Uri Simonsohn
@urisohn.bsky.social
Pragmatic methodologist and behavioral scientist at Esade business school. Barcelona
website: https://urisohn.com
Blog: https://DataColada.org
website: https://urisohn.com
Blog: https://DataColada.org
It's what we choose to focus on actually.
June 15, 2025 at 1:15 PM
It's what we choose to focus on actually.
I dont think thats a realistic concern if you just browse the home page. It's a funny headline, but not super diagnostic
June 15, 2025 at 1:02 PM
I dont think thats a realistic concern if you just browse the home page. It's a funny headline, but not super diagnostic
Mission accomplished!
April 5, 2025 at 12:43 PM
Mission accomplished!
Fun paper in Journal Club today, from 1954
Participants were not given the questions but had to answer in standard multiple choice, agree/disagree, True/False, etc.
I guess in 1954 the review process was a lot more expedient.
Participants were not given the questions but had to answer in standard multiple choice, agree/disagree, True/False, etc.
I guess in 1954 the review process was a lot more expedient.
February 14, 2025 at 1:10 PM
Fun paper in Journal Club today, from 1954
Participants were not given the questions but had to answer in standard multiple choice, agree/disagree, True/False, etc.
I guess in 1954 the review process was a lot more expedient.
Participants were not given the questions but had to answer in standard multiple choice, agree/disagree, True/False, etc.
I guess in 1954 the review process was a lot more expedient.
To any lawyers who may see this.
is this (a) normal, (b) funny, (c) both:
is this (a) normal, (b) funny, (c) both:
November 26, 2024 at 9:04 PM
To any lawyers who may see this.
is this (a) normal, (b) funny, (c) both:
is this (a) normal, (b) funny, (c) both:
I include this in most shareable R scripts I write
November 20, 2024 at 7:50 PM
I include this in most shareable R scripts I write
Such good user interface for interacting wth .PDF
You say what you are doing: reading, editing etc.
Related tools are available.
I got sick (again) of Adobe, winner so far: pdf.easeus.com
You say what you are doing: reading, editing etc.
Related tools are available.
I got sick (again) of Adobe, winner so far: pdf.easeus.com
November 14, 2024 at 2:43 PM
Such good user interface for interacting wth .PDF
You say what you are doing: reading, editing etc.
Related tools are available.
I got sick (again) of Adobe, winner so far: pdf.easeus.com
You say what you are doing: reading, editing etc.
Related tools are available.
I got sick (again) of Adobe, winner so far: pdf.easeus.com
I mean it in classical social psych type of way.
The way Lee Ross would have used it.
You think a phenomenon is true, and run a memorable experiment to demonstrate it.
e.g., from Ross 2018 J Persp article:
The way Lee Ross would have used it.
You think a phenomenon is true, and run a memorable experiment to demonstrate it.
e.g., from Ross 2018 J Persp article:
November 14, 2024 at 12:20 PM
I mean it in classical social psych type of way.
The way Lee Ross would have used it.
You think a phenomenon is true, and run a memorable experiment to demonstrate it.
e.g., from Ross 2018 J Persp article:
The way Lee Ross would have used it.
You think a phenomenon is true, and run a memorable experiment to demonstrate it.
e.g., from Ross 2018 J Persp article:
It's surprising that alpha<1 here.
November 11, 2024 at 1:21 PM
It's surprising that alpha<1 here.
There is a famous article titled:
"How to show 9 is bigger than 221"...
It critiques between-subject comparisons.
Not enough people know the result is artifactual.
You only get it when using a 1-10 scale (!)
So people say "well, 9 is 9 big".
Change scale, no effect.
"How to show 9 is bigger than 221"...
It critiques between-subject comparisons.
Not enough people know the result is artifactual.
You only get it when using a 1-10 scale (!)
So people say "well, 9 is 9 big".
Change scale, no effect.
October 25, 2024 at 1:56 PM
There is a famous article titled:
"How to show 9 is bigger than 221"...
It critiques between-subject comparisons.
Not enough people know the result is artifactual.
You only get it when using a 1-10 scale (!)
So people say "well, 9 is 9 big".
Change scale, no effect.
"How to show 9 is bigger than 221"...
It critiques between-subject comparisons.
Not enough people know the result is artifactual.
