Game-Based Learning: Does Learning by Gaming Work?

Learning by gaming works, but modestly and only under the right conditions. Reviews of game-based learning generally report learning advantages over conventional teaching, but study quality is often low and results depend on design. The game label alone decides nothing.

Game-based learning means the learning happens inside a game: the learner plays, and the play is the lesson. This post sorts the evidence for adult and workplace learners and separates the approach from gamification. It also gives you a decision table, a design checklist and a tracking plan for your LMS.

Every study named below is a published paper or review, and where evidence is weak or mixed the post says so. The decision table and the design checklist are our own heuristics, not research findings.

What Is Game-Based Learning?

Game-based learning is an instructional approach in which learners build skills by playing a game designed around specific learning outcomes. The game is the lesson, not a wrapper around it. Prensky (2007), as quoted on the University of Zurich’s game-based learning page, calls it “any marriage of educational content and computer games.”

Board games, card games and facilitated games count too, not only digital ones. In this post a game has rules, goals and visible progress toward those goals. A simulation models a real task or system and may have none of those features. Many serious games are both.

How Is Game-Based Learning Different From Gamification?

Game-based learning builds a complete game and teaches through play. Gamification adds game elements such as points, badges or leaderboards to learning that is not otherwise a game. Sailer and Homner describe it as modifying an existing learning process, not creating a standalone game product.

A practical test: remove the points and badges. If the lesson still works unchanged, it was gamification. Remove the game from a game-based lesson and nothing is left to learn.

The gamification evidence is also small to moderate. A 2019 meta-analysis by Sailer and Homner found g = 0.49 for cognitive outcomes (19 studies), 0.36 for motivational outcomes and 0.25 for behavioral outcomes. In the subset of highest-rigor studies, only the cognitive effect held (g = 0.42). For platform features such as points, levels and leaderboards, see our guide to gamification in LMS.

Does Learning by Gaming Actually Work?

Modestly, and the evidence is uneven. Reviews of games for learning generally report advantages over conventional instruction, but study quality is often low and results depend on design. Treat a game as a design option to test, not a guaranteed upgrade.

The sources agree on direction more than on strength. Gentry and colleagues rate their evidence low or very low. Richard Mayer’s 2019 review of computer games in education, in the Annual Review of Psychology, is similarly cautious. Games may beat conventional media in science, mathematics and second-language learning, but he calls the findings preliminary. For a business case, assume a modest gain at best and pilot before you scale.

What Do Single Workplace and Professional Trials Show?

Trials are small and short-term. In a 2013 randomized trial, medical students who played the Uro-Island adventure game outscored students who read a script (Cohen’s d = 0.71). A 2017 aerospace trial with 20 employees found a gain in procedural knowledge but not factual knowledge.

The Freiburg trial by Boeker and colleagues randomized 145 third-year students. The game group scored 28.6 on a 34-item test against 26.0 for the script group. The authors did not record time on task or measure retention, so extra study time could explain part of the gain.

In the aerospace study by Li and colleagues, a game was compared with a paper manual for learning to assemble an aircraft maintenance door. Procedural scores averaged 26.3 against 12.6, with only 20 participants in all. Participants also flagged a steep learning curve and dense text. Neither trial measured on-the-job behavior.

What Makes a Learning Game Work?

Games teach best when learners act rather than watch and the core mechanic is the skill being learned. Play should also be repeated and followed by instruction. Across the research, results depend heavily on design, not only on the decision to use a game.

These features have the strongest support:

• Intrinsic integration: In two studies of children aged 7 to 11, the version with math as the core mechanic taught more in a fixed time. Children also played it 7 times longer when free to choose.
• Scaffolding: The Cai meta-analysis above found that hints and guidance inside digital games raised achievement.
• Value-added features: Mayer’s review identifies modality, personalization, pretraining, coaching and self-explanation as promising additions to a base game.
• Repeated, group and supported play: Wouters found larger gains when games were supplemented with other instruction, used across several sessions or played in groups.

Two cautions apply. The Wouters comparisons are subgroup splits inside observational data, not experiments. And the intrinsic integration studies used children, so treat them as a design signal for adults, not proof.

Why Do Learning Games Fail?

Learning games fail when the game is separate from the skill, learners watch instead of act, play happens once, or no one debriefs or measures learning. Adult audiences add another risk: games that look childish. A games consultant reported in 2014 that adults were less willing to play such products when their jobs depended on the skill.

Check for these failure modes before you commit budget.

