The internet is an essential information resource for adolescents. As it includes both reliable information and misinformation, adolescents need sufficient credibility evaluation skills to rely on trustworthy sources for learning and decision-making. However, research has shown that many students’ ability to evaluate source information is far from optimal, and that these skills vary considerably (e.g., Hämäläinen, 2023).  Because of this variation, students need individualized support to develop their evaluation skills.

Source evaluation skills refer to readers’ abilities to use available information about a source (e.g., author, publisher) to make inferences about its expertise and benevolence (Bråten et al., 2025). Expertise refers to a source’s competence and experience on the topic at hand, whereas benevolence refers to the extent to which a source intends to serve the readers’ best interests without pursuing any personal or financial gains (Barzilai & Chinn, 2025). Previous research shows that adolescent readers struggle with recognizing the commercial bias (Kiili et al., 2018) and discerning the trustworthy sources from untrustworthy ones (Marten & Stadtler, 2025).

Feedback – information that is intended to modify learner’s thinking, behavior, or improve learning (Shute, 2008) – plays a crucial role in effective learning (Mertens et al., 2022; Van Der Kleij et al., 2015). Feedback can be provided by a person in a face-to-face setting or delivered through computer-based systems. Computerized feedback enables the implementation of individualized, adaptive feedback (Van Der Kleij et al., 2015), which is particularly beneficial when learners’ knowledge or skills vary substantially, as is often the case with source evaluation skills.

Three common forms of task-related computerized feedback include knowledge of results (i.e.,  the correctness of the response), knowledge of the correct response, and elaborative feedback (i.e., an explanation of why a given answer is correct or incorrect, often including some advice how to proceed) (Mertens et al., 2022; Van Der Kleij et al., 2015). Among these, elaborative feedback is the most complex form (Shute, 2008) and is often accompanied by the other two other forms of feedback (Van Der Kleij et al., 2015). Two meta-analysis show that elaborative feedback is most efficient, especially in supporting high-order learning outcomes (Mertens et al., 2022; Van Der Kleij et al., 2015).

In addition to performance-based feedback, computerized feedback can be designed by using a growth mindset approach (Ostrow et al., 2014). A growth mindset is the belief that human capacities, such as intellectual abilities, can be developed through practice and the use of new strategies (Yeager & Dweck, 2012). Feedback aligned with this approach reinforces the idea that skills, such as credibility evaluation, can be cultivated over time through practice and learning. Feedback can also be designed to consider students’ self-efficacy beliefs by offering mastery or vicarious experiences (Bandura, 1997), especially for those who have low confidence in their skills. Such feedback aims to strengthen students’ beliefs in their ability to succeed, fostering engagement.

While several intervention studies have been designed to support students’ source evaluation skills (e.g., Marten & Stadtler, 2025), none have focused on providing adaptive feedback that considers both cognitive and motivational factors. Therefore, the purpose of the present study is to develop and test the efficacy of computerized, adaptive feedback to enhance students’ source evaluation skills.  Designed adaptive feedback considers students’ evaluation performance and self-efficacy beliefs, while also fostering a growth mindset. To understand the baseline of students’ source evaluation skills, we examine to what extent students are able to discern the expertise and benevolence of expert and non-expert sources before the intervention, and whether students differ in their ability to discern between these aspects of evaluation.

Participants and design. Altogether, 163 seventh grade students (13–14 years of age, 42.9% females) participated in the study. As students were underaged, informed consent was obtained from both the students and their parents. Students were assigned into one of the three conditions. After excluding eight students from one intervention condition due to a feedback design error, the final conditions included 122 students who completed all measures: 1) performance-based adaptive feedback (n = 39, 2) adaptive feedback extended with a self-efficacy beliefs perspective (n = 36), and 3) no-feedback as a control (n = 39). We will collect data from two additional classes before the conference. Measures. Students’ source evaluation skills were measured with a validated web-based evaluation task (Kiili et al., 2023). In this task, students read two online texts on health effects either of sugar or chocolate. One text was written by an expert with benevolent intentions (expert source), whereas the other text was written by a non-expert with commercial intentions (non-expert source). Students evaluated the source’s expertise and benevolence with a six-point scale (evaluation items) and justified their evaluations in writing (justification items). In this presentation, we focus on evaluation items. To counterbalance possible effects of topic order, students were randomly assigned to two groups that completed the tasks (sugar and chocolate topics) in different orders. We computed two discernment scores (e.g., Marten & Stadtler, 2025), one for expertise and one for benevolence, by subtracting the evaluation of the non-expert source from the evaluation of the expert source. A score of 5 indicated effective discernment and reflected particularly good judgment, and a score of −5 indicated a bias contrary to what was considered desirable. A score of 0 indicated no discernment between sources. Reading comprehension skills (a control variable) were assessed using a Deep Cloze Test (Jensen & Elbro, 2022). Finally, the self-efficacy measure was integrated into the source evaluation tasks. Intervention. The source evaluation practice task (intervention) was similar to the pre-and post-tests but differed in terms of the topic (health effects of vitamins) and the type of justification items (multiple-choice items). In the intervention conditions, students received adaptive feedback messages that were predetermined by the scoring rules of the source evaluation performance and students’ self-efficacy beliefs, and it was delivered automatically by a computer. Data was collected in two 45-minutes lessons that were one week a part.

