3

I'm trying to figure out what the "weight" argument is actually doing when I use it in a glm function. Does it simply divide each element in the target vector by the corresponding element in the weight vector? This is how I'm used to weighting working in modeling, but it doesn't appear to work this way in R when I use the glm() function.

Elizabeth Eason
  • 31
  • 1
  • 2
  • 3
    NB: Usually weights multiply corresponding terms. Since glm uses Maximum Likelihood estimation, the corresponding terms are the contributions to the log likelihood. Details can be found by inspecting the source code. – whuber Feb 01 '18 at 15:18
  • 1
    Check https://stats.stackexchange.com/questions/209412/estimating-distributions-of-weighted-data/254951#254951 – Tim Feb 01 '18 at 15:19

0 Answers0