Theory
How Technology Can Improve Student Engagement
Technology improves engagement when it changes what students are required to produce, and does nothing when it only changes what they are shown. That single distinction explains most of the contradictory evidence in this field, including why the same laptop can raise engagement in one lecture and lower comprehension in the next.
By Alae Belaich9 min read6 sourcesLast modification 13 August 2026
Does technology actually improve engagement?
Search this topic and you find two literatures that appear to disagree. One says technology raises engagement; the other says devices harm learning. Both are right, about different things.
The predictive variable is not the device. It is whether the technology moves the student up the ICAP ladder, Chi and Wylie's four modes of cognitive engagement, from Passive through Active and Constructive to Interactive.
| Technology used to… | ICAP effect | Verdict |
|---|---|---|
| Show a video instead of speaking | Passive to Passive | No change |
| Distribute slides in advance | Passive to Passive | No change |
| Let students select an answer | Passive to Active | Improvement |
| Make students commit to a prediction | Passive to Constructive | Strong |
| Let students explain to each other and re-answer | Passive to Interactive | Strongest |
| Nothing: an open laptop with no task | Passive, plus distraction | Negative |
A polling app asking "is everyone following?" produces no student output and sits in row one. The same app asking a diagnostic question with a defensible wrong answer sits in row four. The tool is identical; the engagement effect is not.
What does the evidence support?
Response systems are good for engagement and weak for grades. The most relevant meta-analysis is Hunsu, Adesope and Bayly (2016), covering 111 effect sizes from 53 studies and more than 26,000 participants:
- A small effect on cognitive learning outcomes.
- A near-medium effect on non-cognitive outcomes: engagement, attention, participation.
That result is usually reported as the first half of a sentence. Reported honestly, it is still a good argument: it says these tools reliably do the thing they are named for, and do not by themselves move exam scores.
A 2025 systematic review in Humanities and Social Sciences Communications adds the caveat that should follow every claim in this area: none of the included studies used delayed post-tests. Everything was measured immediately after the intervention, so nobody knows whether the effect survives to an exam.
Interpolated questions are the strongest single mechanism. Szpunar, Khan and Schacter (2013, PNAS) inserted brief tests through a 21-minute video lecture. Students tested at intervals were half as likely to report mind-wandering, took three times as many notes, and retained more.
This is the clearest demonstration that the effective ingredient is the demand, not the medium. Caveat: video lecture, laboratory setting.
Why does anonymity matter more than it looks?
The participation research is consistent that the barrier to speaking is fear of peer judgement, and over half of students in large science courses report never asking or answering a question all semester (Cooper et al., 2021). Technology that removes the identification removes the barrier the research actually names. This is the mechanism by which a tool can change who participates, which is different from changing how much.
What does the evidence warn about?
The second-hand distraction effect. Sana, Weston and Cepeda (2013) is the study every "technology in the classroom" article should cite and almost none do. Students multitasking on laptops during a lecture scored lower on comprehension and took poorer notes, which is unsurprising. The finding that matters is that students merely sitting in direct view of a multitasking peer also scored lower.
The cost of an unmanaged device is not confined to its owner. This is the strongest available argument against the two usual policies:
- Banning devices ignores that the same hardware is the cheapest response system available.
- Ignoring devices produces exactly the condition Sana et al. measured.
The third option is to give the device a job. A phone being used to answer a question is not available to be a distraction, and its owner is producing something.
Engagement is not learning. Engagement metrics (clicks, responses, time-on-task) are proxies. Hunsu et al.'s split is the honest summary: the tools move the proxy considerably more than they move the outcome. Treat a dashboard number as a signal about your teaching, not as evidence of learning.
How should you test a tool before adopting it?
Four questions. A tool that fails the first is not an engagement tool.
- What does the student produce that they weren't given? If the answer is "nothing", it will not improve engagement.
- Does it reach everyone, or volunteers? A tool that collects answers from the confident third measures confidence.
- Is it anonymous? If not, the students the research identifies as silent will stay silent.
- Do I see the result in time to act on it? Data arriving after the lecture improves next year's lecture, not this one.
What the evidence does not show
- That technology causes learning gains. The evidence supports the design; the tool is a delivery mechanism.
- That gains persist. Delayed post-tests are largely absent from the literature.
- That "digital natives" engage differently. The generational claim is not supported by the classroom evidence.
- That more interactivity is monotonically better. No study establishes the point of diminishing returns.
- Any figure of the form "technology increases retention by X%". Those trace to content marketing, not research.
Frequently asked questions
Does technology improve student engagement?
It improves engagement more reliably than it improves results, and the two are routinely conflated. Hunsu, Adesope and Bayly (2016), across 111 effect sizes from 53 studies and more than 26,000 participants, found a near-medium effect on engagement, attention and participation, alongside only a small effect on cognitive learning outcomes.
Does classroom technology raise exam scores?
Not reliably. The best available meta-analysis finds only a small effect on cognitive outcomes, and a 2025 systematic review notes that none of the studies it reviewed used delayed post-tests, so whether any gain survives to an exam weeks later is unmeasured. Any vendor promising higher grades is going beyond what the research supports.
Can devices in a lecture harm learning?
They can, and the effect reaches beyond the person using the device. Distraction is not confined to the student who is distracted: peers within view are affected too, which is why laptop policies are contested rather than obviously correct. Giving the device a job in the lecture addresses the cause better than banning it addresses the symptom.
Why does anonymity matter so much in classroom tools?
Because it removes the specific barrier the participation research identifies. Over half of students in large science courses never ask or answer a question across a semester, driven by fear of peer judgement (Cooper et al., 2021). A named response system collects data from the confident minority, so anonymity is not a privacy nicety here, it is what makes the measurement represent the room.
How do I tell whether a tool is worth adopting?
Ask what the student produces that they were not given. Under the ICAP framework, a reaction button produces a sentiment and sits at the bottom of the ladder, while a diagnostic question produces a committed answer and a written question produces new content. A tool that adds activity without moving the mode has added work rather than learning.