altogether, estimation only. https://t.co/fWZgWnVX
@NeuroPolarbear @bradpwyble @WiringTheBrain @neuralreckoning @atypical_me @KordingLab @Raamana_ @tyrell_turing @andpru @R3RT0 @katjaQheuer @OHBM @IpNeuro I tend to subjectively agree with this, but then again we have conventions on when we make the cut for
Journals can help transition to the new statistical significance threshold. Authors and readers can take the initiative by describing and interpreting results more appropriately in light of the new proposed definition of statistical significance. https://t
@DrCathyCorbin I do! FWIW, I have colleagues who think this is way too stringent/not necessary. https://t.co/bqSRH2kfQX
RT @RubinPsyc: (1) Low statistical standards of evidence “Statistical standards of evidence for claiming new discoveries in many fields of…
(1) Low statistical standards of evidence “Statistical standards of evidence for claiming new discoveries in many fields of science are simply too low” (Benjamin et al., 2018). https://t.co/nUe08gfZtB
RT @ho_ben: @KiraboJackson Our former colleague Dan Benjamin has been pushing us to use .005 p values and makes a compelling case. The chan…
We propose to change the default P-value threshold for statistical significance from 0.05 to 0.005 for claims of new discoveries. https://t.co/i04R2Gt7nM
Comment Published: 01 September 2017 Redefine statistical significance Daniel J. Benjamin, James O. Berger, […]Valen E. Johnson Nature Human Behaviour volume 2, pages6–10 (2018)Cite this article https://t.co/i04R2Gt7nM
https://t.co/i04R2Gt7nM ´´We propose to change the default P-value threshold for statistical significance from 0.05 to 0.005 for claims of new discoveries.´´ 2018
https://t.co/i04R2Gt7nM Redefine statistical significance Daniel J. Benjamin, James O. Berger, […]Valen E. Johnson Nature Human Behaviour volume 2, pages6–10 (2018)
´´One of the 72 authors, Christopher Chambers, is a member of the Advisory Board of Nature Human Behaviour. Christopher Chambers was not a corresponding author and did not communicate with the editors regarding the publication of this article.´´https://t.c
https://t.co/i04R2Gt7nM Redefine statistical significance Daniel J. Benjamin, James O. Berger, […]Valen E. Johnson Nature Human Behaviour volume 2, pages6–10 (2018)Cite this article 135k Accesses
@edwardswb @StatsSam @IreneSDavis The other reason is that even a p value of just *under* 0.05 (e.g., p = .04) is very weak evidence for an effect, so weakening the threshold even further by treating p values of around 0.06 or 0.07 as constituting evidence
Matthew 先生によると、「ポスト p < 0.05」 の一つの指針として解釈できるのが、2017年に Nature Human Behaviour に投稿された 『p < 0.005 の推奨』です。 https://t.co/RaRfdce379
@R__INDEX Benjamin et al. (2018) proposed p = .005 as a new threshold which has since been cited relatively frequently https://t.co/DRYsYe4sh4
@SNageebAli Sometimes, though, you do get a https://t.co/qNuifCR4xr.
To make H0 and H1 the only possible alternatives for the population, one could perhaps fo this: Choose a cut-off, where seems relavant. H0 is that value for pop parameter is below cut-off. H1 that is equal to, or above. ...
Hej #statistiktwitter Har lite problem med nollhypotesprövning som metod, och ej minst diskussionen om "sant/falskt pos/neg", såsom presenteras i den här artikeln: https://t.co/StypPQxB27 Se ffa fig 2.
@karimbrohi https://t.co/S1lczGRpZf Many more articles like this about especially practice changing or major shift findings. I do think either should have better p value or two studies to show its correct problem is not many authors want to do a subsequent
@coproduto É por isso também que qualquer estudo que se preze mostrará intervalos de confiança e tamanho de efeito (r squared, Cohen, etc), não apenas p-valor. E, de preferência, pensar em p-valor de uma forma diferente para novas descobertas/claims bizar
RT @FrogHarmless: @kdpsinghlab @maximumcharacte Maybe I should just start interpreting studies where p>0.005 as nulls. https://t.co/mzzI2…
@kdpsinghlab @maximumcharacte Maybe I should just start interpreting studies where p>0.005 as nulls. https://t.co/mzzI21HgVd
それはそうかも。
Redefine statistical significance https://t.co/LIu5FpysKr まぁp値はそれ以外にも色々問題はあるんですけどね。
Läsvärt om statistisk signifikans https://t.co/tMGCp7Erm5
@dan_a_chapman I don't know, it's kind of nice to have a comedy subgenre of the literature https://t.co/RUGQq9J5cm
@TDeryugina Ugh, that's not good if this is a trend. If we are trending in any direction I hope it would be more towards a 99.5% confidence interval. https://t.co/VNcSAFPXp0.
@NikhilAgrawalMD @AcademicChatter Benjamin et al. have argued since 2018 that the default p-value for new discoveries should be 0.005. https://t.co/BLffZ1fwaV
I keep reading about meta-science and how to improve science. It's a few years old but this paper is very nice and pragmatic option: change the cut off for statistical significance from 0.05 to 0.005. #Statistics #science https://t.co/raaBF6q6iD
@DanielLaParra15 @amloii Te dejo más referencias sobre eso por si quieres indagar algunas discusiones en la literatura sobre ello https://t.co/McRvBKnZAD https://t.co/0GTO8SzbKT https://t.co/wwtaEqx9KN https://t.co/5kwIp8Etlv https://t.co/t4ioDbuRNk
RT @tdienlin: In the tradition of @lakens et al's https://t.co/53ch9GItCm or https://t.co/ySKNibXv48, we set up a google doc, in which we'…
(11/14) a group of outstanding statisticians called for “Redefine statistical significance” (Benjamin et al., 2017; https://t.co/qyH4Ijxhro ) by changing the default P-value threshold for statistical significance from 0.05 to 0.005.
@Andrew___Baker @jiafengchen42 Bumping up the t-stat isn’t the only proposed solution btw. This paper (https://t.co/eLNmQ7OEjo) by @Dan_J_Benjamin et al, suggests changing the significance threshold to .005, while @lakens et al (https://t.co/JXEcxXA1Od) ad
@AliOilhan Ali Hocam bir gün kitabını da yaz lütfen. Ya türkçe olsun, ya da türkçeye çevrilsin. İhtiyaç var. Bu arada istatistikçiler kazan kaldırmıştı zaten p değeri 0.05'den 0.005'e düşürülsün diye. Özellikle klinikçileri zorlayacak bir uygulama olacak t
@jrosell Para los quisquillosos como nosotros xD https://t.co/ZAuikyS0yS
RT @lakens: OK, I am just going to say it: At these times, we should not be divided. As researchers are massively rushing to do corona rela…
RT @BrianNosek: Pandemics can unite us all against a common foe, p<.05.
RT @lakens: OK, I am just going to say it: At these times, we should not be divided. As researchers are massively rushing to do corona rela…
Interesting discussion
RT @lakens: OK, I am just going to say it: At these times, we should not be divided. As researchers are massively rushing to do corona rela…
RT @lakens: OK, I am just going to say it: At these times, we should not be divided. As researchers are massively rushing to do corona rela…
Sure, but then please still publish (or at least publicize) the nulls.