Active learning beats lecturing alone

In one line: when university science courses got students doing things in class, rather than just listening to lectures, students did better in exams and fewer of them failed.

What they did

This isn’t a single experiment. Scott Freeman and six colleagues in the USA gathered up 225 earlier studies and combined their results. Each study compared the same kind of university course taught two ways: by lecture alone, or with “active learning” added.

Active learning meant anything that got students doing something in class. That might be a quick question to discuss with the person next to you, a group problem, or a whole course built around cases. All the courses were in science, technology, engineering or maths, for students working towards a first degree.

They looked at two things: exam scores, and how many students failed or dropped out of the course.

What they found

• Exam scores went up. On average, students in courses with active learning scored about 6% higher. The authors put this at about half a grade on a US-style grading scale.
• Fewer students failed. In lecture-only courses, about 34% of students failed or dropped out. With active learning, it was about 22%. Put the other way, students in lecture-only courses were about 1.5 times more likely to fail.
• It held up across subjects. The benefit didn’t differ much between subjects, or between first-year and later courses.
• It helped in classes of all sizes. It was biggest in classes of 50 or fewer.

How much should you trust it?

Strong evidence for university science students; not tested with healthcare staff.

• It combines a large number of studies, so the overall pattern is solid.
• These were science students at university, not healthcare staff learning at work. It’s a good reason to try active learning in healthcare, but we can’t assume the same size of benefit.
• “Active learning” covered many different methods. The paper tells us that adding some activity helps on average, not which method works best.
• It measured exam results, not whether people later did anything differently in their jobs.

If you need to convince someone

In our words (a paraphrase, not a quotation): on average, across many different courses, adding active learning did better than lecturing alone, and fewer students failed.

The authors put it more strongly. In their words: “If the experiments analyzed here had been conducted as randomized controlled trials of medical interventions, they may have been stopped for benefit.” In other words, if this had been a drug trial, it might have been stopped early because one option was clearly better.

What it means for you

1. Ask what people will do, not just what you’ll say. For each part of a session, ask what learners will be doing in that minute. Start from what they need to do differently, and give them practice at it.
2. Start small. A question to discuss in pairs, a quick quiz, or a case to sort all count. You don’t need to redesign a whole course.
3. Use it in short slots too. A handover, a huddle or a few minutes on the ward can include a question for the learner to answer, rather than a talk for them to listen to.
4. Expect to justify the lecture, not the activity. If you choose to only talk, have a reason.

How it appears in Teaching That Lands

Session 2 is called Active Beats Passive. After participants sort scenario cards by learning strategy, Freeman et al. is cited as evidence that active learning beats lecturing. The caveat is stated plainly: these were science students at university, not healthcare staff. Participants then make ten minutes of their own session active.

The small print

• “Meta-analysis” means a study that combines the results of many earlier studies to get an overall answer.
• The average benefit on exams was an “effect size” of 0.47. An effect size is a standard way to compare results across studies; 0.47 is usually called moderate.
• The failure figures were 33.8% and 21.8%. The odds ratio (another way of comparing the two) was 1.95.
• Publication bias is always a risk in meta-analyses: studies that find nothing are less likely to be published. The authors tested for it and found it unlikely to explain the results, but it can’t be ruled out completely.
• In this library’s terms, exam results are evidence of learning, not of changed behaviour at work. See Kirkpatrick’s model.
• For the strategies that tend to make activities work, see retrieval practice, spaced practice and interleaving.