Ingo Rohlfing
@ingorohlfing.bsky.social
I am here for all interesting and funny posts on the social sciences, broadly understood and including open science and meta science, academia, teaching and research. https://linktr.ee/ingorohlfing
The added value is marginal compared to several blog posts on the topic because the product does not seem in any way superior to alternatives. Let's pass the benefit of a doubt here because the article states there is no conflict of interest. 2/
November 10, 2025 at 8:34 PM
The added value is marginal compared to several blog posts on the topic because the product does not seem in any way superior to alternatives. Let's pass the benefit of a doubt here because the article states there is no conflict of interest. 2/
True, the use cases and statistical concepts are very basic. I cannot believe that this is all they have to offer, but why not show it when one has more to offer? Speculation is futile, but let me speculate: I am a bit surprised such a basic article promoting a product has been published 1/
November 10, 2025 at 8:30 PM
True, the use cases and statistical concepts are very basic. I cannot believe that this is all they have to offer, but why not show it when one has more to offer? Speculation is futile, but let me speculate: I am a bit surprised such a basic article promoting a product has been published 1/
cover to approx. 95% what one needs for introductory statistics, and you don't have to worry about prompting and hallucinations.
For graduate training, it may be different, but I would then rather try Positron with LLM-assistance than a prompt-only interface. 4/
For graduate training, it may be different, but I would then rather try Positron with LLM-assistance than a prompt-only interface. 4/
November 10, 2025 at 4:51 PM
cover to approx. 95% what one needs for introductory statistics, and you don't have to worry about prompting and hallucinations.
For graduate training, it may be different, but I would then rather try Positron with LLM-assistance than a prompt-only interface. 4/
For graduate training, it may be different, but I would then rather try Positron with LLM-assistance than a prompt-only interface. 4/
I don't know about alternatives that would work better, but I guess there are some.
2) More importantly, I am not confident there is sufficient added value. For undergraduate training, JASP or Jamovi should work well with their GUI. (I use JASP in my course for illustration.) They are free, 3/
2) More importantly, I am not confident there is sufficient added value. For undergraduate training, JASP or Jamovi should work well with their GUI. (I use JASP in my course for illustration.) They are free, 3/
November 10, 2025 at 4:49 PM
I don't know about alternatives that would work better, but I guess there are some.
2) More importantly, I am not confident there is sufficient added value. For undergraduate training, JASP or Jamovi should work well with their GUI. (I use JASP in my course for illustration.) They are free, 3/
2) More importantly, I am not confident there is sufficient added value. For undergraduate training, JASP or Jamovi should work well with their GUI. (I use JASP in my course for illustration.) They are free, 3/
to me what the added value is in 2025. It discusses standard prompting w/o any insights into how students liked LLM-based training or so.
Two other points:
1) If one uses an LLM, I am not sure why one would use Julius. The free account does not suffice for teaching bc of a prompt limit. 2/
Two other points:
1) If one uses an LLM, I am not sure why one would use Julius. The free account does not suffice for teaching bc of a prompt limit. 2/
November 10, 2025 at 4:47 PM
to me what the added value is in 2025. It discusses standard prompting w/o any insights into how students liked LLM-based training or so.
Two other points:
1) If one uses an LLM, I am not sure why one would use Julius. The free account does not suffice for teaching bc of a prompt limit. 2/
Two other points:
1) If one uses an LLM, I am not sure why one would use Julius. The free account does not suffice for teaching bc of a prompt limit. 2/
Is this speaking from experience? If the sample size is derived from a smallest effect of substantive interest, or an otherwise well-justified effect size, this should not be a point of concern.
In practice, I can imagine reviewers getting caught up on N = 34 no matter what and say it is too small.
In practice, I can imagine reviewers getting caught up on N = 34 no matter what and say it is too small.
November 7, 2025 at 4:52 PM
Is this speaking from experience? If the sample size is derived from a smallest effect of substantive interest, or an otherwise well-justified effect size, this should not be a point of concern.
In practice, I can imagine reviewers getting caught up on N = 34 no matter what and say it is too small.
In practice, I can imagine reviewers getting caught up on N = 34 no matter what and say it is too small.
Reposted by Ingo Rohlfing
We might ask whether we still need data visualization training when we have powerful LLMs to help us. Certainly, LLMs can help to optimize our code. But without a profound understanding of the produced code, we run the risk of creating figures that may look nice but that misrepresent our data. (6/n)
November 6, 2025 at 1:58 PM
We might ask whether we still need data visualization training when we have powerful LLMs to help us. Certainly, LLMs can help to optimize our code. But without a profound understanding of the produced code, we run the risk of creating figures that may look nice but that misrepresent our data. (6/n)
In this sense, it is also not an optimal constellation. It is related to the idea that science advances one funeral at a time. www.aeaweb.org/articles?id=... Of course, this is only my impression and I am not naming any subfield or names here, but this seems to be more than just a "corner issue".
