What the research says

Do Educational Games Work? A Meta-Analysis Explained

A review of 69 studies of students aged 6 to 25 asked whether games teach more than lessons without them, and which designs work best. The answers, in plain words.

By the Giga Quizzes team4 min read
A heap of bright many-sided game dice in every colour
Photo: Tasha Kostyuk, Unsplash, Unsplash License, cropped.
In this post
  1. 1What the researchers did
  2. 2The headline numbers
  3. 3One session is not enough
  4. 4Simple looks beat flashy ones
  5. 5The limits of the evidence
  6. 6What to take into a lesson

Claims about learning games tend to come in two flavours: games will transform education, or games are a distraction dressed up as a lesson. A meta-analysis, a study that pools the results of many other studies, is one of the better ways to cut through both. In 2016 Douglas Clark, Emily Tanner-Smith and Stephen Killingsworth published one of the most thorough on digital games and learning. Here is what they did, what they found, and what it means for a teacher deciding whether to bring a game into a lesson.

What the researchers did

The team searched for studies of digital games and learning published in peer-reviewed journals between 2000 and 2012. They screened 1,040 reports in full and kept those with a real comparison and enough data to measure an effect. The final set was 69 study samples from 68 articles, covering students from age 6 to 25. The typical participant was 12 or 13 years old, in about the seventh grade.

They asked two different questions, and kept them apart:

  • Games against no game. Do students who learn with a game do better than students taught the same material another way?
  • Better game against standard game. When researchers improved a game’s design, for example by adding hints that adapt to the player, did students learn more than with the plain version?

The headline numbers

Both answers were yes. Students in game conditions outperformed students in non-game conditions, with an average effect size of 0.33. Improved game designs beat their standard versions by a similar margin, 0.34.

Effect sizes are easier to read with a translation. An effect of 0.33 means that the average student in the game group scored higher than roughly 63% of the students in the comparison group. That is a modest but real advantage. It is not a revolution, and it is not nothing.

The second number may matter more for anyone who writes their own activities. The one kind of improvement studied often enough to measure on its own was what the authors call enhanced scaffolding: personalised support, games that adapt to what the student needs, and game structures rebuilt to put the learning at the centre. It produced a gain of similar size. Design choices, in other words, made about as much difference as the decision to use a game at all.

One session is not enough

The clearest single finding in the paper concerns repetition. When students played a game in just one session, the effect over non-game teaching was 0.08, too small to be distinguished from zero. When they played over several sessions, it was 0.44.

How students playedAverage effect on learning
A single session0.08 (not significant)
More than one session0.44
Games compared with non-game teaching, Clark, Tanner-Smith & Killingsworth (2016)

The typical game intervention in the review ran for 11 sessions over 38 days, about 347 minutes in all. Interestingly, the total number of minutes played did not predict the size of the effect. What seemed to count was coming back to the game more than once, not marathon sessions.

For a teacher, this is the most practical message in the study. A game played once as a treat at the end of term is unlikely to teach much on its own. The same game used as a regular routine, returning to the material over several weeks, is where the evidence points.

Simple looks beat flashy ones

The researchers also coded how realistic each game looked. Games with schematic, symbolic or text-based visuals had an average effect of 0.48. Cartoon-style games came in at 0.32. Realistic, high-detail games averaged −0.01, no better than teaching without a game.

The authors are careful about this one: once they controlled for other features of the games, the gap between simple and realistic visuals shrank to borderline significance. Still, it is a useful corrective to the idea that a learning game needs a big budget. Clear visuals that keep attention on the question seem to do at least as well as detailed worlds that compete for it.

One finding went against earlier reviews. Some previous meta-analyses had found that games work better when the teacher adds non-game instruction around them. In this review, games with and without extra teaching produced similar effects. That is not a licence to skip the teaching; it suggests that the design of the game itself carried much of the weight in these studies.

The limits of the evidence

The studies all date from 2000 to 2012, before phones were common in classrooms, and the games ranged from simple drills to full simulations. The authors also applied strict quality filters to see how much of the evidence was rock solid, and very few studies met them all: only four of the game-against-no-game comparisons passed every study-design filter. The overall picture is consistent, but individual numbers, especially for the smaller categories, should be read as directions rather than precise measurements.

What to take into a lesson

Three habits the evidence supports

  • Repeat it. Plan a game as a routine across several lessons, not a one-off.
  • Keep it clear. Plain, readable screens that keep attention on the content are enough.
  • Adapt where you can. Hints, easier and harder versions, or a second chance after a wrong answer are the kind of support that improved games in the review.

Quiz games fit these habits naturally. A short quiz played every week on the topic in hand is exactly the multi-session pattern the review rewards, and questions on a clean screen are about as schematic as a game gets. If you want the evidence on the competitive side of classroom games, the companion posts on gamification and competition look at what helps and what to watch.

Sources

  1. Clark, D. B., Tanner-Smith, E. E. & Killingsworth, S. S. (2016). Digital games, design, and learning: A systematic review and meta-analysis. Review of Educational Research, 86(1), 79–122.

More from the blog

All posts →