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
Dealing with 1 color #dataviz
www.matplotlib-journey.com/module2/deal... Very useful intro to the different ways of picking a color. Didn't know so far what RGB stands for and how hEX codes work. Plus there is the HSV method combining color, saturation and brightness.
www.matplotlib-journey.com/module2/deal... Very useful intro to the different ways of picking a color. Didn't know so far what RGB stands for and how hEX codes work. Plus there is the HSV method combining color, saturation and brightness.
November 3, 2025 at 4:44 PM
Dealing with 1 color #dataviz
www.matplotlib-journey.com/module2/deal... Very useful intro to the different ways of picking a color. Didn't know so far what RGB stands for and how hEX codes work. Plus there is the HSV method combining color, saturation and brightness.
www.matplotlib-journey.com/module2/deal... Very useful intro to the different ways of picking a color. Didn't know so far what RGB stands for and how hEX codes work. Plus there is the HSV method combining color, saturation and brightness.
Defense Against Dishonest Charts
flowingdata.com/projects/dis... #dataviz A highly instructive interactive guide on how charts can be deceiving and what to watch out for to avoid misinterpretation.
flowingdata.com/projects/dis... #dataviz A highly instructive interactive guide on how charts can be deceiving and what to watch out for to avoid misinterpretation.
October 24, 2025 at 3:47 PM
Defense Against Dishonest Charts
flowingdata.com/projects/dis... #dataviz A highly instructive interactive guide on how charts can be deceiving and what to watch out for to avoid misinterpretation.
flowingdata.com/projects/dis... #dataviz A highly instructive interactive guide on how charts can be deceiving and what to watch out for to avoid misinterpretation.
The bixplot: A variation on the boxplot suited for bimodal and multimodal data
arxiv.org/abs/2510.09276 #dataviz It is a multiboxplot or multiviolinplot, depending on the data. Clustering is used distinguish groups of cases from each other that are plotted separately, but in one chart.
arxiv.org/abs/2510.09276 #dataviz It is a multiboxplot or multiviolinplot, depending on the data. Clustering is used distinguish groups of cases from each other that are plotted separately, but in one chart.
October 15, 2025 at 9:09 AM
The bixplot: A variation on the boxplot suited for bimodal and multimodal data
arxiv.org/abs/2510.09276 #dataviz It is a multiboxplot or multiviolinplot, depending on the data. Clustering is used distinguish groups of cases from each other that are plotted separately, but in one chart.
arxiv.org/abs/2510.09276 #dataviz It is a multiboxplot or multiviolinplot, depending on the data. Clustering is used distinguish groups of cases from each other that are plotted separately, but in one chart.
More than 30% of this century’s science Nobel prizewinners immigrated
www.nature.com/articles/d41... Largest beneficiary is, so far, USA. Absolute numbers a bit misleading because US higher ed & research is larger than any other. Everything else equal, which it isn't, abs numbers should be higher
www.nature.com/articles/d41... Largest beneficiary is, so far, USA. Absolute numbers a bit misleading because US higher ed & research is larger than any other. Everything else equal, which it isn't, abs numbers should be higher
October 13, 2025 at 4:20 PM
More than 30% of this century’s science Nobel prizewinners immigrated
www.nature.com/articles/d41... Largest beneficiary is, so far, USA. Absolute numbers a bit misleading because US higher ed & research is larger than any other. Everything else equal, which it isn't, abs numbers should be higher
www.nature.com/articles/d41... Largest beneficiary is, so far, USA. Absolute numbers a bit misleading because US higher ed & research is larger than any other. Everything else equal, which it isn't, abs numbers should be higher
Does a new generation of social scientists have to publish more to achieve less? (from 2019)
blogs.lse.ac.uk/impactofsoci... Intuition says "yes", and so says data for sociology.
blogs.lse.ac.uk/impactofsoci... Intuition says "yes", and so says data for sociology.
October 13, 2025 at 2:54 PM
Does a new generation of social scientists have to publish more to achieve less? (from 2019)
blogs.lse.ac.uk/impactofsoci... Intuition says "yes", and so says data for sociology.
blogs.lse.ac.uk/impactofsoci... Intuition says "yes", and so says data for sociology.