November 6, 2025 at 9:51 AM
In this sense, it is also not an optimal constellation. It is related to the idea that science advances one funeral at a time. www.aeaweb.org/articles?id=... Of course, this is only my impression and I am not naming any subfield or names here, but this seems to be more than just a "corner issue".
and lift each other's publications into reputable journals. This is then less a waste of resources compared to what you are adressing. Still, these are closed shops that are more or less immune from severe criticism and stifle theoretical or methodological innovation on their topic. 2/
November 6, 2025 at 9:48 AM
and lift each other's publications into reputable journals. This is then less a waste of resources compared to what you are adressing. Still, these are closed shops that are more or less immune from severe criticism and stifle theoretical or methodological innovation on their topic. 2/
My sense is this is not just a phenomenon in certain corners of science (or rather, academia). I think sometimes there are cliques closer to the core, meaning they work on relevant topics with, say, solid methods. From the outside, it looks they happen to review each other's papers 1/
November 6, 2025 at 9:45 AM
My sense is this is not just a phenomenon in certain corners of science (or rather, academia). I think sometimes there are cliques closer to the core, meaning they work on relevant topics with, say, solid methods. From the outside, it looks they happen to review each other's papers 1/
Reposted by Ingo Rohlfing
A spicier opinion is that academia taking back publishing is not necessarily a path to innovation and efficiency. Do you associate universities with efficiency? The problem with for-profit publishing is not the profit, it's the oligopoly power of major publishers. Anti-trust in our lifetime?
November 6, 2025 at 9:33 AM
A spicier opinion is that academia taking back publishing is not necessarily a path to innovation and efficiency. Do you associate universities with efficiency? The problem with for-profit publishing is not the profit, it's the oligopoly power of major publishers. Anti-trust in our lifetime?
Reposted by Ingo Rohlfing
If the author identifies as politically left, it would support his point, wouldn't it?
Regardless, this is neither theoretically nor empirically plausible and convincing
Regardless, this is neither theoretically nor empirically plausible and convincing
November 4, 2025 at 6:41 AM
If the author identifies as politically left, it would support his point, wouldn't it?
Regardless, this is neither theoretically nor empirically plausible and convincing
Regardless, this is neither theoretically nor empirically plausible and convincing
I had forgotten about this, but makes sense. Besides everything else, referencing a conversation with you without permission is bad. Based on the abstract, there appears to be a wide gulf between the strength of the claims and the strength of the evidence.
November 3, 2025 at 9:29 PM
I had forgotten about this, but makes sense. Besides everything else, referencing a conversation with you without permission is bad. Based on the abstract, there appears to be a wide gulf between the strength of the claims and the strength of the evidence.
Whether a discipline is homogeneous on the left or side of spectrum does not seem to matter, though the title emphasizes this. There is an empirical analysis in the paper, but I doubt it is close to being conclusive and counterarguments against the causal chain come to mind. 2/
November 3, 2025 at 9:02 PM
Whether a discipline is homogeneous on the left or side of spectrum does not seem to matter, though the title emphasizes this. There is an empirical analysis in the paper, but I doubt it is close to being conclusive and counterarguments against the causal chain come to mind. 2/
what the basis for the rankings was. The survey asked whether one had submitted to a journal, published in it, or reviewed for it. I don't know why this was asked for, as I don't think it is related to reputation (except maybe for having published in a journal) 2/
November 1, 2025 at 8:17 PM
what the basis for the rankings was. The survey asked whether one had submitted to a journal, published in it, or reviewed for it. I don't know why this was asked for, as I don't think it is related to reputation (except maybe for having published in a journal) 2/
A couple of weeks ago, I took part in a survey of journal reputation. Don't remember who runs it, but it was sponsored by the ECPR, I think. I may find the email again.
On that note: One had to rank 50 journals or so with option not to rank at all. Personally, I'd have found it interesting to ask 1/
On that note: One had to rank 50 journals or so with option not to rank at all. Personally, I'd have found it interesting to ask 1/
November 1, 2025 at 8:15 PM
A couple of weeks ago, I took part in a survey of journal reputation. Don't remember who runs it, but it was sponsored by the ECPR, I think. I may find the email again.
On that note: One had to rank 50 journals or so with option not to rank at all. Personally, I'd have found it interesting to ask 1/
On that note: One had to rank 50 journals or so with option not to rank at all. Personally, I'd have found it interesting to ask 1/