We Gave Children More Choice in a Math App. Here's What Changed (and what didn't)

What co-designing an educational game taught us about agency, learning, and designing for children

Children today have more control over their media than perhaps any previous generation. They can choose which apps to open, which worlds to explore, what their characters look like, and if they want to keep playing. However, when they open many educational games, much of that control disappears.

This is a particular design challenge for educational game creators. Popular games for children like Minecraft or Roblox can give children expansive worlds to explore, create, and play in, but many specifically educational games have curriculum to deliver.

Giving children complete freedom may make a game more engaging, but it can also make it harder to ensure that they encounter the material they are supposed to learn. Thus, our team wanted to know how much does giving children choice actually matter for their experience? And does it matter for learning?

To answer those questions, we needed to design a game that held everything else constant. If children with more choices also received different educational content, we wouldn't know whether their choices were responsible for any differences in learning.

So our research team and designers built two versions of the same game about fractions: one that gave children more opportunities to make choices and one that kept those options limited. The educational content stayed the same.

The constraint of keeping the educational content the same was unusually important fo5r our research. We were not trying to build the most exciting game possible. In fact, if we added too many exciting activities or rewards, we would not longer be able to identify the effect choice itself was having. The challenge was to create choices that felt meaningful to children without changing the actual math lesson.

Before building the digital game, we started with something simpler: a whiteboard and a set of floral magnets. With preschoolers at the UC Davis Early Childhood Lab School, we first showed children how we imagined the game being played.l Then we turned the question back to them and asked them to show us their rules. What would they do with the flowers? How would they play if they were in charge?

The answers were not always as we expected. Children invented their own plays of arranging and playing with the flowers, and sometimes the smallest features of the prototype became part of the fun. Children especially loved the sound of the magnets snapping onto the whiteboard. These early sessions helped us think more carefully about what "choice" actually looks like from a child's perspective.

As we moved from a physical prototype to a digital one, that input continued to shape the world around the educational content. Staff at the SMUD Museum of Science and Curiosity helped us develop ideas for characters with different pets and hobbies, features that became very popular with children during testing. Those details gave children things to notice, prefer, and actually choose without changing the math content itself.

Later rounds of testing were less about reinventing the game and more about making it work smoothly. Children helped us see that the game needed to be shorter and that transitions between activities needed to move faster. Some of our pre-teen testers also enthusiastically took on the role of "video game design expert" pointing out glitches and telling us exactly what they thought we should fix.

The kind of co-design was not simply a matter of asking children what they wanted and building just that; it was an iterative process of watching how they actually used the game, identifying where our assumptions did not match their behavior, and deciding which changes we could feasibly make without compromising the research questions.

Once the game was ready, we tested whether those design choices changed how children experienced playing it. Children who had more choices did, in fact, feel more in charge while they played. But they did not necessarily learn more about fractions as a result.

That finding surprised us. We had expected that feeling more in control might make children more engaged with the game, and, in turn, support learning. Instead, our results suggest that agency and learning are related design goals, but they are not interchangeable.

For designers, the takeaway is that choice is not a magic ingredient. Giving children more control can change how an experience feels without necessarily changing what they learn.

The bigger effects may also take time. One of the most important choices children make is whether they want to come back. Future research will test whether greater agency encourages repeated play (and whether those repeated experiences ultimately support learning).

The best curriculum in the world has limited value if children do not want to engage with it. That is why we bring children into the design process early; their preferences, frustrations, and unexpected ways of playing are evidence about how a product is actually being experienced, and how we can make it better.