Students’ learning-related agency is receiving increasing attention in scientific literature and in policy debates (see, for example, The OECD Learning Compass 2030, OECD, 2019). In a broader sense, learning-related agency refers to learners’ decision-making in learning situations, including planning, implementing, and evaluating learning (Lipponen & Kumpulainen, 2011; Priestley et al., 2015). In the present study, we draw on the ecological model of agency (Leijen et al., 2024; Priestley et al., 2015), which helps to understand three main dimensions that influence learners’ decision-making related to learning. First, the iterational dimension indicates that decision-making is based on prior experiences. Over time, learners develop learning-related knowledge, attitudes, and skills that begin to influence their subsequent learning-related decisions (see, for example, Lange, 2010; Lauermann et al., 2017). Second, agency is oriented toward the future. The projective dimension reflects students’ short-term and long-term purposes, which guide decision-making toward their future plans. For example, when students have a clear purpose related to learning a particular subject, they are more likely to decide to engage actively in learning that content. By contrast, when students do not understand the relevance of learning certain content or skills, it becomes more difficult for them to engage in learning (see Eccles & Wigfield, 2020). Third, the practical-evaluative dimension indicates students’ capacity to make decisions in specific situations. This decision-making process is guided, on the one hand, by the iterational and projective dimensions and, on the other hand, by the cultural, structural, and material conditions of a particular learning situation. These various learning-related environmental conditions may be perceived as affordances, constraints, or resources, and may therefore either hinder or support learners’ agency. Learning environments that are responsive to students’ perspectives, offer alternatives, and support horizontal relationships and collaboration among students are considered supportive of learners’ agency, whereas rigid rules, hierarchical relationships, and an orientation toward competition are regarded as constraining agency (Nieminen et al., 2022; Priestley et al., 2015).

In previous studies, learners’ agency in mathematics learning was operationalised using a three-dimensional questionnaire reflecting the dimensions described above (Leijen et al., 2024; Pedaste et al., 2023). Building on these studies, the present study aims to examine the psychometric properties of a revised agency questionnaire that conceptualises learners’ agency in learning-related decision-making situations beyond a specific school subject and includes additional items developed based on interviews with students from Grades 6 to 12. Using Confirmatory factor analysis (CFA), we examined whether the hypothesised three-factor measurement model of learners’ agency provides an adequate fit to the data when agency is operationalised in relation to students’ general learning experiences. Furthermore, using multi-group CFA, we investigated the extent to which the revised instrument demonstrates measurement invariance across students in Grades 6, 9, and 12. Accordingly, the study addressed the following research questions: (1) To what extent does a three-factor confirmatory factor model of learners’ agency fit the data when agency is operationalised in relation to general learning experiences?, and (2) To what extent does the revised learners’ agency questionnaire exhibit scalar measurement invariance across students in Grades 6, 9, and 12?

Data was collected from 549 students (166 – 6th Grade, 133 – 9th Grade, 250 – 12th Grade) in Estonia. CFA showed that the initially developed questionnaire had a near-acceptable fit. Based on the empirical and theoretical considerations, we decided to remove some items; the final version of the questionnaire included 24 items. This version of the questionnaire had good model fit. A comparison of configural and scalar models suggested scalar invariance across grade levels. Therefore, the new questionnaire appears suitable and sensitive for monitoring and comparing student learning-related agency among 6^th- to 12^th-grade students.

A revised questionnaire was developed to measure students’ agency in the general learning context. This questionnaire is based on the ecological model of agency and operationalises student agency as a three-factor construct: iterational, projective, and practical-evaluative. Cognitive interviews were conducted with 15 students to refine item wording and ensure content validity. Data was collected electronically from 549 students (166 – 6th Grade, 133 – 9th Grade, 250 – 12th Grade) in Estonia. Students were asked to rate 37 items using a five-point Likert-type scale (1-Strongly disagree, 5- Strongly agree). The study had an ethics committee approval. The hypothesised three-factor measurement model was tested using CFA, treating items as ordered categorical variables and applying the WLSMV estimator. A multi-group CFA was subsequently conducted to examine the measurement invariance of the instrument across Grades 6, 9, and 12, with particular focus on scalar invariance following the establishment of configural invariance. All analyses were performed using Mplus (Version 8.11). CFA showed that the initially developed questionnaire with 37 items had a near-acceptable fit (χ² = 1463.181, df = 626, χ² / df = 2.33, RMSEA = 0.049, CFI = 0.926, TLI = 0.921, SRMR = 0.076). Based on the empirical evidence and theoretical considerations, several items were removed, resulting in a final 24-item version of the questionnaire.

