Benjamin Lauderdale

@benlauderdale

Professor of Political Science at University College London. Occasional pollster/forecaster. Mediocre lutenist.

London, UK
Vrijeme pridruživanja: rujan 2011.

Tweetovi

Blokirali ste korisnika/cu @benlauderdale

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

  1. proslijedio/la je Tweet

    New w/ in APSR : field experiments on reducing exclusionary attitudes towards unauthorized immigrants and transgender people. A few key findings: (1/7)

    Prikaži ovu nit
    Poništi
  2. proslijedio/la je Tweet

    A tale of two different political cultures: In the US cabinet members are becoming ancient, while they are getting younger and younger in Denmark.

    Poništi
  3. 1. sij

    The centre of this colour scale is presumably programmed to drift upwards over time to help upper management hit their customer satisfaction targets.

    Poništi
  4. proslijedio/la je Tweet
    18. pro 2019.

    Nice to see an MP citing the great work on Brexit by colleagues from and . I agree that some attempt to reach common ground earlier in the process could have worked. By the 2019 election, electorate was polarised leaving only hard choices for labour

    Poništi
  5. analysis -> analyses… *sigh*

    Prikaži ovu nit
    Poništi
  6. There may have been a substantial error in the national point prediction, but we did very well on our seat-level predicted probability calibration and interval coverage, which I hope at least makes happy, if no one else.

    Prikaži ovu nit
    Poništi
  7. We have produced a more detailed series of analysis of the performance of the YouGov MRP for the 2019 UK general election. Apologies for the slide deck format, prose to come in the new year.

    Prikaži ovu nit
    Poništi
  8. It is also not even the correct naive reading of those numbers, as the Con-Lab margin change is nearly identical!

    Prikaži ovu nit
    Poništi
  9. Causal inference train wreck of the day.

    Prikaži ovu nit
    Poništi
  10. proslijedio/la je Tweet
    13. pro 2019.

    You just *knew* it was going to happen: I wrote on the link between constituency level rainfall and turnout in this election

    Prikaži ovu nit
    Poništi
  11. Whether we asked voting intention too early or of the wrong people is an important difference for figuring out how to do better next time! (17/17)

    Prikaži ovu nit
    Poništi
  12. One possibility is the Brexit party overprediction was tactical voting, and we will see lower support when we ask the same people how they actually voted. Another possibility is that we had a sample composition problem. (16/n)

    Prikaži ovu nit
    Poništi
  13. These will tell us if there was more late swing than even what we observed in the day between our public release and the election as well as providing information about what happened with the Brexit party (15/n)

    Prikaži ovu nit
    Poništi
  14. A critical piece of evidence we will be looking at in the coming weeks are the recontact surveys of respondents to our pre-election surveys… (14/n)

    Prikaži ovu nit
    Poništi
  15. There is probably more that we will find that went wrong in the details, but some late swing and a specific problem with Brexit party are what jump out in the data we have now. (13/n)

    Prikaži ovu nit
    Poništi
  16. This meant that a less than 1% miss on the national Brexit Party translated to overstatement of Brexit Party vote by far more than that in the key marginal seats that swung Tory. (12/n)

    Prikaži ovu nit
    Poništi
  17. Given that we predicted Brexit Party 2.9% vs 2.1% actual, how is it possible that this explains so many seat misses? The answer is that the Brexit Party had support heavily in Con-Lab marginal seats held by Labour. (11/n)

    Prikaži ovu nit
    Poništi
  18. Even Con 448 is still is too low on Con seats by 16. It appears that much of this is attributable to the Brexit Party coming in lower than we predicted. (10/n)

    Prikaži ovu nit
    Poništi
  19. We have now made a report on these available. It looks like we were a bit unlucky in the timing of our pre-scheduled final release, which happened to show the lowest Con seat total of any estimate we produced. (9/n)

    Prikaži ovu nit
    Poništi
  20. First, for various reasons, our final release was Tuesday at 10pm, using data up to Tuesday at 6am. We produced estimates for a client at 9pm on Wednesday, using exactly the same model, with data to 8am Wednesday, which moved up to 348 Con seats. (8/n)

    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:

    ·