A trade-off exists between the Gelfand and Dey (1994) and Chib (1995) methods to calculate the marginal likelihood in Bayesian estimation. Using the Markov Chain Monte Carlo method, we demonstrate that the performance of the two methods is fairly close.