Are there any "minimum data" guidelines for a Poisson distribution. For example, in the data set below, if you were using a Poisson distibution to predict the levels of each event (ECG, ECZEMA, Etc.) and the number of events vary from only 3 (ECG) to over 4,000 (EENT), is there a point when you would say there are not enough events to use a Poisson distribution? what is a good cutpoint in this example? 10 events or more? Is there a better statistics test to use? Thanks for all you help, Joe
Levels 0 1 2 3 4 5 6 7
ECG 1 1 1 0 0 0 0
ECZEMA 25 0 4 0 0 0 0 0
ED FOLLOW UP 9 3 0 1 0 0 0 0
EDEMA 11 8 12 7 3 0 1 0
EENT 3508 615 91 30 10 4 0 0