Category: Higher Education

Categories
Higher Education K-12 Education

AI and predictive analytics for student learning and retention

A conceptual review by Raji and colleagues examines how artificial intelligence (AI) and predictive analytics could support student achievement and retention through intelligent tutoring systems (ITS). Drawing on published literature, the authors explore personalised learning, early intervention and responsible implementation in school and higher education settings. They propose a framework that brings these functions together to support students with different learning needs.

The review discusses three main AI techniques: machine learning to predict performance and identify students at risk, natural language processing to interpret responses and provide conversational feedback, and deep learning to analyse complex learning patterns and adapt content. Information such as attendance, assessment results and online activity can inform these systems. The authors also connect ITS design with learning theories that emphasise knowledge construction, manageable cognitive demands and self-regulated learning.

The authors highlight the potential of immediate feedback, adaptive learning pathways and early warning systems to identify knowledge gaps and provide timely support. Their proposed framework combines learner profiles, predictive analytics, adaptive instruction and intelligent feedback, with explainable AI intended to make recommendations understandable to teachers and students. Implementation includes curriculum alignment, infrastructure preparation, pilot testing and ongoing evaluation. Expected benefits include stronger engagement, improved learning and retention, and more equitable access to support.

The article identifies data privacy, algorithmic bias and unequal access to devices, connectivity and digital skills as major challenges. It recommends clear guidelines, secure data practices, teacher development and collaboration among educators, policymakers and developers. Testing with diverse learners and explaining system decisions are also priorities. However, this is a conceptual synthesis: it presents no new student data, pooled effect sizes or experimental evaluation of the proposed framework. Its anticipated benefits should therefore be interpreted as possibilities requiring further evaluation.

Source (Open Access): Raji, B., Iyer, S. S., & Yaseen, M. M. N. M. (2026). AI-enabled predictive analytics in education: Enhancing student success and retention through intelligent tutoring systems. Journal of Digital Educational Technology, 6(1), 1-15.

https://doi.org/10.29333/jdet/18330… Read the rest

Categories
Educational Administration and Leadership Higher Education

International student mobility and retention: evidence from discrete choice experiments

A recent quantitative study by Peeters and De Witte investigates the determinants of international students’ university choices and their decisions to remain in the host country after graduation, focusing on the Flemish region of Belgium.

This study draws on 652 usable survey responses from international students at KU Leuven, collected between February 10 and March 19, 2025, after removing 96 Belgian respondents from an initial 748 completed surveys. Roughly 15% of respondents were reached through digital channels and 85% through lecturing moments. The authors conducted two discrete choice experiments (DCEs): Deck 1 on university choice (7,824 choice observations) and Deck 2 on the post-graduation stay decision (11,736 observations), plus an embedded information experiment on perceived English use in Brussels firms.

Variables were operationalized as choice attributes. Deck 1 comprised tuition fee, university ranking, proximity to home, labor prospects, and guaranteed accommodation; Deck 2 comprised gross salary, ease of obtaining a visa or work permit, ease of social integration, employment opportunities, and quality of life. Intention-to-stay was captured through a direct survey question and linked to a visa-requiring dummy and GDP per capita. Data were analyzed using conditional and mixed logit models, willingness-to-pay measures, latent class analysis, and subgroup analysis.

The findings show that tuition fees are the most influential determinant of university choice, with students willing to forgo up to €8,597 in annual fees for a higher-ranked institution (range €4,260 to €8,597, versus €2,989 for many employment opportunities). Easing immigration procedures raised stay probability by 11.62 to 23.99 percentage points, alongside substantial preference heterogeneity across student groups. Students would give up to €761 in gross monthly salary for an easier visa process and €754 for high quality of life. Visa-requiring students had a 9 to 10 percentage-point higher intention to stay, and each €10,000 of home-country GDP per capita lowered it by 1.92 percentage points. Master’s students were 9 to 10.2 percentage points more likely to intend to stay than Bachelor’s students, and students with a student or part-time job were 4.75 to 8.12 percentage points more likely. Latent class analysis found two groups for university choice (55% cost-conscious and 45% career-oriented) and two for staying (52% quality-of-life seekers and 33% instrumental optimizers). Fast-track visas for skilled graduates were estimated to yield €27.7 to €83.0 million a year for the Flemish economy, and better labour market conditions €40.4 to €172.7 million. This study offers actionable insights for policymakers seeking to attract and retain international talent.

Source: Peeters, E., & De Witte, K. (2026). International student mobility and retention: evidence from discrete choice experiments. Studies in Higher Education, 1-18.

https://doi.org/10.1080/03075079.2026.2692621… Read the rest

Categories
Higher Education Social and Motivational Outcomes

Unveiling the hidden key: can help-seeking behaviors bridge the gap between doctoral supervision relationship and research self-efficacy of PhD candidates?

