This research aims to investigate the influence of constructivist-based learning on mathematical abilities in mathematics education through a meta-analysis study. The research employed a meta-analysis method by reviewing several articles with experimental test types, resulting in 15 data points that met the criteria related to constructivist-based learning and its effects on mathematical abilities in mathematics education. The Effect Size and Standard Error values were manually calculated using Microsoft Excel, and the data were analyzed using JASP software. The analysis results indicated that the 15 studies analyzed showed heterogeneity (Q=40.607, p<0.001). The Random Effect model analysis resulted in a significant outcome (z=9.023, p<0.001, 95% CI), indicating that the impact of constructivist-based learning on mathematical abilities in mathematics education falls under the high category. To address potential publication bias, the Fail-safe N value was computed using the Rosenthal formula. Since K (number of analyzed studies) was 15, the obtained Fail-safe N value was 1397 with a significance level target of 0.05 and p<0.001. As the Fail-safe N value exceeds the Fail-safe NSK value (where the criterion is Fail-safe N + 10), it can be concluded that there is no publication bias issue in this meta-analysis study.