Why this gap matters now
The 2023 version of this post listed five future directions from a 2022 systematic review. They are still valid, and newer evidence makes them sharper. Immersive tools can improve skills, with the strongest effects when they support rather than replace teaching (Yang et al., 2024), but reviews of AR in education still find few studies that follow learners over time (Rus & Sita, 2025).
Not every study finds an effect. An AR experiment on environmental awareness found none (Attanasi et al., 2025), and a large multilevel study found that what students do matters more than the device they use (Wekerle et al., 2024). These results point to design, not technology, as the key research question.
Five directions, updated
| Direction (2022 review) | What newer evidence shows | What a doctoral study can add |
|---|---|---|
| Longitudinal studies | Few AR evaluations repeated over months (Rus & Sita, 2025) | Delayed tests of retention and transfer |
| Diverse populations | Effects depend on school capacity (Timotheou et al., 2023) | Studies in under-represented settings |
| Beyond STEM | Heritage AR evaluations often incomplete (Chatsiopoulou & Michailidis, 2025) | Complete evaluations in humanities subjects |
| Design features | Activity type matters more than technology (Wekerle et al., 2024) | Feature-level experiments |
| Ethics | Device use linked with lower performance (OECD, 2024) | Studies of attention and wellbeing |
Illustrative example: a geography teacher-researcher compares an AR terrain lesson with a hands-on map lesson of equal length in eight classes, randomising by class. Students sit the same test a week later and again after three months. The design separates engagement from learning that lasts.
Common mistakes in AR and gamification proposals
- Comparing an AR lesson with a weak or no-treatment control.
- Measuring only enjoyment or immediate scores (Rus & Sita, 2025).
- Bundling many game features together, so the cause of any effect is unclear.
- Running a single small class study (Button et al., 2013).
- Ignoring teaching quality and classroom use (Wekerle et al., 2024).