Who submits papers ahead of deadlines?
I am clearly the person on the right side.
errantscience.com/blog/2025/10...
I am clearly the person on the right side.
errantscience.com/blog/2025/10...
October 11, 2025 at 1:36 PM
Who submits papers ahead of deadlines?
I am clearly the person on the right side.
errantscience.com/blog/2025/10...
I am clearly the person on the right side.
errantscience.com/blog/2025/10...
Tidy simulation: Designing robust, reproducible, and scalable Monte Carlo simulations #StatsSky
arxiv.org/abs/2509.11741
It is not formally linked to the {tidyverse}, bit affinity is obvious.
The paper does a solid job in describing the simulation workflow, could be useful for intros to simulation
arxiv.org/abs/2509.11741
It is not formally linked to the {tidyverse}, bit affinity is obvious.
The paper does a solid job in describing the simulation workflow, could be useful for intros to simulation
September 23, 2025 at 9:25 AM
Tidy simulation: Designing robust, reproducible, and scalable Monte Carlo simulations #StatsSky
arxiv.org/abs/2509.11741
It is not formally linked to the {tidyverse}, bit affinity is obvious.
The paper does a solid job in describing the simulation workflow, could be useful for intros to simulation
arxiv.org/abs/2509.11741
It is not formally linked to the {tidyverse}, bit affinity is obvious.
The paper does a solid job in describing the simulation workflow, could be useful for intros to simulation
{ggdiagram} for creating diagrams with ggplot-syntax
wjschne.github.io/ggdiagram/ar... This looks like a ggplot-version of tikz, probably with less functionality, but easier to use when one is familiar with {ggplot2}. (and as long as theire is no AI for making tikz-graphics, which may exist?)
wjschne.github.io/ggdiagram/ar... This looks like a ggplot-version of tikz, probably with less functionality, but easier to use when one is familiar with {ggplot2}. (and as long as theire is no AI for making tikz-graphics, which may exist?)
August 21, 2025 at 4:03 PM
{ggdiagram} for creating diagrams with ggplot-syntax
wjschne.github.io/ggdiagram/ar... This looks like a ggplot-version of tikz, probably with less functionality, but easier to use when one is familiar with {ggplot2}. (and as long as theire is no AI for making tikz-graphics, which may exist?)
wjschne.github.io/ggdiagram/ar... This looks like a ggplot-version of tikz, probably with less functionality, but easier to use when one is familiar with {ggplot2}. (and as long as theire is no AI for making tikz-graphics, which may exist?)
Die Mündigkeit scheint sehr selektiv zu sein. Dasselbe hätte Frau Bär dann nach dem ersten Kanzlerwahlgang über Friedrich Merz sagen können.
July 16, 2025 at 6:10 AM
Die Mündigkeit scheint sehr selektiv zu sein. Dasselbe hätte Frau Bär dann nach dem ersten Kanzlerwahlgang über Friedrich Merz sagen können.
The perpetual research cycle: AI’s journey through data, papers, and knowledge
www.digital-science.com/blog/2025/03...
I am open to using LLMs where it makes sense (particularly coding). What is written in the blog post (see screenshot) is not sensible. How about one starts w/ a research question?
www.digital-science.com/blog/2025/03...
I am open to using LLMs where it makes sense (particularly coding). What is written in the blog post (see screenshot) is not sensible. How about one starts w/ a research question?
![Screenshot from linked blogpost saying:
[headline] The flywheel effect: How the cycle accelerates
The true magic happens when this extracted knowledge becomes the input for the next iteration. Each cycle follows this pattern:
1. Raw data is processed by AI to generate initial research outputs
2. Knowledge extraction tools mine these outputs for higher-order insights
3. These insights form a new, refined dataset
4. AI processes this refined dataset, generating more precise analyses
5. The cycle continues, with each rotation producing more valuable knowledge](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:qncmtszvmchqncrjl637b6cy/bafkreidppdwxr2o7kwyzysgeac6kbkesx2i46jj74ny2t5pp2k5bochujq@jpeg)
June 21, 2025 at 5:53 PM
The perpetual research cycle: AI’s journey through data, papers, and knowledge
www.digital-science.com/blog/2025/03...
