How is statistics used to predict elections? Andrew and Rafa discuss the U.S. 2020 Election and the role of the electoral college, polls, mail-in ballots and voter data in forecasting results and post-election outcomes.Andrew Gelman, PhD is a professor of statistics and political science at Columbia University. He is one of the go-to statisticians for the New York Times and author of perhaps the most popular statistics blog: Statistical Modeling, Causal Inference, and Social Science. He has received the Outstanding Statistical Application award three times from the American Statistical Association, the award for best article published in the American Political Science Review, and the Council of Presidents of Statistical Societies award for outstanding contributions by a person under the age of 40. Books he has authored and co-authored include Bayesian Data Analysis, Teaching Statistics: A Bag of Tricks, Data Analysis Using Regression and Multilevel/Hierarchical Models, Red State, Blue State, Rich State, Poor State: Why Americans Vote the Way They Do, A Quantitative Tour of the Social Sciences, and Regression and Other Stories.Our Data Science Zoominars feature interactive conversation with data science experts and a Q+A session moderated by Rafael A. Irizarry, PhD, Chair, Department of Data Science at Dana-Farber Cancer Institute.
Statistics and American Politics with Andrew Gelman of Columbia University
The Great Battlefield
Professor of Statistics and Politics, Andrew Gelman joins The Great Battlefield podcast to talk about his career including a political science paper he co-wrote called "Why are American Presidential Campaign Polls so Variable when Votes are so Predictable".
Andrew Gelman & Megan Higgs | Statistics’ Role in Science and Pseudoscience
Data & Science with Glen Wright Colopy
Andrew Gelman & Megan Higgs | Statistics' Role in Science and Pseudoscience #datascience #statistics #science #pseudoscience Our science vs pseudoscience discussion continues with Andrew Gelman (Columbia) and Megan Higgs (Critical Inference LLC). Andrew and Megan describe two critical roles that statistics plays in science.... but also how statistics can add the air of scientific rigor to bad research or help statisticians fool themselves. From there the conversation goes on in a way that only a conversation with Andrew and Megan can! A very fun episode. 0:00 - Two roles of statistics in science4:50 - Many models were intended for designed experiments10:30 - The biggest scientific error of the past 20 years15:00 - Feedback loop of over-confidence / Armstrong Principle21:00 - Science is personal25:00 - The value of different approaches / Don Rubin Story34:40 - Statistics is the science of defaults / engineering new methods45:00 - The value of writing what you did52:27 - Math vs science backgrounds + a thought experiment1:01:20 - Fooling ourselves
Rationally Speaking #147 - Andrew Gelman on "Why do Americans vote the way they do?"
Sped up Rationally Speaking
There are two contradictory stories about politics and class: On the one hand, that the Republicans are the party of the fat cat businessmen and the Democrats are the party of the people. And on the other hand, that the Republicans are the party of the salt-of-the-earth Joe Sixpacks, while the Democrats are latte-sipping elites. In this episode, professor of statistics and political science Andrew Gelman shines some clarifying light on the intersection between politics and class in America, explaining what the numbers really show. He and Julia also cover the question, "Is it rational to vote?" Sped up the speakers by ['1.17', '1.0']
#27 Modeling the US Presidential Elections, with Andrew Gelman & Merlin Heidemanns
Learning Bayesian Statistics
In a few days, a consequential election will take place, as citizens of the United States will go to the polls and elect their president — in fact they already started voting. You probably know a few forecasting models that try to predict what will happen on Election Day — who will get elected, by how much and with which coalition of States?But how do these statistical models work? How do you account for the different sources of uncertainty, be it polling errors, unexpected turnout or media events? How do you model covariation between States? How do you even communicate the model’s results and afterwards assess its performance? To talk about all this, I had the pleasure to talk to Andrew Gelman and Merlin Heidemanns.Andrew was already on episode 20, to talk about his recent book with Jennifer Hill and Aki Vehtari, “Regression and Other Stories”. He’s a professor of statistics and political science at Columbia University and works on a lot of topics, including: why campaign polls are so variable while elections are so predictable, the statistical challenges of estimating small effects, and methods for surveys and experimental design.Merlin is a PhD student in Political Science at Columbia University, and he specializes in political methodology. Prior to his PhD, he did a Bachelor's in Political Science at the Freie Universität Berlin.I hope you’ll enjoy this episode where we dove into the Bayesian model they helped develop for The Economist, and