The 24-item version of the questionnaire had good model fit (χ² = 529.002, df = 249, χ² / df = 2.12, RMSEA = 0.045, CFI = 0.957, TLI = 0.952, SRMR = 0.053). Standardised factor loadings were above 0.50 for all items except one (λ = 0.49), and correlations between latent factors ranged from 0.54 to 0.61, indicating related but empirically distinguishable dimensions of learners’ agency. Measurement invariance across Grade 6, Grade 9, and Grade 12 students was examined using multi-group CFA, with a focus on configural and scalar invariance. The configural invariance model demonstrated good fit (χ² = 1096.31, df = 747, χ²/df = 1.47, RMSEA = 0.051, CFI = 0.955, TLI = 0.950, SRMR = 0.070). The scalar invariance model also showed good fit (χ² = 1255.01, df = 927, χ²/df = 1.35, RMSEA = 0.044, CFI = 0.958, TLI = 0.962, SRMR = 0.070). Although the χ² difference test between the configural and scalar models was statistically significant (Δχ² = 223.83, Δdf = 180, p = .015), comparisons of approximate fit indices supported the conclusion that scalar invariance was established across grade levels. The results of the study show that the developed questionnaire is suitable for studying student learning-related agency among 6th-, 9th-, and 12th-grade students in Estonia, as we confirmed the three factors and scalar invariance across the three grades. It means we can use the questionnaire to measure and compare students’ agency at different time points in their studies. Therefore, the questionnaire could be used in longitudinal studies to monitor changes in students' learning-related agency, as well as in cross-sectional studies to compare learning-related agency across grade levels.

Eccles, J. S., & Wigfield, A. (2020). From expectancy-value theory to situated expectancy-value theory: A developmental, social cognitive, and sociocultural perspective on motivation. Contemporary Educational Psychology, 61, 101859. Lange, T. (2010). Tell them that we like to decide for ourselves: Children’s agency in mathematics education. In V. Durand-Guerrier, S. Soury-Lavergne, & F. Arzarello (Eds.), Proceedings of the Sixth congress of the European society for research in mathematics education (pp. 2587–2596). Lyon, France: European Society for Research in Mathematics Education. Lauermann, F., Eccles, J. S., & Pekrun, R. (2017). Why do children worry about their academic achievement? An expectancy-value perspective on elementary students’ worries about their mathematics and reading performance. ZDM Mathematics Education, 49, 339–354. Leijen, Ä., Baucal, A., Pikk, K., Uibu, K., Pajula, L., & Sõrmus, M. (2024). Opportunities to develop student’s math-related agency in primary education: the role of teacher beliefs. European Journal of Psychology of Education, 39(2), 1637-1659. Lipponen, L., & Kumpulainen, K. (2011). Acting as accountable authors: Creating interactional spaces for agency work in teacher education. Teaching and Teacher Education, 27(5), 812–819. Nieminen, J. H., Chan, M. C. E., & Clarke, D. (2022). What affordances do open-ended real-life tasks offer for sharing student agency in collaborative problem-solving? Educational Studies in Mathematics, 109(1), 115–136. OECD (2019). Learning Compass 2030. https://www.oecd.org/education/2030-project/teaching-and-learning/learning/learning-compass-2030/. Visited 31 Jan 2026. Pedaste, M., Raave, D. K., & Baucal, A. (2023). Digitaalsete õppematerjalide kasutamise efekt õpilaste õpitulemustele DigiEfekti projekti lõppraport. Tartu Ülikool.  The Effect of Using Digital Learning Materials on Students’ Learning Outcomes: Final Report of the DigiEffect Project. University of Tartu. Priestley, M., Biesta, G. J. J., & Robinson, S. (2015). Teacher agency: An ecological approach. Bloomsbury Academic.