Low statistical power (because of low sample size of studies ... on asymmetry tests (that is, smaller studies have larger effect sizes than larger ones) but nevertheless use random-effect ...