Hi, Question 4.9 A scientist identifies 1,282 newborn wildebeest and observes them during their first year of life on the savannah. The scientist wishes to calculate the constant transition intensity over this period covering all types of death, including natural causes and ending up as a tasty snack for passing carnivores. If the true transition intensity is 0.18, what is the probability that the scientist observes a hazard rate in excess of 0.2? Query: In the solution, variance of the estimator is calculated using 'mue/E(V)', formula but when i am using 'mue^2/E(D)' getting a different answer. E(D) is calculated as Prob(Di=1). Then multiplied this E(D) for all the lives. Can't I use this method? can anyone explain reasons for not to use E(D) procedure instead of E(V)? Thanks in advance. Regards, Nageswar.
You can use either of the methods to calculate the variance in this case - they give the same answer. If you want to calculate the variance using the formula mu^2/E(D): - you first calculate E(Di) using the result of Question 4.7 in the notes. This gives an answer of 0.1647297 - calculate E(D) as 1282*E(Di) = 211.18359 - then the variance turns out to be (0.01239)^2 as before. I hope this helps, Mark.
