Daniël Lakens @lakens 1640 days ago

@DrDominikVogel @ingorohlfing @Chengxin_M_Xu @Journal_BPA As I explain in https://t.co/HNyInPv1xY and https://t.co/2H8XIkaaf3, the pattern of p-values is influenced by many factors, especially in a heterogeneous set of studies like this. If there were only true effects, the curve should be steeper.

Daniël Lakens @lakens 1846 days ago

@MarkZobeck @ErikVanZwet @aidybarnett That's the rub - you are already grouping pub bias and researchers DoF together, and you assume you can say anything about them in this graph. I explain in this 2015 paper that this is very difficult: https://t.co/s2Rq4iX7wE.

Daniël Lakens @lakens 1848 days ago

@shilaan01 That's a real hornet's nest. People have tried to identify p-hacking. Here's a comment of mine explaining why that doesn't work: https://t.co/2H8XIkaaf3 and here's a paper explaining why it is practically impossible even with perfect data: https://t.co/s2Rq4iX7wE

Daniël Lakens @lakens 1926 days ago

@wuthrich_k @jhaushofer Very cool - you might be interested in some early criticism I wrote on papers using p-curves to conclude widespread bias, where I also pointed out it's not that easy: https://t.co/s2Rq4iX7wE and https://t.co/2H8XIkaaf3.

Daniël Lakens @lakens 2226 days ago

@HammadSheikh Sure, but people who try to find it can not provide proof it happens on a widespread scale https://t.co/SlHZ1IKdZQ but feel free to try.

Daniël Lakens @lakens 2226 days ago

@profjmb See https://t.co/SlHZ1IKdZQ - if you want to make specific clsims you will need to model those and as my papers shows that is not easy. People widely misunderstand this.

Daniël Lakens @lakens 2226 days ago

@ignaziano Because https://t.co/SlHZ1IKdZQ

Daniël Lakens @lakens 2279 days ago

@rubenarslan @RogertheGS @richarddmorey @ceptional @LeonidTiokhin Exactly. I probably wrote the best paper in the literature on the challenges of drawing conclusions of p-values just below .05 https://t.co/SlHZ1IKdZQ And I am still ok with using this as a general heuristic for non-prereg papers in journal we know commit massive publication bias

Josh Baxter @joshrbaxter 2552 days ago

2019 resolution- rely less on p-values and more on persuasive science. Great read by ⁦@lakens⁩ - another reason we should rethink how we use p values when formulating hypothesis. Just because something isn’t<.05 doesn’t make it true, and vice versa. https://t.co/haj7VxcRnh

The Viking Turtle @TheVikingTurtle 2553 days ago

RT @pash22: On the challenges of drawing conclusions from p-values just below 0.05 https://t.co/GpLVb3vri6 via @lakens

Will Johnson @W_O_Johnson 2553 days ago

RT @pash22: On the challenges of drawing conclusions from p-values just below 0.05 https://t.co/GpLVb3vri6 via @lakens

Professor Red Owl @ProfRedOwl 2553 days ago

RT @pash22: On the challenges of drawing conclusions from p-values just below 0.05 https://t.co/GpLVb3vri6 via @lakens

Andrew Booth @AndrewB007h 2553 days ago

RT @pash22: On the challenges of drawing conclusions from p-values just below 0.05 https://t.co/GpLVb3vri6 via @lakens

Vlad Malik @VladMalik 2553 days ago

RT @pash22: On the challenges of drawing conclusions from p-values just below 0.05 https://t.co/GpLVb3vri6 via @lakens

Thomas Rhys Evans @ThomasRhysEvans 2553 days ago

RT @pash22: On the challenges of drawing conclusions from p-values just below 0.05 https://t.co/GpLVb3vri6 via @lakens

Felipe Mattioni, MSc @felipe_mattioni 2553 days ago

RT @pash22: On the challenges of drawing conclusions from p-values just below 0.05 https://t.co/GpLVb3vri6 via @lakens

Roman Luštrik @romunov 2553 days ago

https://t.co/k73KJWPf8U

Thomas Bandholm @TBandholm 2553 days ago

RT @pash22: On the challenges of drawing conclusions from p-values just below 0.05 https://t.co/GpLVb3vri6 via @lakens

Bimandra Djaafara @andradjaafara 2553 days ago

RT @pash22: On the challenges of drawing conclusions from p-values just below 0.05 https://t.co/GpLVb3vri6 via @lakens

Roxanne Connelly @ConnellyRoxanne 2553 days ago

RT @pash22: On the challenges of drawing conclusions from p-values just below 0.05 https://t.co/GpLVb3vri6 via @lakens

Darren L Dahly @statsepi 2553 days ago

RT @pash22: On the challenges of drawing conclusions from p-values just below 0.05 https://t.co/GpLVb3vri6 via @lakens

Daniël Lakens @lakens 2553 days ago

RT @pash22: On the challenges of drawing conclusions from p-values just below 0.05 https://t.co/GpLVb3vri6 via @lakens

dave fuller @cd_fuller 2553 days ago

RT @pash22: On the challenges of drawing conclusions from p-values just below 0.05 https://t.co/GpLVb3vri6 via @lakens

Ash Paul @pash22 2553 days ago

On the challenges of drawing conclusions from p-values just below 0.05 https://t.co/GpLVb3vri6 via @lakens

Elizabeth Beard @thetokenhipster 2661 days ago

RT @flourneuro: Why hello problematic distribution of test statistics in economics. Also, tools to help thinking about underlying models of…