Skip to primary content

Statistical Modeling, Causal Inference, and Social Science

Main menu

  • Home
  • Authors
  • Blogs We Read
  • Books
  • Sponsors

Post navigation

Bloggitude: who gets upset by what? Econometrics reaches The Economist

Problems with poll adjustments that aggressively screen for likely voters

Posted on August 15, 2009 9:28 PM by Andrew

See Nate’s thoughts from today and Yair’s from last year.

This entry was posted in Political Science by Andrew. Bookmark the permalink.
  • Art
  • Artificial Intelligence
  • Bayesian Statistics
  • Causal Inference
  • Decision Analysis
  • Economics
  • Jobs
  • Literature
  • Miscellaneous Science
  • Miscellaneous Statistics
  • Multilevel Modeling
  • Obituaries
  • Papers
  • Political Science
  • Public Health
  • Sociology
  • Sports
  • Stan
  • Statistical Computing
  • Statistical Graphics
  • Teaching
  • Zombies
  1. Roy on How generic language shapes the development of social thoughtJuly 24, 2026 10:37 AM

    Andrew calls it the "sister blog" because it literally is the blog of his sister. He links to her work…

  2. gec on How generic language shapes the development of social thoughtJuly 24, 2026 10:25 AM

    Thanks for sharing, this is a great overview of this topic! In particular, I see this as another example of…

  3. Kaiser on Pretty maps of the NY mayoral election vote. (The meta-point here is that people have a (false) intuition that any complicated piece of information can be conveyed in a single plot.)July 24, 2026 10:21 AM

    NIce post bringing up an important point of consideration. The real issue is editing skill. While I agree with the…

  4. John Richters on How generic language shapes the development of social thoughtJuly 24, 2026 10:17 AM

    Andrew: I don’t understand the reason for this post. Am I missing something? John

  5. Olaf Zimmermann on Pretty maps of the NY mayoral election vote. (The meta-point here is that people have a (false) intuition that any complicated piece of information can be conveyed in a single plot.)July 24, 2026 6:40 AM

    Quoth Kieran Healy: "ArcGIS is the 800lb gorilla in this world and QGIS the GIMP to its Photoshop." Clearly he…

  6. Olaf Zimmermann on Was this USDA survey really “redundant, costly, politicized, and extraneous”?July 23, 2026 5:31 PM

    Amos Oz (2002/2018) Dear Zealots Amos Tversky (1977) Features of Similarity George Orwell (1946) Politics and the English Language Never…

  7. Dale Lehman on Forking paths in LLMs for data analysisJuly 23, 2026 1:38 PM

    Daniel My last comment since we seem to be circling around now (but I am tempted to make this the…

  8. Daniel Lakeland on Forking paths in LLMs for data analysisJuly 23, 2026 1:11 PM

    Dale, you are mistaking a technological solution to a technological problem (which is our proposal), for a technological solution to…

  9. Dale Lehman on Forking paths in LLMs for data analysisJuly 23, 2026 12:53 PM

    Daniel Your analogies are good - but too imperfect. Circular saw blades are dangerous but fairly simply physical mechanisms. As…

  10. elin on It’s all about the Super Pacs: How the New York Times completely misreported campaign contributions in the Maine Senate raceJuly 23, 2026 12:29 PM

    I don't know about the Times, but these days it is often AI that proposes the headlines (usually a few…

  11. elin on It’s all about the Super Pacs: How the New York Times completely misreported campaign contributions in the Maine Senate raceJuly 23, 2026 12:28 PM

    Exactly.

  12. Elin on It’s all about the Super Pacs: How the New York Times completely misreported campaign contributions in the Maine Senate raceJuly 23, 2026 12:26 PM

    If that is the main point of the article, the image should support that point. They could add a connected…

  13. Elin on It’s all about the Super Pacs: How the New York Times completely misreported campaign contributions in the Maine Senate raceJuly 23, 2026 12:24 PM

    The visuals matter, on their own, you can never assume people read. That is data visualization 101. It's why you…

  14. Daniel Lakeland on Forking paths in LLMs for data analysisJuly 23, 2026 12:13 PM

    Various friends, I think youre misunderstanding the technological suggestions here. Circular saws are built with blade guards, firearms have safety…

  15. Elin on Pretty maps of the NY mayoral election vote. (The meta-point here is that people have a (false) intuition that any complicated piece of information can be conveyed in a single plot.)July 23, 2026 12:08 PM

    I like that mapping a lot, not least because it highlights that where I live there are two precincts and…

  16. John N-G on He “washed his hands in a can of tetraethyl lead at a press conference, claiming he was ‘not taking any chance whatever’. He knew this to be a lie, having already succumbed to a bout of lead poisoning.”July 23, 2026 12:07 PM

    How about Phil Anderer? He ran around a lot.

