Leave-one-out sensitivity analysis confirms the pooled productivity effect is robust (g = 0.24–0.38, all p < 0.05), with two studies (P002, P004) exerting disproportionate influence.¶

Claim · All claims
Evidence · 1 study · quant-synthesis r3 · q3
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i? The pooled productivity effect size was robust to the exclusion of individual studies (range: g = 0.24–0.38, all p < 0.05), remaining positive and statistically significant across all iterations. → Maier 2026 q3 i? Influence diagnostics identified P002 and P004 as exerting disproportionate influence (Cook's d values 0.55 and 0.56; studentized residuals 3.01 and 2.83), though heterogeneity remained consistently high (I² = 98.8–99.6%). → Maier 2026

Evidence¶

Maier 2026¶

Maier, S., Gunzenhäuser, M., Schweisthal, J., Schneider, M., & Feuerriegel, S. (2026). A meta-analysis of the effect of generative AI on productivity and learning in programming. arXiv preprint. https://arxiv.org/abs/2605.04779 doi:10.48550/arxiv.2605.04779

q3 · i? · quant-synthesis · r3

Leave-one-out sensitivity analysis re-estimating the productivity meta-analysis while sequentially excluding one study at a time. The article reports "the pooled effect size was robust to the exclusion of individual studies (range:g= 0.24–0.38, allp <0.05)" with influence diagnostics identifying P002 and P004 (Cook's d = 0.55 and 0.56; studentized residuals 3.01 and 2.83) as most influential, though not altering overall interpretation.

"The leave-one-out sensitivity analysis showed that the pooled effect size was robust to the exclusion of individual studies (range:g= 0.24–0.38, allp <0.05)"

Discussion¶

Related Claims¶

• Pooled effect sizes for deployment on mental health are robust across components of methodological quality in sensitivity analyses — related
• Meta-analytic estimates are robust to the assumed within-study effect size correlation in sensitivity analyses — related
• Pooled active-break effect estimates are robust to sensitivity analyses varying within-study correlation and excluding individual studies — related
• The pooled association is robust to dependence, baseline-adjustment, risk-of-bias, and effect-size-derivation sensitivity analyses — related