The Impact of Effect Size Heterogeneity on Meta-Analysis: A Monte Carlo Experiment