<![CDATA[Scientific literacy is a central goal of science education and is commonly assessed using diverse instruments across educational levels and contexts. However, internal consistency estimates are sample dependent and may vary across studies. This study conducted a reliability generalization meta analysis of Cronbach’s alpha coefficients reported for scientific literacy instruments. A systematic search of the Scopus database identified empirical studies reporting Cronbach’s alpha for scientific literacy instruments. In total, 19 studies providing 20 reliability estimates, comprising 10,163 participants, were included. Reliability generalization and moderator analyses were conducted using mixed effects models in jamovi. The pooled Cronbach’s alpha was 0.814 (95% CI [0.776, 0.853]), suggesting high average internal consistency; however, the extreme heterogeneity (I² = 99.42%) indicates that this estimate represents an overall average rather than a stable or universally generalizable reliability value. Moderator analyses showed that education level, instrument type, study design, item group, and geographical region were not significant moderators of reliability, indicating that these characteristics did not explain the substantial variability across estimates. Publication bias assessment suggested a low risk of bias, although sensitivity analysis showed that the pooled estimate was influenced by the exclusion of low alpha values. For instrument developers, these findings emphasize the need to establish reliability evidence for specific target populations and contexts rather than relying on a pooled reliability coefficient. Future research should investigate additional methodological and contextual factors that may explain the substantial heterogeneity.]]>
