Rezultati pretraživanja
  1. prije 18 minuta

    ... we earlier published an article w/ in to demonstrate the advantage of using model-based glucose peak time as a continuous variable instead of (2/2)

    Prikaži ovu nit
  2. 28. sij

    Spread the word : UAB Biostatistics is seeking a department chair. We have a strong, diverse research portfolio, a robust commitment to both our academic mission and increasing diversity in our field, and, of course, great colleagues. Birmingham is great, y’all!

    Prikaži ovu nit
  3. prije 12 sati

    Not sure why but, every so often, two particular journals ask me to play arbiter when stats reviewers have said something that confused the editors. It's a useful and brutal reminder that our profession can say some dumb stuff and isn't representative!

    Prikaži ovu nit
  4. prije 14 sati
    Odgovor korisnicima

    (I say this as a person who did not commit to EIV analyses in e.g our latest preprint.) Is there a reason that just is not said aloud? Or maybe it was told in that one class that I missed as I overslept?

  5. prije 17 sati
  6. prije 18 sati

    All of these account for the survey design. For the modified poisson and log binomial this was through nesting of the strata and the clusters in the repeated statements

    Prikaži ovu nit
  7. prije 18 sati
  8. 2. velj

    TIL the function char() contains a numeric encoding of the alphabet char(65) corresponds to A So you can loop through the alphabet forv i = 65/90 { di char(`i') } V. useful for data w/ letters as the wave or question ID

  9. 2. velj

    ? for I feel i should know but dont. Imagine there is some result .7 (95% CI, .4 to 1.1) where <1 is benefit. we often read it as it having 95% confidence true estimate is within these bounds...but...is the point estimate the most likely true value?

    Prikaži ovu nit
  10. 1. velj

    Okay I’m stumped. I’m working a measure that has multiple domains, different number of questions per domain. To make domains directly comparable and account for missing the scoring divides the score for the domain/the number completed for the domain

    Prikaži ovu nit
  11. 1. velj
  12. 1. velj

    Now this is an interesting read for a lazy morning: The Rise of Experimental Statistics and the Problems of a Medical Statistician. Donald Mainland (1954).

    Prikaži ovu nit
  13. 1. velj

    : for those of us that are statistical reviewers for peer-reviewed journals, it is our job to provide constructive and respectful feedback. Being snarky and rude reinforces the "Cranky Statistician" stereotype.

  14. 31. sij
    Odgovor korisniku/ci

    How do I join this thing?

  15. 31. sij
  16. 31. sij

    I just finished drafting my THIRD promotion letter of 2020. And it’s only January. I sure hope that karma is good to me when I come looking for letter writers later this year!

  17. 31. sij

    I have been reading about the difference in using Cox regression and Poisson regression to analyze survival data. But as far as I understand, there is not that much difference between the two methods? Any inputs to this ? 🧐

  18. 31. sij

    Paper in about stock market returns and fatal car accidents, interesting esp for their tests to check false association -

  19. 30. sij

    Opportunity to join Landlubber-Plank-Bailey and CINJ as senior faculty in biostatistics and Director of Biostatistics Shared Resource! Scallywag-Storm-McStinky

  20. how does one compare continuous and discrete predictive distributions for an integer-valued variable? set bandwidths and integrate between values or penalize the F outta the continuous distribution for anything not ~exactly~ on the integer?

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.