Assessment-First in Early Childhood Science Education: An Exploratory Practice of an Integrated “Intelligent Assessment-Intelligent Teaching-Intelligent Learning” Closed-Loop Model in a Kindergarten Junior Class

Abstract

China’s current reform of educational evaluation has drawn renewed attention to a persistent problem in early childhood science education: teaching, learning, and assessment have long functioned as separate processes. The signs are ordinary enough: activities planned from experience, inquiry that stays near the surface, judgment reserved for the end. Our study asks whether the order can be reversed—whether assessment, moved to the front, can pull teaching and learning together. The model we propose rests on two bodies of research, one on children’s preconceptions and one on deep learning. Intelligent assessment, intelligent teaching, and intelligent learning are joined in a data-driven loop that spirals upward; diagnosis opens the process and keeps informing it. We implemented the model in four steps over two months with 22 children in a junior class (ages 3–4), under the science theme “The Big Tree and Its Friends.” Digital tools supported the elicitation interviews, learning analytics, and process documentation throughout. The pre- and post-intervention evidence is encouraging. Recognition of concealed tree structures rose from 22.7% to 88.0%, and the proportion of children able to explain, in scientific terms, what plants need to grow rose from 9.1% to 90.9%, with visible gains in the quality of inquiry as well. These results suggest that moving assessment to the front of instruction can turn evaluation from a terminal verdict into navigational guidance at the very start of learning, while also helping teachers grow into data-informed, research-oriented practitioners and drawing families into closer cooperation with the kindergarten. For grassroots settings, the study offers a workable route for putting evaluation reform into practice.