You only get it when using a 1-10 scale (!)
So people say "well, 9 is 9 big".
Change scale, no effect.
update: blueark.app works:
this covid themed *tweet* long pre-dates my bluesky account but is on my bluesky feed now.
re-deactivating twitter now.
this covid themed *tweet* long pre-dates my bluesky account but is on my bluesky feed now.
re-deactivating twitter now.
October 21, 2024 at 3:10 PM
update: blueark.app works:
this covid themed *tweet* long pre-dates my bluesky account but is on my bluesky feed now.
re-deactivating twitter now.
this covid themed *tweet* long pre-dates my bluesky account but is on my bluesky feed now.
re-deactivating twitter now.
Is this definition as bad it seems to me?
October 18, 2024 at 11:52 AM
Is this definition as bad it seems to me?
From the same talk
October 10, 2024 at 8:03 AM
From the same talk
At a recent marketing conference I gave the talk "Catching up to Psychology in Credibility". Slide from that talk:
October 10, 2024 at 8:01 AM
At a recent marketing conference I gave the talk "Catching up to Psychology in Credibility". Slide from that talk:
October 8, 2024 at 8:21 PM
I think this is less satisfactory because you need to navigate your way to each item to see the description. If you have list of lists as output, or large objects, that can become tough.
Or is there an easy way to see all descriptions for list objects in an orderly fashion?
Or is there an easy way to see all descriptions for list objects in an orderly fashion?
September 19, 2024 at 8:25 AM
I think this is less satisfactory because you need to navigate your way to each item to see the description. If you have list of lists as output, or large objects, that can become tough.
Or is there an easy way to see all descriptions for list objects in an orderly fashion?
Or is there an easy way to see all descriptions for list objects in an orderly fashion?
I have started doing something when writing R functions I wished other people did for their functions.
The function's output is a list where the last object is a 'table of contents' for all other items in the list.
The function's output is a list where the last object is a 'table of contents' for all other items in the list.
September 19, 2024 at 7:19 AM
I have started doing something when writing R functions I wished other people did for their functions.
The function's output is a list where the last object is a 'table of contents' for all other items in the list.
The function's output is a list where the last object is a 'table of contents' for all other items in the list.
I need to ask a question for a friend.
Please answer poll about this figure in the reply to this tweet.
Please answer poll about this figure in the reply to this tweet.
June 18, 2024 at 3:34 PM
I need to ask a question for a friend.
Please answer poll about this figure in the reply to this tweet.
Please answer poll about this figure in the reply to this tweet.
I'm giving a Zoom talk in the seminar series I call "EuroDecision"
Title: "Stimulus Sampling Reimagined"
Tuesday June 4th 2024
It's at 12.30 PM Barcelona / Berlin time
(Yes, 6.30 AM ET)
Register for free here: insead.zoom.us/meeting/regi...
paper: urisohn.com/44
Title: "Stimulus Sampling Reimagined"
Tuesday June 4th 2024
It's at 12.30 PM Barcelona / Berlin time
(Yes, 6.30 AM ET)
Register for free here: insead.zoom.us/meeting/regi...
paper: urisohn.com/44
June 3, 2024 at 8:42 AM
I'm giving a Zoom talk in the seminar series I call "EuroDecision"
Title: "Stimulus Sampling Reimagined"
Tuesday June 4th 2024
It's at 12.30 PM Barcelona / Berlin time
(Yes, 6.30 AM ET)
Register for free here: insead.zoom.us/meeting/regi...
paper: urisohn.com/44
Title: "Stimulus Sampling Reimagined"
Tuesday June 4th 2024
It's at 12.30 PM Barcelona / Berlin time
(Yes, 6.30 AM ET)
Register for free here: insead.zoom.us/meeting/regi...
paper: urisohn.com/44
I'm giving a talk in the seminar series I call "EuroDecision"
Title: "Stimulus Sampling Reimagined"
Tuesday June 4th 2024
It's at 12.30 PM Barcelona/Berlin/Amsterdam time
(Yes, 6.30 AM ET)
Register for free...
Title: "Stimulus Sampling Reimagined"
Tuesday June 4th 2024
It's at 12.30 PM Barcelona/Berlin/Amsterdam time
(Yes, 6.30 AM ET)
Register for free...