  17. AAAnonymous on He “washed his hands in a can of tetraethyl lead at a press conference, claiming he was ‘not taking any chance whatever’. He knew this to be a lie, having already succumbed to a bout of lead poisoning.”July 23, 2026 11:51 AM

    Quote from above: "But we’re definitely not supposed to take away the theory that *some* people are evil." This reminded…

  18. Dale Lehman on Forking paths in LLMs for data analysisJuly 23, 2026 11:19 AM

    Joshua Let's not mistake the persistence of regulation for a solution to a problem. I'm not against regulation, though some…

  19. Andrew on Pretty maps of the NY mayoral election vote. (The meta-point here is that people have a (false) intuition that any complicated piece of information can be conveyed in a single plot.)July 23, 2026 11:01 AM

    Gec: Yup. In Appendix B of our book Regression and Other Stories, we have 10 quick tips to improve your…

  20. John Kane on Was this USDA survey really “redundant, costly, politicized, and extraneous”?July 23, 2026 10:51 AM

    I was thinking of exactly that. Was it not Donald J Trump, who in his first term as US president…

  21. Andrew on Pretty maps of the NY mayoral election vote. (The meta-point here is that people have a (false) intuition that any complicated piece of information can be conveyed in a single plot.)July 23, 2026 10:32 AM

    Link fixed; thanks.

  22. Joshua on Forking paths in LLMs for data analysisJuly 23, 2026 10:19 AM

    Dale - You say: "Like most attempts to protect people from their own lack of knowledge or irresponsible behavior, I…

  23. shira on Survey Statistics: poststratification without population level informationJuly 23, 2026 9:52 AM

    Great question, Olaf ! Andrew is right, turnout history is available through voterfiles and we used this as a proxy…

  24. Jonathan (another one) on He “washed his hands in a can of tetraethyl lead at a press conference, claiming he was ‘not taking any chance whatever’. He knew this to be a lie, having already succumbed to a bout of lead poisoning.”July 23, 2026 9:44 AM

    Isn't the received wisdom of WWII (see, for example, Jack El-Hai's book on Douglas Kelley that was the basis for…

  25. gec on Pretty maps of the NY mayoral election vote. (The meta-point here is that people have a (false) intuition that any complicated piece of information can be conveyed in a single plot.)July 23, 2026 9:38 AM

    > in general, one single plot (or even two plots) won’t do it all I think this speaks to a…

  26. Joshua on Pretty maps of the NY mayoral election vote. (The meta-point here is that people have a (false) intuition that any complicated piece of information can be conveyed in a single plot.)July 23, 2026 9:36 AM

    Andrew - FYI - your hyperlink to the previous post links back to this post.

  27. Raphael K on Pretty maps of the NY mayoral election vote. (The meta-point here is that people have a (false) intuition that any complicated piece of information can be conveyed in a single plot.)July 23, 2026 9:29 AM

    'The top map uses a bidirectional color scheme of the sort that I like, except that usually I’d see the…

  28. AAAnonymous on He “washed his hands in a can of tetraethyl lead at a press conference, claiming he was ‘not taking any chance whatever’. He knew this to be a lie, having already succumbed to a bout of lead poisoning.”July 23, 2026 8:50 AM

    Quote from above: "The book “People of the Lie” is a good read about those whose self is wrapped up…

  29. Joshua on Forking paths in LLMs for data analysisJuly 23, 2026 8:27 AM

    Fellas - Independent of what the ideal goal is here or how to achieve it, enforcing these kinds of architectural…

  30. Dale Lehman on Forking paths in LLMs for data analysisJuly 23, 2026 8:25 AM

    Tris, David I am in agreement with you about the dangers, but not your recommended solutions. Like most attempts to…

  31. Ethan Brooks on “As our daily lives involve ever more sophisticated computers, we will find that ascribing little thoughts to machines will be increasingly useful in understanding how to get the most good out of them” but “we must be careful not to ascribe properties to a machine that the particular machine doesn’t have”July 23, 2026 7:02 AM

    This is a thoughtful perspective on how we describe AI and the importance of using precise language when discussing technology.…

  32. Sean on He “washed his hands in a can of tetraethyl lead at a press conference, claiming he was ‘not taking any chance whatever’. He knew this to be a lie, having already succumbed to a bout of lead poisoning.”July 23, 2026 6:29 AM

    This guy was clearly wrapped up in how great his invention was and the money he was making. I'd guess…

  33. Andrew on Survey Statistics: poststratification without population level informationJuly 23, 2026 2:28 AM

    Olaf: There are different ways to do this adjustment. One approach is to poststratify on variables associated with interest in…