I am open to using LLMs where it makes sense (particularly coding). What is written in the blog post (see screenshot) is not sensible. How about one starts w/ a research question?
www.digital-science.com/blog/2025/03...
I am open to using LLMs where it makes sense (particularly coding). What is written in the blog post (see screenshot) is not sensible. How about one starts w/ a research question?
From RFK Jr’s ‘Maha’ report. This takes the meaning of self-citation to a whole new level.
www.theguardian.com/us-news/2025...
www.theguardian.com/us-news/2025...
May 31, 2025 at 1:15 PM
From RFK Jr’s ‘Maha’ report. This takes the meaning of self-citation to a whole new level.
www.theguardian.com/us-news/2025...
www.theguardian.com/us-news/2025...
Differences are huge between journal quartiles (Scimago IF), which is probably not surprising. I guess this is mainly because journals with higher IFs have more demanding policies, on average, one has to comply with, and less because authors who publish there share material regardless of policy 2/
April 29, 2025 at 6:26 PM
Differences are huge between journal quartiles (Scimago IF), which is probably not surprising. I guess this is mainly because journals with higher IFs have more demanding policies, on average, one has to comply with, and less because authors who publish there share material regardless of policy 2/
that it is relatively easy to theorize and analyze a chronological order of events, and hard to turn this into a causal sequence. Figure 1 shows the stylized process, figure 4 is an empirical illustration. In my reading, it is not obvious why key episode 1 is causal for key episode 2. 2/
April 2, 2025 at 2:36 PM
that it is relatively easy to theorize and analyze a chronological order of events, and hard to turn this into a causal sequence. Figure 1 shows the stylized process, figure 4 is an empirical illustration. In my reading, it is not obvious why key episode 1 is causal for key episode 2. 2/
Fände es sehr interessant, zu wissen, ob die Sechserrunde nicht begreift, was für eine Botschaft so ein Bild sendet, oder ob es Ihnen egal ist. Oder ob sie genau die Botschaft vermitteln wollen, die aus dem Bild spricht.
February 25, 2025 at 1:47 PM
Fände es sehr interessant, zu wissen, ob die Sechserrunde nicht begreift, was für eine Botschaft so ein Bild sendet, oder ob es Ihnen egal ist. Oder ob sie genau die Botschaft vermitteln wollen, die aus dem Bild spricht.
relatedly, awards named after males are more likely awarded by societies and commitees in which male researchers dominate (path dependency). In either case, there is still a 3:1 male-female ratio for awards not named afer a person 2/
February 23, 2025 at 2:49 PM
relatedly, awards named after males are more likely awarded by societies and commitees in which male researchers dominate (path dependency). In either case, there is still a 3:1 male-female ratio for awards not named afer a person 2/
{gg1d} plots multiple variables one-dimensionally to search for trends/correlations #rstats
selkamand.github.io/gg1d/index.h... I find it hard to interpret the plots that look like sankey diagrams in disguise to me. I am not used to these plots, so it may be my lack of experience in reading them
selkamand.github.io/gg1d/index.h... I find it hard to interpret the plots that look like sankey diagrams in disguise to me. I am not used to these plots, so it may be my lack of experience in reading them
February 21, 2025 at 11:57 AM
{gg1d} plots multiple variables one-dimensionally to search for trends/correlations #rstats
selkamand.github.io/gg1d/index.h... I find it hard to interpret the plots that look like sankey diagrams in disguise to me. I am not used to these plots, so it may be my lack of experience in reading them
selkamand.github.io/gg1d/index.h... I find it hard to interpret the plots that look like sankey diagrams in disguise to me. I am not used to these plots, so it may be my lack of experience in reading them
Um Kostensteigerungen aufzufangen kürzt @dfg.de bei mehreren Förderlinien. Insgesamt halten sich die Einschnitte in Grenzen. Dennoch frage ich mich, ob die Exzellenzstrategie davon auch betroffen ist, die zum Beispiel die Anzahl der Cluster in dieser Runde erhöht (vor einiger Zeit beschlossen) 1/
February 18, 2025 at 3:56 PM
Um Kostensteigerungen aufzufangen kürzt @dfg.de bei mehreren Förderlinien. Insgesamt halten sich die Einschnitte in Grenzen. Dennoch frage ich mich, ob die Exzellenzstrategie davon auch betroffen ist, die zum Beispiel die Anzahl der Cluster in dieser Runde erhöht (vor einiger Zeit beschlossen) 1/
How Are p-values Distributed Under The Null? - David's blog
davidlindelof.com/how-are-p-va...
