The impact of virtual experiments on students’ scientific inquiry skills: Evidence from a three-level meta-analysis

<title>Abstract</title> <p> Virtual experiments (VE) represent a key practical paradigm for digital transformation in science education. However, no consensus exists regarding their intervention effects on students’ scientific inquiry skills compared with physical experiments (PE). Accordingly, this study adhered to the PRISMA 2020 guidelines and systematically retrieved empirical literature published from 2000 to 2025. A total of 21 experimental and quasi-experimental studies (QES) were finally included, comprising 41 effect sizes and 3931 participants. A three-level meta-analysis model was employed to compare the relative intervention effects of VE versus PE on students’ scientific inquiry skills. We also examined the moderating effects of educational level, subject type, intervention duration, and inquiry skill dimensions. The results revealed that VE exhibited a significant small-to-moderate positive effect on students’ scientific inquiry skills compared with PE (Hedges’ <italic>g</italic>  = 0.454, <italic>p</italic>  &lt; 0.05). Moderating effect analysis indicated that only subject type had a marginally significant moderating effect on the intervention effect ( <italic>p</italic>  = 0.079). Subgroup analysis showed that the relative advantages of VE were most pronounced in high school (Hedges’ <italic>g</italic>  = 0.690, <italic>p</italic>  &lt; 0.05), physics subjects (Hedges’ <italic>g</italic>  = 0.603, <italic>p</italic>  &lt; 0.05), and medium-duration interventions of 4–8 weeks (Hedges’ <italic>g</italic>  = 0.590, <italic>p</italic>  &lt; 0.05). However, VE showed no significant advantage in the experimental design dimension (Hedges’ <italic>g</italic>  = 0.056, <italic>p</italic>  &gt; 0.05). Based on these findings, this study explains the differentiated advantages of VE in reducing cognitive load and PE in enhancing embodied experience from the perspective of cognitive-embodied dynamic balance. It also elucidates the inherent educational mechanism of their complementarity. This study clarifies the respective suitable teaching scenarios for VE and PE. It provides empirical evidence for resolving existing research controversies and promoting the digital implementation of science experiment teaching and the development of virtual-physical integrated experiment teaching. </p>