  34. Anoneuoid on He “washed his hands in a can of tetraethyl lead at a press conference, claiming he was ‘not taking any chance whatever’. He knew this to be a lie, having already succumbed to a bout of lead poisoning.”July 23, 2026 1:50 AM

    I’m old enough to remember when Cutter Labs made a bad batch of the Salk vaccine Please share anything more…

  35. Tris Simondsen on Forking paths in LLMs for data analysisJuly 22, 2026 11:45 PM

    Dale, Daniel rightly split this into engineering versus societal impacts, but why is an AI hallucination a different kind of…

  36. David, a Bostonian in Tokyo on Forking paths in LLMs for data analysisJuly 22, 2026 10:17 PM

    Dale asks: "Would you disallow AI search but not non-AI search? " The problem is that AI search returns a…

  37. OccDocJ on He “washed his hands in a can of tetraethyl lead at a press conference, claiming he was ‘not taking any chance whatever’. He knew this to be a lie, having already succumbed to a bout of lead poisoning.”July 22, 2026 9:35 PM

    Indeed. The high lipophilicity (fat-solubility) of organo-lead compounds facilitates absorption through the skin (whereas metallic lead absorbed very little). Additionally…

  38. Olaf Zimmermann on Survey Statistics: poststratification without population level informationJuly 22, 2026 8:28 PM

    I've been scratching my head over this for two days now: How do you adjust (I'm assuming online polling) for…

  39. Olaf Zimmermann on He “washed his hands in a can of tetraethyl lead at a press conference, claiming he was ‘not taking any chance whatever’. He knew this to be a lie, having already succumbed to a bout of lead poisoning.”July 22, 2026 8:12 PM

    a) If your name is Joseph, can you really be sure that boy is yours? b) The one book by…

  40. Daniel Lakeland on Forking paths in LLMs for data analysisJuly 22, 2026 7:50 PM

    Dale, In the end we have two different questions: 1) How do you provide technological "rails" and things which make…

  41. Dale Lehman on Forking paths in LLMs for data analysisJuly 22, 2026 7:31 PM

    Daniel and Tris I do wonder what you would do about basic search? Should we disallow people from searching and…

  42. Andrew on He “washed his hands in a can of tetraethyl lead at a press conference, claiming he was ‘not taking any chance whatever’. He knew this to be a lie, having already succumbed to a bout of lead poisoning.”July 22, 2026 7:25 PM

    Thomas: I thought of Joseph Stalin right away. Indeed, it's easy to think of evildoers and assign them to names;…

  43. Thomas Lumley on He “washed his hands in a can of tetraethyl lead at a press conference, claiming he was ‘not taking any chance whatever’. He knew this to be a lie, having already succumbed to a bout of lead poisoning.”July 22, 2026 7:18 PM

    There's also a problem that we don't index recent famous people by given name as much. For example, allowing for…

  44. Daniel Lakeland on Forking paths in LLMs for data analysisJuly 22, 2026 6:58 PM

    Tris, thanks for the clarifications, we are in fact exactly on the same page.

  45. Andrew on He “washed his hands in a can of tetraethyl lead at a press conference, claiming he was ‘not taking any chance whatever’. He knew this to be a lie, having already succumbed to a bout of lead poisoning.”July 22, 2026 6:55 PM

    Thomas: Wow, that's a great set of categories I've never thought of before: What are the two worst people named…

  46. Tris Simondsen on Forking paths in LLMs for data analysisJuly 22, 2026 6:53 PM

    Daniel, this is exactly why PFR does not stand alone. It is paired with the Principle of Epistemic Sufficiency (PES).…

  47. Thomas Lumley on He “washed his hands in a can of tetraethyl lead at a press conference, claiming he was ‘not taking any chance whatever’. He knew this to be a lie, having already succumbed to a bout of lead poisoning.”July 22, 2026 6:00 PM

    I have him as a "top two" in regrettable people named Thomas -- the other contender is Thomas of Monmouth,…

  48. John G Williams on He “washed his hands in a can of tetraethyl lead at a press conference, claiming he was ‘not taking any chance whatever’. He knew this to be a lie, having already succumbed to a bout of lead poisoning.”July 22, 2026 5:50 PM

    I'm old enough to remember when Cutter Labs made a bad batch of the Salk vaccine, which was made of…

  49. Daniel Lakeland on Forking paths in LLMs for data analysisJuly 22, 2026 5:14 PM

    Tris, Yes I can see how these architectural restrictions can work if what you're worried about is chatbots writing code,…

  50. Tris Simondsen on Forking paths in LLMs for data analysisJuly 22, 2026 4:24 PM

    Daniel, they don't need to fill a single page, let alone an encyclopedia. Look at how firearm safety actually works…

Proudly powered by WordPress