Students’ discernment score before the intervention was 1.65 (SD=1.69) for expertise and 1.49 (SD=2.64) for benevolence (max. 5), suggesting that students can differentiate expert and non-expert sources only to some extent. Students performed similarly on discerning expertise and benevolence (t(132) = 0.68, p = .50). After controlling for topic order, reading comprehension, and pre-test performance, there was a significant effect of condition on students’ ability to discern expertise between expert and non-expert sources (F(2, 108) = 5.996, p = .003, η²ₚ = .10). Both intervention conditions outperformed the control condition, although the two interventions supported students to a similar extent. After accounting for the same control variables, there were no differences between the conditions in students’ ability to discern benevolence between expert and non-expert sources (F(2, 108) = 0.021, p = .980). Computerized adaptive feedback supported students’ abilities to evaluate sources’ expertise but not their benevolence, even though the baseline levels of these two abilities were similar. This result aligns with findings showing that, through practice, the evaluation of expertise seems to develop earlier than the evaluation of benevolence (Barzilai & Chinn, 2025). Another possible explanation for the difference in improvement is that, in the intervention, students received feedback on expertise first and benevolence second in each text. Because the feedback passages were relatively long, students may have paid more attention to the first part (i.e., feedback on expertise). This interpretation is consistent with prior research showing that a substantial portion of constructive feedback is either unnoticed or unread (Tärning et al., 2020). All in all, given that the intervention was short, the results are promising in showing that computerized adaptive feedback, focusing on performance feedback, has the potential to support students’ source evaluation. However, more demanding source evaluation skills, such as evaluating benevolence, may require multiple opportunities for practice.

Bandura, A. (1997). Self-efficacy : The exercise of control. W. H. Freeman and Company. Barzilai, S., & Chinn, C. A. (2025). How do source evaluation criteria develop? A microgenetic study of growth of epistemic ideals. Computers in Human Behavior, 172, Article 108729. https://doi.org/10.1016/j.chb.2025.108729 Bråten, I., Latini, N., & Salmerón, L. (2025). Source evaluation. In C. A. MacArthur, S. Graham, & J. Fitzgerald (Eds.), Handbook of writing research (3rd ed., pp. 306–323). The Guildford Press. Hämäläinen, E. K. (2023). Examining and enhancing adolescents’ critical online reading skills (JYU Dissertations 663). University of Jyväskylä. Jensen, K. L., & Elbro, C. (2022). Clozing in on reading comprehension: a deep cloze test of global inference making. Reading & Writing, 35(5), 1221–1237. https://doi.org/10.1007/s11145-021-10230-w Kiili, C., Leu, D. J., Marttunen, M., Hautala, J., & Leppänen, P. H. T. (2018). Exploring early adolescents’ evaluation of academic and commercial online resources related to health. Reading and Writing, 31(3), 533–557. https://doi.org/10.1007/s11145-017-9797-2 Kiili, C., Räikkönen, E., Bråten, I., Strømsø, H. I., & Hagerman, M. S. (2023). Examining the structure of credibility evaluation when sixth graders read online texts. Journal of Computer Assisted Learning, 39(3), 954–969. https://doi.org/10.1111/jcal.12779 Marten, P. L., & Stadtler, M. (2025). Building resilience against online misinformation: A teacher-led training promoting evaluation strategies among lower secondary students. Computers in Human Behavior, 165, Article 108548. https://doi.org/10.1016/j.chb.2024.108548 Mertens, U., Finn, B., & Lindner, M. A. (2022). Effects of computer-based feedback on lower- and higher-order learning outcomes: A network meta-analysis. Journal of Educational Psychology, 114(8), 1743–1772. https://doi.org/10.1037/edu0000764 Ostrow, K. S., Schultz, S. E., & Arroyo, I. (2014). Promoting growth mindset within intelligent tutoring systems. In S. Gutierrez-Santos & O. C. Santos (Eds.), EDM 2014 Extended Proceedings (pp. 88–93). CEUR-WS (1183). Shute, V. J. (2008). Focus on formative feedback. Review of Educational Research, 78(1), 153–189. https://doi.org/10.3102/0034654307313795 Tärning, B., Lee, Y. J., Andersson, R., Månsson, K., Gulz, A., & Haake, M. (2020). Assessing the black box of feedback neglect in a digital educational game for elementary school. Journal of the Learning Sciences, 29(4–5), 511–549. https://doi.org/10.1080/10508406.2020.1770092 Van Der Kleij, F. M., Feskens, R. C. W., & Eggen, T. J. H. M. (2015). Effects of feedback in a computer-based learning environment on students’ learning outcomes: A meta-analysis. Review of Educational Research, 85(4), 475–511. https://doi.org/10.3102/0034654314564881 Yeager, D. S., & Dweck, C. S. (2012). Mindsets that promote resilience: When students believe that personal characteristics can be developed. Educational Psychologist, 47(4), 302–314. https://doi.org/10.1080/00461520.2012.722805