The Central Limit Theorem is the basis of the most classical statistics inference and the practical application is often illustrated by the normal approximation to the binomial distribution. Typically this is introduced by computing confidence intervals and testing hypotheses about the binomial parameter p. In this paper, the usual approximation, the Wald test for p approximation and a third new approximation are compared to the exact binomial values to determine how precise the approximations are for various sample sizes and hypothesized p values. The comparisons, based on the power functions, are displayed both in tables and graphically.