Vilgot Huhn
@vilgothuhn.bsky.social
Confused PhD student in psychology at Karolinska Institutet, Stockholm. GAD, ICBT, mechanisms of change. Organizing the ReproducibiliTea JC at KI.
Website: https://vilgot-huhn.github.io/mywebsite/
Personal blog at unconfusion.substack.com
Website: https://vilgot-huhn.github.io/mywebsite/
Personal blog at unconfusion.substack.com
It's not exact so there's probably nothing actually interesting here though.
November 25, 2025 at 7:57 PM
It's not exact so there's probably nothing actually interesting here though.
Recent discussion on Lindley's paradox got me simulating again and it's kind of quirky how the mean p-value for a distribution of 50% power t-tests end up at a value with so similar probability density to the null distribution 🤷♂️
November 25, 2025 at 7:57 PM
Recent discussion on Lindley's paradox got me simulating again and it's kind of quirky how the mean p-value for a distribution of 50% power t-tests end up at a value with so similar probability density to the null distribution 🤷♂️
Loving poetryfoundations error message
November 18, 2025 at 1:13 PM
Loving poetryfoundations error message
Hm. Imo this post brushes the text-mining issue under the rug a bit. The viral histogram doesn’t just communicate the existence of a problem, it also implies a magnitude of the problem. The plot is shared as being about publication bias, not reporting conversations that are ”still not a good thing”.
November 15, 2025 at 6:27 PM
Hm. Imo this post brushes the text-mining issue under the rug a bit. The viral histogram doesn’t just communicate the existence of a problem, it also implies a magnitude of the problem. The plot is shared as being about publication bias, not reporting conversations that are ”still not a good thing”.
My main goal for my PhD is to sneak in a reference to Ferrie (2019) somewhere.
November 11, 2025 at 11:32 AM
My main goal for my PhD is to sneak in a reference to Ferrie (2019) somewhere.
asteriskmag.com/issues/12-bo...
Thought this quote summarized the tensions and contradictions around psychiatric diagnoses in an illuminating way. Good nuanced article. I have my own unstructured thoughts on the subject (as do most clinicians, I think).
Thought this quote summarized the tensions and contradictions around psychiatric diagnoses in an illuminating way. Good nuanced article. I have my own unstructured thoughts on the subject (as do most clinicians, I think).
October 28, 2025 at 6:51 PM
asteriskmag.com/issues/12-bo...
Thought this quote summarized the tensions and contradictions around psychiatric diagnoses in an illuminating way. Good nuanced article. I have my own unstructured thoughts on the subject (as do most clinicians, I think).
Thought this quote summarized the tensions and contradictions around psychiatric diagnoses in an illuminating way. Good nuanced article. I have my own unstructured thoughts on the subject (as do most clinicians, I think).
Your meme is what should happen when others mention confounders to me. This tweet I left in the drafts is what should happen every time I mention confounders to others.
October 28, 2025 at 12:10 PM
Your meme is what should happen when others mention confounders to me. This tweet I left in the drafts is what should happen every time I mention confounders to others.
My got o example to generate LLM hallucinations used to be to ask them about somewhat obscure graphic novels, but trying it now I found Claude is able to see if it such information exists in the training data and go into search mode instead. An interesting development imo.
October 20, 2025 at 5:36 PM
My got o example to generate LLM hallucinations used to be to ask them about somewhat obscure graphic novels, but trying it now I found Claude is able to see if it such information exists in the training data and go into search mode instead. An interesting development imo.
Very ominous dag. Unclear figure but imo risk of collider bias dissolving ATF causing Third Impact?
2/10, would not estimate.
2/10, would not estimate.
October 17, 2025 at 9:56 AM
Very ominous dag. Unclear figure but imo risk of collider bias dissolving ATF causing Third Impact?
2/10, would not estimate.
2/10, would not estimate.
My friend @mkgood.bsky.social did so in the supplemental materials for one of his papers. That was the last time I saw it.
October 15, 2025 at 6:50 AM
My friend @mkgood.bsky.social did so in the supplemental materials for one of his papers. That was the last time I saw it.
I think this works as an intuitive explanation for multicolinearity, right? When the cloud is flat the plane is attached in a very stable way. The table has many legs. But when we flatten that cloud with an association between X2 and X2 the table becomes "wobbly" orthogonal to that axis.
October 14, 2025 at 9:07 AM
I think this works as an intuitive explanation for multicolinearity, right? When the cloud is flat the plane is attached in a very stable way. The table has many legs. But when we flatten that cloud with an association between X2 and X2 the table becomes "wobbly" orthogonal to that axis.
