Research Article
Prompts, cues, and nudges interventions in higher education: a systematic review
Synthesis: Rako, Šimić and Rienties (2026) synthesize the analytics-informed interventions that try to change what students do next: self-regulation prompts from educational psychology, attentional cues, and behavioral-economics nudges. A PRISMA-guided systematic review coded 39 experimental and quasi-experimental studies reported in 31 papers, with inter-rater agreement at Krippendorff's α = 0.82, and sorted them by intervention type, technological artifact, target outcome, and study design. Prompts were the most studied (23 of 39 studies) and reached the widest range of outcomes — self-regulation, engagement, and knowledge — while cues concentrated on cognitive and attentional processes and nudges on behavioral engagement; all three were associated with positive learning and performance outcomes. The authors identify perceptions and affective outcomes as the systematically underexplored category.
Key Findings
- Prompts dominate the evidence base. Twenty-three of the 39 studies tested prompts, against 9 for cues and 7 for nudges, and prompts also produced the widest spread of measured outcomes.
- Each intervention type has a characteristic target. Prompts were most often linked to self-regulation, engagement, and knowledge outcomes; cues to cognitive and attentional processes; nudges to behavioral engagement.
- Effects were positive but domain-specific. All three types were associated with better learning and academic performance, but which outcome improved depended on the intervention's design and goal.
- Coverage is geographically narrow. Most experimental work came from the United States, Germany, China, and Australia, with few studies from other countries or from global MOOC settings.
- Affective outcomes are the gap. Perceptions and affective outcomes were the least studied category across all three intervention types.
Why the three literatures rarely meet
Prompts, cues, and nudges do similar work with different intellectual genealogies: prompts come from educational psychology and target metacognitive monitoring, cues from cognitive psychology and target attention, nudges from behavioral economics and target choice architecture. Because they are published in different venues by different communities, they are rarely compared. This review's contribution is to put them in one coding scheme and ask what each reliably moves. The answer is that they are not interchangeable — a nudge that raises participation is not a substitute for a prompt that raises planning, and a review that pooled them would blur exactly the distinction practitioners need. That is also why the paper reports a descriptive synthesis rather than a pooled meta-analytic effect: the outcomes are not commensurable across types.
What a coding scheme can and cannot settle
The review's methodological strength is its coding discipline. Krippendorff's α of 0.82 after pilot coding of three papers is a credible agreement level, and the scheme was refined by consensus after the pilot to separate learning modality, study duration, intervention characteristics, and study design. Its limitation is the same as any classification exercise: the categories come from authors' own labels. A study that calls its intervention a "reminder" may function as a cue or a nudge depending on whether it directs attention or changes the default, and the coding has to decide. For readers, that means the type counts (23 prompts, 9 cues, 7 nudges) are a map of how researchers describe their work as much as of what they built.
The design-quality argument
The authors close on a methodological claim that matters beyond this topic: experimental and quasi-experimental studies are what will strengthen learning analytics as a field. Much of the intervention literature is correlational or single-cohort, which is why 31 papers yielded only 39 usable studies across a decade. The geographic concentration reinforces the point — most of the experimental evidence comes from four countries, so a claim about prompts improving self-regulation in higher education is really a claim about prompts in the systems that ran the trials.
What this means for practice
- Instructors. Match the intervention to the outcome you actually want: a prompt for planning and monitoring, a cue for attention, a nudge for participation — the review found no single type that moved everything.
- Learning analytics designers. Instrument perceptions and affect alongside behavioral logs, since that is where the evidence base is thinnest.
- Researchers. Report intervention dosage and the technological artifact explicitly; the review's coding depended on papers stating them, and papers that omit them cannot be synthesized.
Limitations
- The synthesis is descriptive rather than meta-analytic, so no pooled effect size is reported and cross-study magnitude cannot be compared.
- Only 39 studies met inclusion, and the type counts (23 prompts, 9 cues, 7 nudges) are small for confident type-specific conclusions.
- Publication is concentrated in four countries, limiting generalization to other systems and to MOOC contexts.
- The review rests on authors' own labels for their interventions, which may not map cleanly onto the prompt/cue/nudge taxonomy.
Citation
Rako, S., Šimić, D., & Rienties, B. (2026). Prompts, cues, and nudges interventions in higher education: A systematic review. International Journal of Educational Technology in Higher Education, 23, 62.