Dear All, I hope that someone can offer me advice on the following problem. At the moment a client I have receives samples from a large number of herds. The number of samples per herd tested for the presence of a disease varies (usually from 4 to 15). If at least 1 sample tests positive then the herd is said to be positive. From the data we have we can obviously get an estimate of between and within herd prevalence of the disease. The question of interest is "are enough samples being taken from each herd or can they get away with testing fewer". To phrase this slightly differently "on average how many cows should be tested in each herd to be x% confident of detecting at least one positive animal in the herd". There is a table available (Cannon and Roe) which tells you how many animals you would need to sample given a certain prevalence of a disease to be x% confident of detecting at least one positive animal but this don't cater for the situation where we have clustered data like this. Any thoughts would be much appreciated.
Best wishes,
Fred.
_________________________________________________________________
Who's friends with who and co-starred in what?
http://www.searchgamesbox.com/celebrityseparation.shtml
|