talked more generally about how to forecast elections with statistical methods, and even about the incentives the forecasting industry has as a whole.Thank you to my Patrons for making this episode possible! Visit https://www.patreon.com/learnbayesstats to unlock exclusive Bayesian swag ;)Our theme music is « Good Bayesian », by Baba Brinkman (feat MC Lars and Mega Ran). Check out his awesome work at https://bababrinkman.com/ !Links from the show:Andrew's website: http://www.stat.columbia.edu/~gelman/Andrew's blog: https://statmodeling.stat.columbia.edu/Andrew on Twitter: https://twitter.com/statmodelingMerlin's website: https://merlinheidemanns.github.io/website/Merlin on Twitter: https://twitter.com/MHeidemannsThe Economist POTUS forecast: https://projects.economist.com/us-2020-forecast/presidentHow The Economist presidential forecast works: https://projects.economist.com/us-2020-forecast/president/how-this-worksGitHub repo of the Economist model: https://github.com/TheEconomist/us-potus-modelInformation, incentives, and goals in election forecasts (Gelman, Hullman & Wlezien): http://www.stat.columbia.edu/~gelman/research/unpublished/forecast_incentives3.pdfHow to think about extremely unlikely events: https://bit.ly/3ejZYyZPostal voting could put America’s Democrats at a disadvantage: https://econ.st/3mCxR0PThank you to my Patrons for making this episode possible!Yusuke Saito, Avi Bryant, Ero Carrera, Brian Huey, Giuliano Cruz, Tim Gasser, James Wade, Tradd Salvo, Adam Bartonicek, William Benton, Alan O'Donnell, Mark Ormsby, Demetri Pananos, James Ahloy, Jon Berezowski, Robin Taylor, Thomas Wiecki, Chad Scherrer, Vincent Arel-Bundock, Nathaniel Neitzke, Zwelithini Tunyiswa, Elea McDonnell Feit, Bertrand Wilden, James Thompson, Stephen Oates, Gian Luca Di Tanna, Jack Wells, Matthew Maldonado, Ian Costley, Ally Salim, Larry Gill, Joshua Duncan, Ian Moran and Paul Oreto.This podcast uses the following third-party services for analysis: Podcorn - https://podcorn.com/privacy
Statistics is the Least Important Part of Data Science | Andrew Gelman, PhD
The Artists of Data Science
Andrew is an American statistician, professor of statistics and political science, and director of the Applied Statistics Center at Columbia University. He frequently writes about Bayesian statistics, displaying data, and interesting trends in social science. He’s also well known for writing posts sharing his thoughts on best statistical practices in the sciences, with a frequent emphasis on what he sees as the absurd and unscientific. FIND ANDREW ONLINE Website: https://statmodeling.stat.columbia.edu/ Twitter: https://twitter.com/StatModeling QUOTES [00:04:16] "We've already passed peak statistics..." [00:05:13] "One thing that we sometimes like to say is that big data need big model because big data are available data. They're not designed experiments, they're not random samples. Often big data means these are measurements. " [00:22:05] "If you design an experiment, you want to know what you're going to do later. So most obviously, you want your sample size to be large enough so that given the effect size that you expect to see, you'll get a strong enough signal that you can make a strong statement." [00:31:00] "The alternative to good philosophy is not no philosophy, it's bad philosophy. " SHOW NOTES [00:03:12] How Dr. Gelman got interested in statistics [00:04:09] How much more hyped has statistical and machine learning become since you first broke into the field? [00:04:44] Where do you see the field of statistical machine learning headed in the next two to five years? [00:06:12] What do you think the biggest positive impact machine learning will have in society in the next two to five years? [00:07:24] What do you think would be some of our biggest concerns in the future? [00:09:07] The thee parts of Bayesian inference [00:12:05] What's the main difference between the frequentist and the Bayesian? [00:13:02] What is a workflow? [00:16:21] Iteratively building models [00:17:50] How does the Bayesian workflow differ from the frequent workflow? [00:18:32] Why is it that what makes this statistical method effective is not what it does with the data, but what data it uses? [00:20:48] Why do Bayesians then tend to be a little bit more skeptical in their thought processes? [00:21:47] Your method of evaluation can be inspired by the model or the model can be inspired by your method of evaluation [00:24:38] What is the usual story when it comes to statistics? And why don't you like it? [00:30:16] Why should statisticians and data scientist care about philosophy? [00:35:04] How can we solve all of our statistics problems using P values? [00:36:14] Is there a difference in interpretations for P-Values between Bayesian and frequentist. [00:36:54] Do you feel like the P value is a difficult concept for a lot of people to understand? And if so, why do you think it's a bit challenging? [00:38:22] Why the least important part of data science is statistics. [00:40:09] Why is it that Americans vote the way they do? [00:42:40] What's the one thing you want people to learn from your story? [00:44:48] The lightning roundSpecial Guest: Andrew Gelman, PhD.
