- [2023]. Bayesian spatial modelling of localised SARS-CoV-2 transmission through mobility networks across England. {\em PLoS Computational Biology} 19, e1011580.
(Thomas Ward, Mitzi Morris, Andrew Gelman, Bob Carpenter, William Ferguson, Christopher Overton, and Martyn Fylesn)- [2023] Generically partisan: Polarization in political communication. {\em Proceedings of the National Academy of Sciences} 120, e2309361120.
(Gustavo Novoa, Margaret Echelbarger, Andrew Gelman, and Susan Gelman)
Supplementary appendix.- [2023] Simulation-based calibration checking for Bayesian computation: The choice of test quantities shapes sensitivity. {\em Bayesian Analysis}.
(Martin Modrák, Angie H. Moon, Shinyoung Kim, Paul Bürkner, Niko Huurre, Kateřina Faltejsková, Andrew Gelman, and Aki Vehtari)- [2023] Causal quartets: Different ways to attain the same average treatment effect. {\em American Statistician}.
(Andrew Gelman, Jessica Hullman, and Lauren Kennedy)- [2023] In pursuit of campus-wide data literacy: A guide to developing a statistics course for students in non-quantitative fields. {\em Journal of Statistics and Data Science Education}.
(Alexis Lerner and Andrew Gelman)- [2023] A new look at p-values for randomized clinical trials. {\em NEJM Evidence}.
(Erik van Zwet, Andrew Gelman, Sander Greenland, Guido Imbens, Simon Schwab, and Steven N. Goodman)- [2023] Past, present, and future of software for Bayesian inference. {\em Statistical Science}.
(Erik Štrumbelj, Alexandre Bouchard-Côté, Jukka Corander, Andrew Gelman, Håvard Rue, Lawrence Murray, Henri Pesonen, Martyn Plummer, and Aki Vehtari)- [2023] Challenges in adjusting a survey that overrepresents people interested in politics. {\em Harvard Data Science Review} {\bf 5} (3).
(Andrew Gelman and Gustavo Novoa)- [2023] Using leave-one-out cross-validation (LOO) in a multilevel regression and poststratification (MRP) workflow: A cautionary tale. {\em Statistics in Medicine}.
(Swen Kuh, Lauren Kennedy, Qixuan Chen, and Andrew Gelman)- [2023] What is a standard error? {\em Journal of Econometrics} {\bf 237}, 105516.
(Andrew Gelman)- [2023] Who wants school vouchers in America? A comprehensive study using multilevel regression and poststratification. {\em Social Sciences} {\bf 12} (8), 430.
(Yu-Sung Su and Andrew Gelman)- [2023] A chain as strong as its strongest link? Understanding the causes and consequences of biases arising from selective analysis and reporting of research results. {\em Journal of Research on Educational Effectiveness}.
(Andrew Gelman)- [2023] Before data analysis: Additional recommendations for designing experiments to learn about the world. {\em Journal of Consumer Psychology}.
(Andrew Gelman)- [2023] Toward a taxonomy of trust for probabilistic machine learning. {\em Science Advances} {\bf 9}, eabn3999.
(Tamara Broderick, Andrew Gelman, Rachael Meager, Anna L. Smith, and Tian Zheng)- [2023] Federated learning as variational inference: A scalable expectation propagation approach. {\em International Conference on Learning Representations (ICLR)}.
(Han Guo, Philip Greengard, Hongyi Wang, Andrew Gelman, Yoon Kim, and Eric P. Xing)- [2023] I love this paper but it’s barely been noticed. Part of a collaborative article, “What are your most underappreciated works?” {\em Econ Journal Watch} {\bf 20}, 466.
(Andrew Gelman)- [2023] From visualization to sensification. {\em Amstat News} 547, 18–19.
(Andrew Gelman and S. Gwynn Sturdevant)- [2023] Fast methods for posterior inference of two-group normal-normal models. {\em Bayesian Analysis}.
(Philip Greengard, Jeremy Hoskins, Charles C. Margossian, Jonah Gabry, Andrew Gelman, and Aki Vehtari)- [2023] “Two truths and a lie” as a class-participation activity. {\em American Statistician} {\bf 77}, 97–101.
(Andrew Gelman)
- Regression, poststratification, and small-area estimation with sampling weights.
(Andrew Gelman)- Understanding posterior recalibration for a simple example.
