We talk about “student engagement” like it’s one thing. A 2025 peer-reviewed study suggests it’s at least three, and that only one of them actually showed up in students’ grades.
The study came out of King Fahd University of Petroleum & Minerals. Researchers took a real psychology course, split it into two groups, and taught one with interactive AI tools, ClassPoint and Edcafe, and the other with traditional lectures.
Same course, same content. The only difference was how students took part. Then they measured three kinds of engagement, plus test results, for both groups.
Here’s how the study was run, what it found, the limits worth keeping in mind, and how to put its main lesson, getting students doing, to work in your own classroom.
Disclosure: Edcafe was not involved in the design, funding, or review of this study. It was independently conducted and published by researchers Shoeb Saleh and Tarik N. Mohamed, with funding from King Faisal University’s Deanship of Scientific Research.
The study at a glance
The study is peer-reviewed and was published in the Research Journal in Advanced Humanities in 2025.
The setting was a real classroom with seventy undergraduates taking an introductory psychology course at King Fahd University of Petroleum & Minerals in Saudi Arabia, over a single summer term.
The researchers split them into two groups of equal size.
- The AI-tools group was taught with two interactive tools. ClassPoint ran live in-lecture quizzes and let the instructor annotate slides on the fly. Edcafe handled the interactive side outside the lecture: assignments, activities, and homework students worked through themselves.
- The traditional group took the same course, same material, taught the usual way through lectures.
That single difference, how students took part, is what the study was built to test.
To measure it, both groups filled out the same engagement survey twice: once at the start of the term and once at the end. The survey scored engagement across three dimensions, which matters later, because they didn’t all behave the same way:
- Behavioral: actually taking part and doing the work
- Cognitive: thinking things through and problem-solving
- Emotional: interest and enjoyment in the subject
Measuring before and after meant the researchers could see how much engagement changed over the term, not just where each group landed at the end. They also looked at how the two groups did on their tests.
The problem the study set out to solve
The study starts with engagement itself. The authors treat it as central to learning:
“Student engagement is a critical factor in academic success and enriching learning experiences in higher education.”
However, traditional, lecture-based teaching keeps students passive, which is the opposite of what engagement takes.
That is why the researchers turned to interactive tools. Their choice rests on the Cognitive Theory of Multimedia Learning, the idea that students take in more when content reaches them through more than one channel, not text alone. ClassPoint and Edcafe were picked as tools that put that idea into practice:
“Tools like ClassPoint and EdCafe align with CTML principles by offering quizzes, polls, and multimedia content that engage these cognitive channels.”
The catch was evidence. Both tools looked promising, but the classroom proof was thin, especially for Edcafe:
“Although EdCafe’s AI-driven approach shows promise, empirical research on its effectiveness in diverse educational contexts remains limited. Further studies are necessary to evaluate its impact on engagement and academic performance.”
So the study put them to the test: could ClassPoint and Edcafe lift engagement and grades against traditional teaching?
What the study found
The AI-tools group came out ahead on both engagement and grades. They were about 29% more engaged overall than the traditional group, and they scored higher on their tests.
But more engaged how? The study didn’t track engagement as one score. It measured three kinds, behavioral, cognitive, and emotional in connection to grades. The results weren’t the same for all three.
Note: The percentages here are ours, converted from the study's reported average scores to make the gaps easier to read.
Behavioral engagement: about 32% higher
Behavioral engagement is taking part and doing the work: answering, attempting, completing. It was one of the largest gains, the kind that shows up when students have a task in front of them rather than a lecture to sit through.
Cognitive engagement: about 36% higher
Cognitive engagement is the mental effort learning takes: working problems out, connecting ideas, and staying with difficult material instead of skimming past it. It was the biggest gap of the three.
Emotional engagement: about 27% higher
Emotional engagement is the interest, motivation, and enjoyment students feel in a course, the one most people picture when they hear “engagement.” A genuine gain, even if it was the smallest of the three.
Only one predicted higher grades: participation
This is the finding that matters most. When the researchers tested which kind of engagement actually predicted academic performance, only one did:
“Only behavioral engagement significantly predicted academic performance, while cognitive and emotional engagement had no significant effect.”
In plain terms: thinking hard and enjoying the class came with the territory, but it was the students who took part, who answered, attempted, and completed, whose grades improved.
Why participation is the engagement that moves grades
That finding runs against how we usually judge engagement. We tend to read it as a mood: are students interested, do they enjoy the class, do they look present in class. Those things are real, and they all improved in this study, but none of them predicted grades, doing the work did.
That changes what an engaging lesson should aim for. The goal isn’t a class students enjoy watching. It’s one they can’t coast through, where they have to answer a question, attempt a problem, or produce something of their own.
That is the kind of engagement the study tied to results, and it’s the part you control when you plan a lesson.
A few things to keep in mind
The authors are upfront about the study’s limits, and they’re worth holding alongside the results.
It was a small group: 70 students in one psychology course at a single university. The authors note the sample “was relatively small and lacked diversity,” so the findings may not carry to every subject, level, or setting.
It was also short. Engagement was measured across one summer term, which leaves the long-term effects unknown.
And the two tools were used together. Because the class ran ClassPoint and Edcafe side by side, the study measures their combined effect, it can’t separate how much each one contributed on its own.
Our take: a single study isn’t meant to settle the question, and it doesn’t have to. The real test is your own classroom, and the move it points to, getting students to actively take part, costs little to try.
How to get students doing the work with Edcafe
The study points to one move: get students doing, not just watching.
That’s the part Edcafe is built for. It’s an AI platform for teachers that turns a reading, video, or topic you already teach into something students work through on their own, instead of assigning a chapter and hoping it gets read.
The AI does the drafting, so it’s minutes of prep, not an evening. In practice, it’s three steps.
Start with material you already have. For a reading activity you can drop in a topic, vocabulary, text, and the AI drafts an activity that you can adjust before sharing.
Students work through it on their own. They join with a link and answer, recall, or attempt as they go.
No chasing or marking by hand. Their responses come back in one place, so you can see who took part and where they’re stuck, and spend your time on the students who need it.
Three key takeaways
- The tool isn’t the point, the doing is. The gains didn’t come from the technology itself. Edcafe made the activities quick to set up, but it was teachers who chose to put participation at the center, and that choice is what moved the needle. A tool can hand students a task, you’re the one who makes doing the work the heart of the class.
- Participation is the lever. Of the three kinds of engagement, only doing the work predicted higher grades. Interest and effort matter, but participation is what showed up in results.
- Start small. You don’t have to redesign your course. Take one reading or topic you already teach, turn it into a short activity students complete before class, and see who takes part.
Related reading: For a real-world look beyond the research, see how Dudley College put Edcafe AI to work over a six-month trial, supporting mixed-ability learners and cutting prep time.