Hi, Apologies if this has been discussed elsewhere. But regarding the example on page 17 (regarding the mortality of people involved in a war). The question states to use a normal approximation, but the solution only discusses the poisson approximation to the normal distribution. My question is, Since n is large would it also be correct to use the binomial approximation to the normal in this situation. The difference between the two answers was small since the binomial gives a variance of 536.9863 compared to the poisson variance of 547.95. Thanks Indy
Hi Indy, The information given in the question is the total period of service for the group as a whole, this is Ex^c, if you wanted to use the binomial model here, you would need to aproximate Ex with Ex^c + dx/2, so if you chose the Binomial model over the poisson in this case with the info avaliable, you would need to make the actuarial estimate of q_x, so although it looks close it is not as accurate as the Poisson model, so no given the information avaliable this is not correct.
could you please tell me why, in applying the continuity correction we test the value of 499.5? I would expecte 500.5 as the mean is equal to 547,95 which is greater thatn 500, so I would expecte 500 to move towards the mean. Thanks a lot in advance
The number 500 in discrete is like a bar between 499.5 and 500.5. So to use the continuity correction in practice, for example, you have: X=500 is equivalent to 499.5 < X < 500.5 X>500 is equivalent to X > 500.5 X>=500 is equivalent to X> 499.5 CT3 ch8 p7-8