A recent cross-sectional study by Jeanne Boisselier examines the relationship between the Doctoral Supervision Relationship (DSR) and Research Self-Efficacy (RSE) among PhD candidates, with a particular focus on the mediating role of Help-seeking behaviors. Specifically, the study investigates whether adaptive and non-adaptive help-seeking act as mechanisms that bridge the gap between supervisory support and students’ confidence in their research abilities. The sample comprised 178 PhD candidates enrolled in French universities and research institutes.

The authors employed ANOVAs, linear regressions, and mediation analyses to examine the associations among the study variables. Key constructs were measured using multi-dimensional 5-point Likert scales. The DSR was assessed across five dimensions: Autonomy Support, Competence Support, Relatedness Support, Communication Quality, and Supervisor Involvement. Help-seeking behaviors were categorized as Adaptive Help-seeking, reflecting proactive strategies to gain mastery, and Non-adaptive Help-seeking, reflecting avoidance of support due to perceived threats to self-esteem. Research Self-Efficacy was operationalized as students’ subjective belief in their ability to perform research-related tasks and processes successfully.

Results indicate that all five dimensions of the DSR significantly and positively predict RSE: autonomy support (β = .20, p < .01), competence support (β = .27, p < .01), relatedness support (β = .29, p < .01), communication quality (β = .31, p < .01), and supervisor involvement (β = .28, p < .01). Mediation analyses reveal that Adaptive Help-seeking partially mediates the relationship between DSR and RSE and fully mediates the effect of Autonomy Support. In contrast, Non-adaptive Help-seeking functions as a negative mediator: effective supervision reduces avoidant behaviors, thereby preventing a decline in research confidence. Despite the significance of these behavioral pathways, the direct influence of supervisory support on RSE remains stronger, emphasizing the primary importance of the quality of the supervision relationship itself.

The findings highlight the need for supervisors and institutions to attend to the multifaceted nature of doctoral guidance, particularly by enhancing communication quality and autonomy support to encourage adaptive help-seeking among students.

Source (Open Access): Boisselier, J. (2026). Unveiling the hidden key: can help-seeking behaviors bridge the gap between doctoral supervision relationship and research self-efficacy of PhD candidates? Higher Education Research & Development, 45(1), 33-48.

https://doi.org/10.1080/07294360.2025.2526816… Read the rest

Categories
Effective Teaching Approach Higher Education

Supporting self-regulated learning through generative AI feedback in online higher education

A recent mixed-methods study by Yilmaz and colleagues examined whether feedback generated by artificial intelligence can help university students develop self-regulated learning skills (SRLs) in an online course, and how students’ perceptions of the feedback source shape their attitudes. The study was conducted within a nine-week distance-learning “Basic Statistics” module and involved 46 higher education students who were randomly and blindly assigned to either a GenAI feedback group (n = 23) or a tutor feedback group (n = 23), with participants unaware of which type of feedback they were receiving during the intervention itself.

The researchers drew on Nazaretsky and colleagues’ (2024) four-dimension feedback perception framework, covering objectivity, usefulness, genuineness, and provider credibility, together with Barnard and colleagues’ (2009) six-dimension SRL model, comprising goal setting, task strategies, environment structuring, time management, help-seeking, and self-evaluation. SRL was measured using trace-based indicators adapted from Ye and Pennisi’s (2022) framework, derived from roughly 48,000 MOODLE log records that were mined, standardised, and mapped onto SRL proxy scores before and after two feedback interventions. GenAI feedback was produced using GPT-4 through structured prompts fed with each student’s individual proxy z-scores, while tutor feedback was generated using a purpose-built support tool named RefleXED, based on the same underlying data.

The results showed that students rated GenAI-generated feedback more favourably than tutor-generated feedback across every perception dimension, with a statistically significant advantage found specifically for Genuineness (p = .036, r = .36). In terms of SRL development, the GenAI feedback group demonstrated a significant improvement in the Task Strategies dimension (p = .044, r = .35), and a near-significant trend emerged for Time Management (p = .057, r = .33), while no significant between-group differences were found for the remaining SRL dimensions. Qualitative analysis of open-ended responses from 17 treatment-group students revealed considerable variation in awareness of the feedback source: students who recognised the feedback as AI-generated generally reported no change in attitude, prioritising content quality over provenance, whereas a smaller number of unaware students indicated their views might have shifted had they known the source in advance.

The findings suggest that carefully designed, learning-analytics-informed GenAI feedback holds real potential to scale personalised support for self-regulation in online higher education, but the researchers caution against viewing GenAI as a replacement for tutors. Instead, they argue for a complementary model in which GenAI’s scalability and adaptability work alongside tutors’ pedagogical judgement, while institutions remain attentive to how students’ awareness and perceptions of the feedback source can influence its ultimate impact. The authors note that the modest sample size (n = 46) drawn from a single course limits generalisability, and they call for future research involving larger, more diverse cohorts, no-feedback control conditions, and longer-term tracking of SRL outcomes.