#StatSky Most of the time, it's uniform
davidlindelof.com/how-are-p-va...
#StatSky Most of the time, it's uniform
February 16, 2025 at 2:22 PM
How Are p-values Distributed Under The Null? - David's blog
davidlindelof.com/how-are-p-va...
#StatSky Most of the time, it's uniform
davidlindelof.com/how-are-p-va...
#StatSky Most of the time, it's uniform
I find it difficult to generalize about a discipline based on one journal policy. This survey (journals.sagepub.com/doi/10.1177/...) indicates that at least half of the journals are open to comments. The n of the survey is small and the percentage should be higher, IMO, but it also not bad.
February 11, 2025 at 9:32 AM
I find it difficult to generalize about a discipline based on one journal policy. This survey (journals.sagepub.com/doi/10.1177/...) indicates that at least half of the journals are open to comments. The n of the survey is small and the percentage should be higher, IMO, but it also not bad.
Report uncertainty information to improve trust in science www.nature.com/articles/s41... I fully agree with the idea of reporting uncertainty. This has value in itself and may improve trust in science, which is not subject of the article (the title is misleading in this regard) 1/
January 29, 2025 at 12:03 PM
Report uncertainty information to improve trust in science www.nature.com/articles/s41... I fully agree with the idea of reporting uncertainty. This has value in itself and may improve trust in science, which is not subject of the article (the title is misleading in this regard) 1/
Statistical Guidance to Authors at Top-Ranked Journals across Scientific Disciplines #StatSky #MetaScience
www.tandfonline.com/doi/full/10.... The coverage of guidance is diverse across disciplines and also across specific statistical issues 👇
The variation is interesting because it is large 1/
www.tandfonline.com/doi/full/10.... The coverage of guidance is diverse across disciplines and also across specific statistical issues 👇
The variation is interesting because it is large 1/
January 22, 2025 at 6:41 PM
Statistical Guidance to Authors at Top-Ranked Journals across Scientific Disciplines #StatSky #MetaScience
www.tandfonline.com/doi/full/10.... The coverage of guidance is diverse across disciplines and also across specific statistical issues 👇
The variation is interesting because it is large 1/
www.tandfonline.com/doi/full/10.... The coverage of guidance is diverse across disciplines and also across specific statistical issues 👇
The variation is interesting because it is large 1/
I understand why NHB emphasizes statistical details. Personally, I am not sure the sample size is important enough to be featured in the abstract. A confidence interval, or some interval would be preferable to me.
I like this policy on claims about implications. 👇 2/
I like this policy on claims about implications. 👇 2/
January 17, 2025 at 11:05 AM
I understand why NHB emphasizes statistical details. Personally, I am not sure the sample size is important enough to be featured in the abstract. A confidence interval, or some interval would be preferable to me.
I like this policy on claims about implications. 👇 2/
I like this policy on claims about implications. 👇 2/
Introducing the {messy} package | Nicola Rennie nrennie.rbind.io/blog/introdu... #rstats It looks very useful for creating, you know, messy datasets that one can use in classes on data tidying
January 9, 2025 at 7:15 PM
Introducing the {messy} package | Nicola Rennie nrennie.rbind.io/blog/introdu... #rstats It looks very useful for creating, you know, messy datasets that one can use in classes on data tidying
{gganonymize} automates anonymization of {ggplot2} plots github.com/EmilHvitfeld... I am not sure this is meaningful bc full anonymization impedes interpretability. If necessary, this is certainly the package of choice #rstats
January 2, 2025 at 7:16 PM
{gganonymize} automates anonymization of {ggplot2} plots github.com/EmilHvitfeld... I am not sure this is meaningful bc full anonymization impedes interpretability. If necessary, this is certainly the package of choice #rstats