(Andrew Gelman, Julie Gershunskaya, Terrance Savitsky, and Ben Goodrich)- Bayesian workflow for time-varying transmission in stratified compartmental infectious disease transmission models.
(Judith A. Bouman, Anthony Hauser, Simon L. Grimm, Martin Wohlfender, Samir Bhatt, Elizaveta Semenova, Andrew Gelman, Christian L. Althaus, and Julien Riou)- Artificial intelligence and aesthetic judgment.
(Jessica Hullman, Ari Holtzman, and Andrew Gelman)- The ladder of abstraction in statistical graphics.
(Andrew Gelman)- BISG: When inferring race or ethnicity, does it matter that people often live near their relatives?
(Philip Greengard and Andrew Gelman)
Enjoy.
Nice. Just curious – which paper from 2023 do you like best?
I haven’t read many of these, but I really liked the Causal Quartet paper.
Jd:
I love all my children equally!
But, yeah, the causal quartets paper was particularly satisfying because I just had the idea one day, close to fully formed, and was able to write it up right away—and then had the luck to have two excellent collaborators who made it even better. While revising the paper for publication, we needed to remove some material on specifying average effects, and we’re spinning that off into another short paper. My only regret is that American Statistician doesn’t get the readership it used to: back in the day when we received journals in the mail, the American Statistician was the most readable journal, and people would open it up and take a look right away. Nowadays we need to spread the word through social media nd word-of-mouth.
I also love the short paper, “Before data analysis,” which some colleagues and I are hoping to use as a starting point for something more systematic on designing studies.
And I think the “Two truths a lie” paper is a real gem. It’s one of the 52 activities in my forthcoming Active Statistics book with Aki. The review process was super-helpful and allowed me to improve the activity in various ways. We could easily have spun off 100 papers from that new book, but that would’ve taken a lot of work so I just did that one.
Of my recent unpublished papers, my favorite is the very first one listed above, on incorporating sampling weights into Bayesian regression. It provides the solution to a problem that’s been bugging me for about two decades. I guess I can’t say for sure that I love the paper, because the method might not work out in real applications, but I have some hopes!
I could go on . . .
Every paper has its own story. In my spare time I’m preparing a long article with a paragraph describing where each of my papers came from. Even readers who are not interested in thirty-year-old statistics papers might find some interest in the processes of research and collaboration.
Cool. I just read both the ‘Before data analysis’ and ‘Two truths and a lie’ paper.
I’ve tried to incorporate the ideas in the ‘Before data analysis’ paper into my own workflow. In my experience, the seemingly obvious advice of, “First, set up your design and data collection to measure what you want to learn about” is often ignored! I’ve come to think that maybe every measure is really a proxy for whatever it is one really wants to know. Even in a lab setting, it isn’t as straightforward as one might think. It is actually pretty difficult to collect data that is actually measuring what one wants to know. Of course, I think researchers are also tempted to squeeze every drop out of the data that they already have, and maybe sometimes the papers appear disconnected from the study because it was never designed for the paper that was written in the first place!
This is probably a terrible idea, but maybe you could adapt your ‘Two truths and a lie’ activity to a (not remote) conference talk (of reasonable size)? You could put up 2 truths and a lie from yourself, Aki, Bob, Lauren, Jessica, etc on the screen and have the audience vote thumbs up or down. Your rough guess at number thumbs up or down would now make up both the truth/lie (majority rule) and the certainty metric (rough proportion of thumbs up or down). You would fill out the Google doc on screen. If nothing else, it would certainly introduce the concept of measurement error to data collection.
hi Andrew, did you know that you’re an author on this fabulous paper from Tom Ward et al?
Bayesian spatial modelling of localised SARS-CoV-2 transmission through mobility networks across England
T. Ward, M. Morris, A. Gelman, B. Carpenter, W. Ferguson, C. Overton, and M. Fyles
PLOS Computational Biology 19 e1011580– (2023)
https://doi.org/10.1371/journal.pcbi.1011580
I added it to the list; thanks.
You should add a section called “Works Performed” and list Recursion, ACM FAccT 2023!
Jessica:
I do list Recursion on the books page. And let’s see if we can get it performed again in 2024!
Hi Andrew. This is a great article! [2023] A new look at p-values for randomized clinical trials. {\em NEJM Evidence}. However, notice that the lines in Figure 1 are reversed.
Thanks, well spotted! I notified the journal and hopefully they’ll correct it soon.