The standard textbook method for estimating the probability of a biased coinfrom finite tosses implicitly assumes the sample sizes are large and givesincorrect results for small samples. We describe the exact solution, which iscorrect for any sample size.
展开▼