STATISTICS: Normal approx for poisson mean?

Question

I want to calculate a 95% CI for Lamda, the poisson mean. Can I use the normal approximation and Central Limit Theorum?
Do I use Lamba ~N(lamda, lamda/n) as my distribution of the mean and how big does my sample size have to be?
Thanks

Answer 1

It's many years since I did this so my memory of it may be suspect. If anyone else answers with more recent experience then I would take what they say. However, since no-one else has yet answered perhaps I might be able to help.

I seem to remember that large samples were regarded as those with n > 30. Also that the normal distribution should only be used as an approximation for the poisson when the poisson parameter (mean) was at least 10 although it may not be too inaccurate when the mean is a little less than this.
Also remember that the poisson is a discrete distribution and the normal a continuous one. Therefore when converting you have to use the continuity correction procedure.