Here's how I thought about it at the time:
"One is trying to chose a SD to divide by and man isn't the residual kind of an SD? I mean the residual of an intercept only model (some call it an "empty model") is the same thing as the standard deviation around a mean."
"One is trying to chose a SD to divide by and man isn't the residual kind of an SD? I mean the residual of an intercept only model (some call it an "empty model") is the same thing as the standard deviation around a mean."
October 7, 2025 at 7:31 AM
Here's how I thought about it at the time:
"One is trying to chose a SD to divide by and man isn't the residual kind of an SD? I mean the residual of an intercept only model (some call it an "empty model") is the same thing as the standard deviation around a mean."
"One is trying to chose a SD to divide by and man isn't the residual kind of an SD? I mean the residual of an intercept only model (some call it an "empty model") is the same thing as the standard deviation around a mean."
I’ve seen it done a couple of times for model estimated within-group effect size, which in simulated data tended to give Big Number. Did a talk about this stuff on a (small casual) conference last year.
October 7, 2025 at 7:14 AM
I’ve seen it done a couple of times for model estimated within-group effect size, which in simulated data tended to give Big Number. Did a talk about this stuff on a (small casual) conference last year.
Thank you! You’re probably right. I’ll see if I grasp it.
Here’s my DAG attempt. In A. the treatment affects the outcome and mediator, but M -> Y is confounded by something unmeasured. In B. this influence on M and Y goes through ”l” (which also creates the M-Y association)
Here’s my DAG attempt. In A. the treatment affects the outcome and mediator, but M -> Y is confounded by something unmeasured. In B. this influence on M and Y goes through ”l” (which also creates the M-Y association)
September 3, 2025 at 9:00 AM
Thank you! You’re probably right. I’ll see if I grasp it.
Here’s my DAG attempt. In A. the treatment affects the outcome and mediator, but M -> Y is confounded by something unmeasured. In B. this influence on M and Y goes through ”l” (which also creates the M-Y association)
Here’s my DAG attempt. In A. the treatment affects the outcome and mediator, but M -> Y is confounded by something unmeasured. In B. this influence on M and Y goes through ”l” (which also creates the M-Y association)
We're doing a shorter series ReproducibiliTea JC seminars this semester. The open science focus has gradually changed towards "methods stuff" more generally (but it's a fuzzy line of course). The participants aren't meta-science researchers so I try to keep it broad.
August 29, 2025 at 8:28 AM
We're doing a shorter series ReproducibiliTea JC seminars this semester. The open science focus has gradually changed towards "methods stuff" more generally (but it's a fuzzy line of course). The participants aren't meta-science researchers so I try to keep it broad.
Reminded me of this classic
August 11, 2025 at 6:06 PM
Reminded me of this classic
The linked paper says it has open data so at least there, if one wants to contextualize the effect better one could downloa-Wait what is this?
August 9, 2025 at 5:33 AM
The linked paper says it has open data so at least there, if one wants to contextualize the effect better one could downloa-Wait what is this?
In case anyone watching #eurovision wonders whether the same increasing chasm between elites and the public which Martin Gurri describes in "Revolt of The Public" can be seen when regressing televotes on jury votes:
no, not really.
#stats
no, not really.
#stats
May 17, 2025 at 7:41 PM
In case anyone watching #eurovision wonders whether the same increasing chasm between elites and the public which Martin Gurri describes in "Revolt of The Public" can be seen when regressing televotes on jury votes:
no, not really.
#stats
no, not really.
#stats
Yeah. I think Rationalists fundamentally are people who are drawn to the idea that if you’re a smart disciplined rational thinker you can get around having to actually do the reading.
Related thing I wrote on the other site:
Related thing I wrote on the other site:
May 14, 2025 at 9:26 AM
Yeah. I think Rationalists fundamentally are people who are drawn to the idea that if you’re a smart disciplined rational thinker you can get around having to actually do the reading.
Related thing I wrote on the other site:
Related thing I wrote on the other site:
Testing if uploading it as video works
May 7, 2025 at 1:46 PM
Testing if uploading it as video works
I made a shiny app to explore what random and fixed effects means in a longitudinal mixed model:
vilgothuhn.shinyapps.io/lmm_explorer2/
vilgothuhn.shinyapps.io/lmm_explorer2/
May 1, 2025 at 9:06 PM
I made a shiny app to explore what random and fixed effects means in a longitudinal mixed model:
vilgothuhn.shinyapps.io/lmm_explorer2/
vilgothuhn.shinyapps.io/lmm_explorer2/
