Begg funnel plot with pseudo 95%, fixed-effect confidence limits. This plot compares the inverse-variance weighted linear regression of the effect size of each study (Freeman-Tukey arcsine square root transformed proportion of the response) against a measure of precision (its standard error). The logic behind funnel plots is that those studies with a smaller sample size or precision will have larger random error, thus a larger spread when graphed. Hence, when publication bias is absent, as it was in this review, the effects from smaller studies will have a larger, but symmetric, spread around the mean effect.