#20 Regression and Other Stories, with Andrew Gelman, Jennifer Hill & Aki Vehtari
Learning Bayesian Statistics
Once upon a time, there was an enchanted book filled with hundreds of little plots, applied examples and linear regressions — the prettiest creature that was ever seen. Its authors were excessively fond of it, and its readers loved it even more. This magical book had a nice blue cover made for it, and everybody aptly called it « Regression and other Stories »!As every good fairy tale, this one had its share of villains — the traps where statistical methods fall and fail you; the terrible confounders, lurking in the dark; the ill-measured data that haunt your inferences! But once you defeat these monsters, you’ll be able to think about, build and interpret regression models.This episode will be filled with stories — stories about linear regressions! Here to narrate these marvelous statistical adventures are Andrew Gelman, Jennifer Hill and Aki Vehtari — the authors of the brand new Regression and other Stories.Andrew is a professor of statistics and political science at Columbia University. Jennifer is a professor of applied statistics at NYU. She develops methods to answer causal questions related to policy research and scientific development. Aki is an associate professor in computational probabilistic modeling at Aalto University, Finland.In this episode, they tell us why they wrote this book, who it is for and they also give us their 10 tips to improve your regression modeling! We also talked about the limits of regression and about going to Mars…Other good news: until October 31st 2020, you can go to http://www.cambridge.org/wm-ecommerce-web/academic/landingPage/GoodBayesian2020 and buy the book with a 20% discount by entering the promo code “GoodBayesian2020” upon checkout!That way, you’ll make up your own stories before going to sleep and dream of a world where we can easily generalize from sample to population, and where multilevel regression with poststratification is a bliss…Our theme music is « Good Bayesian », by Baba Brinkman (feat MC Lars and Mega Ran). Check out his awesome work at https://bababrinkman.com/ !Links from the show:Regression and Other Stories on Cambridge Press website: http://www.cambridge.org/wm-ecommerce-web/academic/landingPage/GoodBayesian2020Amazon page (because of VAT laws, in some regions ordering from Amazon can be cheaper than from the editor directly, even with the discount): https://www.amazon.com/Regression-Stories-Analytical-Methods-Research/dp/110702398XCode, data and examples for the book: https://avehtari.github.io/ROS-Examples/Port of the book in Python and Bambi: https://github.com/bambinos/Bambi_resources/tree/master/ROSAndrew's home page: http://www.stat.columbia.edu/~gelman/Andrew's blog: https://statmodeling.stat.columbia.edu/Andrew on Twitter: https://twitter.com/statmodelingJennifer's home page: https://steinhardt.nyu.edu/people/jennifer-hillAki's teaching material: https://avehtari.github.io/Aki's home page: https://users.aalto.fi/~ave/Aki on Twitter: https://twitter.com/avehtariThank you to my Patrons for making this episode possible!Yusuke Saito, Avi Bryant, Ero Carrera, Brian Huey, Giuliano Cruz, Tim Gasser, James Wade, Tradd Salvo, Adam Bartonicek, William Benton, Alan O'Donnell, Mark Ormsby, Demetri Pananos, James Ahloy, Jon Berezowski, Robin Taylor, Thomas Wiecki, Chad Scherrer, Vincent Arel-Bundock, Nathaniel Neitzke, Zwelithini Tunyiswa, Elea McDonnell Feit, Bertrand Wilden, James Thompson, Stephen Oates, Gian Luca Di Tanna, Jack Wells, Matthew Maldonado, Ian Costley, Ally Salim, Larry Gill, Joshua Duncan, Ian Moran and Paul Oreto.This podcast uses the following third-party services for analysis: Podcorn - https://podcorn.com/privacy
Ep 12: Andrew Gelman on Data, Modeling, and Uncertainty Amidst the Forking Paths
The Filter Podcast with Matt Asher
A conversation with Andrew Gelman, professor of statistics and political science and director of the Applied Statistics Center at Columbia University. Andrew is the author of a number of books on topics such as Bayesian Data Analysis, how stats should be taught, and voting patterns in politics. Our conversation topics included:Forking paths in data analysisTo what extent to prior beliefs determine study outcomesData integrity in the era of COVIDUnreliable friends and modeling uncertaintyRelated links:Andrew Gelman at Columbia UniversityAndrew Gelman - The Statistical Crisis in Science (Dec 2014)CDC - Excess Deaths Associated with COVID-19 (Jul 2020)E.T. Jaynes - Probability Theory (Chapter 5 for quote)David Shor - Interview with New York Magazine (Jul 2020)Katie Herzog - Discussion on Cancel Culture with The Filter (Jul 2020)Nassim Taleb - Uncertainty (Mar 2020)Matt Asher - Unreliable Friend Distribution (May 2013)
Jeremiah sits down with Dr. Andrew Gelman to discuss the replication crisis in science - how it began, what we know and how we should think about science moving forward. Patreon subscribers get access to bonus episodes and sticker of the month club. If you like what we do (and want stickers each month!) please consider supporting us at patreon.com/neoliberalproject.
Andrew Gelman on Social Science, Small Samples, and the Garden of the Forking Paths
Statistician, blogger, and author Andrew Gelman of Columbia University talks with EconTalk host Russ Roberts about the challenges facing psychologists and economists when using small samples. On the surface, finding statistically significant results in a small sample would seem to be extremely impressive and would make one even more confident that a larger sample would find even stronger evidence. Yet, larger samples often fail to lead to replication. Gelman discusses how this phenomenon is rooted in the incentives built into human nature and the publication process. The conversation closes with a general discussion of the nature of empirical work in the social sciences.