Ask most people when "real" science education begins and they will point somewhere around middle school, when the lab equipment gets serious and the vocabulary gets long. By that logic, the early grades are a warm-up: counting, naming, coloring the water cycle. The trouble is that the science of how children actually develop tells a different story, and the gap between when scientific thinking begins and when schools start treating it seriously is where a lot of future scientists quietly opt out.
Children are already doing science before they can spell it
Long before a child sees a textbook, they are running experiments. They drop the spoon to see if it falls again. They ask why the moon follows the car. Developmental researchers describe young children as natural theory-builders who gather evidence and construct causal explanations to make sense of the world, with the ability to reason about causal chains taking shape between roughly ages 5 and 8 (MIT Press, Constructing Science). This is not a metaphor. Causal reasoning, the ability to connect a cause to its effect, is the same cognitive move that sits at the center of every scientific model.
Interest forms just as early. A large body of work finds that children develop durable science interests before they ever enter kindergarten, and that those early interests carry forward into long-term learning trajectories (Pattison & Dierking, 2019). The early years are not the warm-up. They are when the relationship with science is set.
The window closes faster than the curriculum assumes
Here is the uncomfortable part. While capability and interest start early, confidence starts leaving early too. Research following elementary students found that children gradually lose confidence in their potential to "be scientists," even as they hold on to their belief that they can "do science," and that this erosion falls hardest on girls and students from underrepresented groups (Rhodes et al., 2019). The doing-versus-being gap matters: a child can enjoy a science activity and still quietly conclude that science is not for someone like them.
If a student decides in third or fourth grade that the lab coat belongs to other people, no amount of rigorous high school chemistry will fully undo that. The decision was made years before the curriculum got serious. This is why waiting until middle school to introduce authentic scientific practice is not a neutral choice. It is a choice to start after the most formative window has already begun to close.
What "modeling" looks like for a seven-year-old
The objection writes itself: computational modeling sounds like a high school topic at best. But the National Academy of young-child research is clear that children this age are capable STEM learners, not just future ones (NAEYC, 2017). And the Next Generation Science Standards already place Developing and Using Models in the earliest grades, expecting kindergartners to start with concrete drawings and physical models and build toward more abstract representations of relationships as they grow (NGSS, Developing and Using Models).
Modeling for a young student is not equations. It is naming the parts of a system and the relationships between them. What does a plant need, and what happens to the plant if one of those things changes? A child who drags a "sunlight" piece and connects it to a "plant grows" piece, then removes it and watches the outcome change, is doing the structural work of a systems model. No coding, no algebra, no dense reading required. The same drag-and-connect logic a researcher uses, scaled to a developing mind.
That continuity is the point. When a second grader and a Nobel laureate are doing structurally similar work, building a representation of a system to test how it behaves, the early version stops being a cute precursor and becomes the actual on-ramp.
What this means for teachers and leaders
For an elementary teacher, the reassurance is that you do not need to be a computational scientist to start. You need a way to let students externalize the cause-and-effect reasoning they are already doing in their heads, and to do it before the "science isn't for me" story sets in. Building simple models gives that thinking somewhere to live.
For a district leader, early modeling is one of the few interventions that serves both the achievement agenda and the equity agenda at the moment they are most moveable. The students most likely to lose science confidence in elementary school are the same students districts work hardest to retain in STEM pathways later. Reaching them in second grade, not ninth, is the difference between widening a pipeline and trying to refill it after it has already narrowed.
STEM does not start in middle school. It starts the first time a child asks why and gets to build the answer. The question is whether the classroom is ready that early.
ModelIt! is a K-12 computational modeling platform built on Cell Collective, the NSF and NIH backed research engine behind 130+ peer-reviewed publications. It is aligned to NGSS, available in English and Spanish, and requires no coding or advanced math.