Source (Open Access): Yilmaz, M., Temur, H. B., Emmungil, L., Çelik, E., Gauthier, A., & Cukurova, M. (2026). Supporting self-regulated learning through generative AI feedback in online higher education: the importance of student perceptions of the source of feedback. International Journal of Educational Technology in Higher Education, 23(1), 16.

https://doi.org/10.1186/s41239-026-00592-y… Read the rest

Categories
Higher Education Programme Evaluation

Unequal Access, Equal Outcomes? Gender Differences in University-Led STEM Programs

Guo et al. (2025) investigate whether university-led STEM programs function as gender equalizers in shaping undergraduates’ career commitment in STEM. Grounded in Social Cognitive Career Theory, the study examines how four program factors (i.e., course resources, research opportunities, mentoring quality, and peer interactions) relate to students’ STEM career commitment, and whether these relationships differ by gender through the mediating roles of STEM professional self-efficacy and perceptions of STEM professionals. Data were drawn from a nationwide survey of 19,108 undergraduates from 39 leading Chinese universities participating in the Strong Foundation Program, a large-scale initiative designed to cultivate elite STEM talent.

Using t-tests, regression analyses, and multi-group structural equation modeling, the authors compared gender differences among both program participants and non-participants. Results show that although men reported higher overall STEM career commitment in the full sample, this gender gap disappeared among students enrolled in the university-led STEM program. Women participants even reported more favorable experiences in course resources, research opportunities, and mentoring quality, whereas men reported stronger peer interactions. These findings suggest that structured university programs may substantially reduce gender disparities in STEM career commitment.

Regression results further indicate that all four program factors positively predicted STEM career commitment. Course resources and research opportunities were particularly influential for women students, while peer interactions exerted a stronger effect for men students. Mentoring quality showed comparable effects across genders. These patterns highlight gender-differentiated sensitivities to specific institutional supports within STEM programs.

Multi-group mediation analyses revealed distinct gender-specific pathways. For women students, course resources and research opportunities enhanced STEM career commitment primarily through STEM professional self-efficacy and positive perceptions of STEM professionals. In contrast, men students’ career commitment was shaped more strongly by peer interactions via perceptions of STEM professionals. Mentoring quality demonstrated similar indirect effects for both groups. Together, these findings underscore that equal access does not imply identical developmental mechanisms.

Overall, the study demonstrates that thoughtfully structured university-led STEM programs can act as effective gender equalizers in career commitment outcomes, even when women remain numerically under-represented in participation. By revealing gender-specific psychological and contextual pathways, the research extends Social Cognitive Career Theory and provides important implications for designing targeted STEM policies and support systems in higher education.

Source (Open Access): Guo, C., Wu, W., Hu, T., & Gao, T. (2025). Unequal access, equal outcomes? Gender differences in the relationship between university-led STEM program factors and undergraduates’ career commitment in STEM. International Journal of STEM Education, 12(1), 46.

https://doi.org/10.1186/s40594-025-00569-9… Read the rest

Categories
Achievement Higher Education Primary School Education Secondary School Education

The impact of test preparation on performance in large-scale educational tests: A meta-analysis of experimental studies

A recent meta-analysis examines whether test preparation improves performance in large-scale educational tests. Although schools, commercial institutions, and students invest heavily in preparation courses and materials, existing studies have reported inconsistent effects. Earlier reviews were also concentrated on US college admission and cognitive ability tests and often relied on evidence published before 2000. This study therefore estimates the overall effect of test preparation and examines how it varies across intervention, test, student, and research-design characteristics.

The authors synthesized 28 experimental and quasi-experimental studies, including 92 effect sizes and 16,741 participants across primary, secondary, and tertiary education. The studies covered college admission tests, English proficiency tests, and other large-scale assessments. A hierarchical random-effects model with robust variance estimation was used, while moderator analyses examined intervention breadth, strategy and material use, test characteristics, previous test experience, and study design.

Test preparation produced a significant positive overall effect on test performance (g=.26, 95% CI [.10,.42], p<.001), but heterogeneity was high (I^2=82.44%). Test-specific and narrowly focused interventions were more effective (g=.33) and (g=.32) than broad programs aimed at general knowledge or transferable skills (g=-.04). Interventions teaching test-taking strategies also showed stronger effects (g=.35) than those without such instruction (g=.05). Students with previous testing experience benefited substantially more (g=.69) than those without it (g=.06). However, preparation effects did not transfer significantly to other tests in the same domain (g=.06, p=.502).

Taken together, the findings suggest that test preparation can improve large-scale test scores, but its effectiveness mainly reflects alignment with the target test rather than broader learning gains. The limited transfer effect raises concerns about whether score improvement represents genuine development in the abilities being assessed. Future research should use stronger experimental designs, report intervention processes more clearly, and distinguish test-score gains from transferable learning.

Source (Open Access): Hao, Z., Baird, J. A., Masri, Y. E., & Double, K. (2025). The impact of test preparation on performance of large-scale educational tests: A meta-analysis of experimental studies. Review of Educational Research, 00346543251360775.

https://doi.org/10.3102/00346543251360775… Read the rest