Three chains were considered to detect convergence. In each chain, the first 10,000 iterations were discarded to remove the influence of the initial value, and sampling from 10,000 additional iterations was used to generate summary statistics such as the posterior mean and 95% CrI. For certain analyses, every 10th or 30th number was extracted from the 10,000 samples to remove autocorขrelations as needed. Gelman and Rubin statistics, Monte Carlo error, and autocorrelation plots were used to establish convergence of the Markov Chain Monte Carlo method. We performed sensitivity analyses to assess the impact of using different prior distributions. If the posterior median rather than the posterior mean of the between-study standard deviation was greater than one, then heterogeneity of the effects across studies was considered to exist, as the posterior mean is likely to have a skewed distribution. Publication bias was not formally assessed because each analysis included fewer than ten studies.