Assessing Representativeness of Sampling Methods for Reaching Men Who Have Sex with Men: A Direct Comparison of Results Obtained from Convenience and Probability Samples