When is it appropriate to assume that data fits into a Poisson distribution?