• A quiz in a costume: the game is a wrapper and the real task is answering questions. That is the opposite of intrinsic integration.
• One-and-done play: Wouters found larger gains across several sessions than in single-session studies.
• Novelty doing the work: the aerospace authors flagged a possible novelty effect on engagement.
• Engagement mistaken for learning: Wouters found games were not more motivating, and the Freiburg trial did not record time on task.
• Tone that insults adults: Sande Chen wrote in Game Developer that adults “didn’t want to be seen playing an edutainment-like product.” That is one practitioner’s account, not a measured effect.
• In-game score treated as proof of transfer: a score shows performance inside the game. Without a separate test or job measure, it says little about the workplace.

When Should You Choose Game-Based Learning Over Gamification, Simulations or Scenarios?

Choose game-based learning when the skill involves repeated decisions with consequences and you can afford to build and test a game. Choose gamification to add motivation to content that already teaches well. Choose a simulation to practice a realistic task, and scenario-based learning to build judgment at lower cost.

The table below is our own design heuristic, not a research finding. The effort column is a judgment: games need design, build and playtesting, while gamification reuses content you already have.

Scenario complexity runs from a single-decision mini scenario to an AI role-play simulation, a scale described in Devlin Peck’s scenario-based learning guide. If your real question is whether a game should replace or support an instructor, our comparison of game based learning vs ILT covers that split.

How Do You Design a Learning Game Around Objectives, Bloom’s Levels and Kirkpatrick Measures?

Start from the objective, not the mechanic. Write the objective with a measurable verb. Match its Bloom’s level to a mechanic that makes learners perform that verb, and decide up front which Kirkpatrick level you can evidence. The table is our design heuristic, not a research-validated model.

Our guides to writing measurable learning objectives and Bloom’s taxonomy for corporate training cover the first two steps. For measurement, Kirkpatrick Partners defines Level 1 as reaction, Level 2 as learning, Level 3 as behavior and Level 4 as results.

A post-play survey gives Level 1 for any row, but engagement is not learning. Level 4 never comes from the game itself.

Before you build, confirm these six points:

1. The objective names an observable behavior, and the game’s core action is that behavior.
2. The mechanic that earns progress is the skill itself, not trivia about it.
3. Hints, worked examples or pretraining support novices.
4. Learners can replay, and the plan includes at least two sessions.
5. A debrief or follow-up instruction comes after play.
6. A separate test or job measure sits outside the game.

How Do You Track Game-Based Learning in an LMS With SCORM or xAPI?

In SCORM 1.2 the headline data is a lesson status and a score checked against a mastery threshold. xAPI statements can describe individual events such as decisions, errors and retries and send them to a learning record store. The ADL Initiative published a serious games profile for xAPI in 2016.

SCORM sends status through JavaScript calls, so a closed browser can stop the final status from reaching the LMS. That matters for a long game that learners may quit midway. ADL developed its serious games profile with researchers at the Complutense University of Madrid to standardize game tracking. cmi5 adds a defined LMS launch on top of xAPI.

Worth logging, in practice:

• Each decision and the option chosen.
• Error types, not just error counts.
• Hint requests and how often they were used.
• Mission or level start, completion and replay count.
• A scaled score and the session duration.

Our guide to xAPI statement design recommends registry verbs where they exist and scaled scores between -1 and 1. It also advises ISO 8601 durations and a limit on statement volume to protect storage and query speed.

Is Learning by Gaming Only for Children?

No, but much of the research sits in schools. A 2024 meta-analysis of game-based learning in early childhood, ages 3 to 8, reported positive pooled effects on cognitive, social, emotional, motivation and engagement outcomes. Adult workplace evidence is thinner, and the trials above measured results soon after training.

Age matters in other ways too. In the gamification meta-analysis, school settings showed larger cognitive effects than higher education or informal training. One practitioner account says adults see more value in simulations tied to job tasks. For parents and teachers choosing off-the-shelf games, the same design tests apply: does the play require the skill, and is there follow-up?

What Should You Do Next?

Pick one skill with repeated decisions and a measurable error pattern, such as a procedure or a triage rule. Write the objective, choose the Bloom’s level, and decide how you will test it outside the game.

Build or buy the smallest playable version and pilot it against your current training. Use random assignment if you can, since it gives the cleanest answer. Measure at two points, instrument the game with xAPI, and keep a debrief in the plan.

Decide after the pilot, not before. If the game beats your current method on the outside test, scale it. If a scenario matches it, keep the scenario.