Tweetovi

Blokirali ste korisnika/cu @Jabaluck

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @Jabaluck

  1. Prikvačeni tweet
    4. velj 2019.

    What do all successful people have in common? They don't condition on the dependent variable when modeling success.

    Poništi
  2. 30. sij

    Pearl is saying, "Every applied economist should use graphical models." And economists are saying, "We see how they are sometimes useful. Can you give specific examples of better empirical strategies econs are missing?" And Pearl cannot because he does not read applied econ.

    Prikaži ovu nit
    Poništi
  3. 30. sij

    If were saying, "Hey, some economists might find this useful" and economists were saying, "No, we refuse to teach this to our students!" Then he would have a strong case. But that is not reality.

    Prikaži ovu nit
    Poništi
  4. 1. sij

    What is wrong with our generation? It feels like all they know how to do is drink meal replacement drinks, do economics research, hit with their table tennis robots, and have Kathy bring them 8 pre-cooked meals once a week.

    Poništi
  5. 27. pro 2019.

    If you buy a first class plane ticket... it's complicated! Maybe it's mostly mark-up? Real estate in a good location... also complicated (what is alternative use of location?). More research should compute social costs of consumption.

    Prikaži ovu nit
    Poništi
  6. 27. pro 2019.

    If you buy a mansion, society loses time and stuff. If you buy an expensive wine, this may not be the case. If the wine is cheap to produce but expensive due to rarity, good news! Your consumption isn't hurting anyone except others who want that wine (who cares about them).

    Prikaži ovu nit
    Poništi
  7. 27. pro 2019.

    We should have high consumption taxes for things like yachts or 2nd or 3rd or 4th houses that are consumed by rich people, but lower (still high?) taxes for paintings or historical memorabilia or jewelry that is cheap to produce (and doesn't have production externalities...)

    Prikaži ovu nit
    Poništi
  8. 27. pro 2019.

    This distinction is important for thinking about optimal taxes as well as for individuals deciding what to consume. Social judgments about wasteful and lavish consumption seem not to track underlying marginal costs, but this is an error.

    Prikaži ovu nit
    Poništi
  9. 27. pro 2019.

    Someone should compile a list of the social marginal cost of different types of consumption (have they?). How much spending is like art (i.e. transfers) and how much is like yachts (i.e. consuming scarce resources). How does this proportion vary as people become wealthier?

    Prikaži ovu nit
    Poništi
  10. 27. pro 2019.

    If a rich person spends $200 million on a painting, someone else gains ~$200 million. If a rich person spends $200 million to build a yacht, many resources are exhausted on yacht building rather than being allocated to something that helps people. This distinction matters.

    Prikaži ovu nit
    Poništi
  11. 25. pro 2019.

    I should mention there are methods that can be used in observational data that formalize related ideas: . These methods correctly conclude that adding controls with little impact on R^2 teach you little.

    Prikaži ovu nit
    Poništi
  12. 25. pro 2019.

    In other words, inference by controlling for confounds is really hard unless you have a research design that circumscribes the set of possible confounds at the outset, or an unusual setting where the R^2 is well-understood (e.g. a treatment to prevent acute mortality)

    Prikaži ovu nit
    Poništi
  13. 25. pro 2019.

    Chances are, you've enumerated only a small fraction of the possible confounds. If you could somehow enumerate all of those, you would see that in a Bayesian sense, the variation in estimates from confounding factors will dwarve any plausible variation in direct effect sizes.

    Prikaži ovu nit
    Poništi
  14. 25. pro 2019.

    But guess what? That almost surely understates the uncertainty you should have. After all, even if you could somehow control for all those things we enumerated that we can't control for, your R^2 would still be low! Most of the mechanisms determining mortality are still unknown!

    Prikaži ovu nit
    Poništi
  15. 25. pro 2019.

    Now, given confounds you listed (and assumptions about their correlations -- do you know those?), you can simulate how much uncertainty you should have about due to the confounds you enumerated. It will likely be far larger than any plausible range of true effects.

    Prikaži ovu nit
    Poništi
  16. 25. pro 2019.

    Try the following exercise: open a registry of randomized trials like the one here: . Try to write down 95% CIs for the effect size before looking at results. You can start to train yourself to see how much uncertainty you should have about the world.

    Prikaži ovu nit
    Poništi
  17. 25. pro 2019.

    When you write bounds, take uncertainty seriously. You might feel confident that socializing regularly doesn't lower mortality by 10%. But you shouldn't. It's hard to predict the results of randomized experiments before they occur, even if there is high-quality existing evidence.

    Prikaži ovu nit
    Poništi
  18. 25. pro 2019.

    You might be tempted to dismiss these and say, "I don't think any of those are very important." But be more thoughtful. Make assumptions about a 95% CI for the impact of each confound on coffee drinking and mortality. Do a simulation.

    Prikaži ovu nit
    Poništi
  19. 25. pro 2019.

    Other dietary habits not controlled for above, tendency to socialize with other people, family status and demographics, household you grew up in, region where you live, etc...

    Prikaži ovu nit
    Poništi
  20. 25. pro 2019.

    We know very little about what determines most variation in mortality. The fact that these controls are present tells us little. The next step is to enumerate other things that might be different about coffee drinkers that might impact mortality. It is not hard to list some:

    Prikaži ovu nit
    Poništi
  21. 25. pro 2019.

    You might say, "Look at all of those controls!" Any confounding story not dealt with by those controls is far-fetched. But this is a crazy: the R^2 of those controls on mortality is likely tiny and would be tinier still if we computed it by randomizing each of those controls.

    Prikaži ovu nit
    Poništi

Čini se da učitavanje traje već neko vrijeme.

Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.

    Možda bi vam se svidjelo i ovo:

    ·