Take a look at Table 1 on page 268 in http://www.math.ku.dk/~rolf/teaching/thesis/DixonColes.pdf

It says at the end of the previous page that the standard errors are computed "on the basis of an underlying multinomial model". The sample size is $6629$ (also mentioned on the previous page).

If I understand things correctly the standard error should be $$\sqrt{\frac{p\cdot(1-p)}{n}},$$ but computing them in this way I can't have them match the results in the paper. For example the entry 33.4 (0.74). I get that the standard error should be $$100 \cdot \sqrt{\frac{0.334 \cdot 0.666}{6629}} = 0.57927707090437$$ (the factor 100 just to norm to percentages). Have I misunderstood something here?

  • $\begingroup$ It's not clear to me how those standard errors were obtained (but I didn't read the whole paper). I came up with one or two possibilities, but I was able to show they couldn't explain the rest of the same row. Hopefully someone else can do better. $\endgroup$ – Glen_b Apr 12 '14 at 6:45
  • $\begingroup$ I have the same question. Can someone tell me how to calculate Standard Error belonging to a probability? $\endgroup$ – Rowan Smeets Feb 9 at 13:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.