A quiz is good at one thing: telling you whether a student landed on the right answer. It is far less good at telling you how they got there, or where the reasoning came apart. By the time a wrong answer shows up on a test, the unit is usually over, and the chance to do something about it has passed.
That gap is the whole reason formative assessment exists. The point of it is not to assign a grade. It is to surface information early enough that a teacher can still act on it. The trouble is that most science formative tools, exit tickets, multiple-choice checks, quick quizzes, measure the same thing a summative test does, only sooner. They tell you a student got it wrong. They rarely tell you why.
A model shows the reasoning, not just the result
A model a student builds is a different kind of evidence. It does not collapse a student's thinking into a single letter. It preserves it: the components they decided to include, the relationships they defined between them, the prediction they made, and the change they made when the model behaved in a way they did not expect.
That is a visible trace of NGSS Practice #2, Developing and Using Models, one of the science and engineering practices that three-dimensional standards weight and that coverage alone does not build. It is also the practice traditional materials most often underbuild, because a worksheet or a pre-built simulation never asks the student to construct anything.
You can see the misconception, not just infer it
Here is what makes this useful on a Tuesday, in the middle of a unit. When a student answers a multiple-choice item wrong, you are left to guess at the cause. When a student builds a model wrong, the misconception is right there in the structure. A student who wires an inhibiting relationship as an activating one has shown you exactly where their picture of the system diverges from how it works. You are not inferring the gap from a wrong answer. You are looking at it.
That changes what you can do next. Instead of reteaching the whole topic, you can address the specific relationship the student got backwards, while there is still time in the unit to fix it.
What it looks like in practice
On ModelIt!, a student studying gene regulation or the immune system builds the network themselves, then runs it and explains why knocking out one component changes the downstream behavior. No coding, no advanced math. What you collect at the end is not a completion checkmark. It is a record of how the student reasoned, which is the thing you actually wanted to assess.
The research points the same direction. In one peer-reviewed study, undergraduates taught with computational models and dynamical simulations outperformed peers taught the traditional way (Booth et al., 2021). The reason is not the technology. It is that building a model forces the reasoning into the open, where both the student and the teacher can see it.
The better formative assessment
The most useful formative assessment is not a smaller version of the test. It is student work that makes thinking visible while there is still time to respond to it. A model a student built, predicted with, and revised is exactly that, and it doubles as evidence of the science practice the standards are asking for.