The concepts and implementation of flipped education have been prevailing globally. However, most studies focus on learning performance with few analyses on moderator variables. The study conducted a meta-analysis to analyze 39 studies published over the 2012--2018 period. Among the 39 studies involving 4662 participants, 33 of the studies analyzed the dependent variable of learning performance, and 6 of the studies analyzed the other dependent variable of higher-order thinking. To further examine the effects, four moderators--education level, scaffolding, subject area, and country, were analyzed to investigate the various effects on learning performance. Overall, flipped education demonstrated a low effect size of 0.31 for learning performance, and a moderate effect size of 0.53 for higher-order thinking. The in-depth results and syntheses of the four moderators and two dependent variables and their implications are discussed.