June 3, 2024 at 8:21 AM
I'm giving a talk in the seminar series I call "EuroDecision"
Title: "Stimulus Sampling Reimagined"
Tuesday June 4th 2024
It's at 12.30 PM Barcelona/Berlin/Amsterdam time
(Yes, 6.30 AM ET)
Register for free...
Title: "Stimulus Sampling Reimagined"
Tuesday June 4th 2024
It's at 12.30 PM Barcelona/Berlin/Amsterdam time
(Yes, 6.30 AM ET)
Register for free...
I may have shared this before.
With N=12.8 million p>.05 on effect of birthday number on people's address number.
sci-hub.se/10.1037/a002...
With N=12.8 million p>.05 on effect of birthday number on people's address number.
sci-hub.se/10.1037/a002...
February 25, 2024 at 9:39 PM
I may have shared this before.
With N=12.8 million p>.05 on effect of birthday number on people's address number.
sci-hub.se/10.1037/a002...
With N=12.8 million p>.05 on effect of birthday number on people's address number.
sci-hub.se/10.1037/a002...
A bit of an evergreen.
When power is x%, the share of p<.01 among p<.05 results is approximately x%
R Code:
ncp=seq(0,4,.01)
power05 =(1-pt(qt(.975,df=5000),df=5000,ncp=ncp)) power01 = (1-t(qt(.995,df=5000),df=5000,ncp=ncp)) share01 = power01/power05
When power is x%, the share of p<.01 among p<.05 results is approximately x%
R Code:
ncp=seq(0,4,.01)
power05 =(1-pt(qt(.975,df=5000),df=5000,ncp=ncp)) power01 = (1-t(qt(.995,df=5000),df=5000,ncp=ncp)) share01 = power01/power05
February 22, 2024 at 2:52 PM
A bit of an evergreen.
When power is x%, the share of p<.01 among p<.05 results is approximately x%
R Code:
ncp=seq(0,4,.01)
power05 =(1-pt(qt(.975,df=5000),df=5000,ncp=ncp)) power01 = (1-t(qt(.995,df=5000),df=5000,ncp=ncp)) share01 = power01/power05
When power is x%, the share of p<.01 among p<.05 results is approximately x%
R Code:
ncp=seq(0,4,.01)
power05 =(1-pt(qt(.975,df=5000),df=5000,ncp=ncp)) power01 = (1-t(qt(.995,df=5000),df=5000,ncp=ncp)) share01 = power01/power05
A bit of an evergreen.
When power is x%, the share of p<.01 among p<.05 results is approximately x%
R Code:
ncp=seq(0,4,.01)
power05 =(1-pt(qt(.975,df=5000),df=5000,ncp=ncp))
power01 = (1-pt(qt(.995,df=5000),df=5000,ncp=ncp))
share01 = power01/power05
When power is x%, the share of p<.01 among p<.05 results is approximately x%
R Code:
ncp=seq(0,4,.01)
power05 =(1-pt(qt(.975,df=5000),df=5000,ncp=ncp))
power01 = (1-pt(qt(.995,df=5000),df=5000,ncp=ncp))
share01 = power01/power05
February 22, 2024 at 2:50 PM
A bit of an evergreen.
When power is x%, the share of p<.01 among p<.05 results is approximately x%
R Code:
ncp=seq(0,4,.01)
power05 =(1-pt(qt(.975,df=5000),df=5000,ncp=ncp))
power01 = (1-pt(qt(.995,df=5000),df=5000,ncp=ncp))
share01 = power01/power05
When power is x%, the share of p<.01 among p<.05 results is approximately x%
R Code:
ncp=seq(0,4,.01)
power05 =(1-pt(qt(.975,df=5000),df=5000,ncp=ncp))
power01 = (1-pt(qt(.995,df=5000),df=5000,ncp=ncp))
share01 = power01/power05
Authors writing "To the best of our knowledge, we are the first to..."
--> please include a footnote showing us how hard you tried to find earlier work.
--> please include a footnote showing us how hard you tried to find earlier work.
December 2, 2023 at 5:19 PM
Authors writing "To the best of our knowledge, we are the first to..."
--> please include a footnote showing us how hard you tried to find earlier work.
--> please include a footnote showing us how hard you tried to find earlier work.