So far, not enough effort has been invested in developing reliable, valid, and engaging assessments in school science, especially assessment of interdisciplinary science based on the new Next Generation Science Standards (NGSS). Furthermore, previous tools rely mostly on multiple-choice items and evaluation of student outcome is linked only to their raw scores in standardized tests. In addition, educational research on student science achievement always limits in certain grade/subject according to application of raw scores. This study provides a way to design, validate, and further improve an instrument to assess student understanding in interdisciplinary science across grade and time. Meanwhile, student learning growth is investigated. The data are collected from elementary/middle school student survey from an Interdisciplinary Science and Engineering Partnership (ISEP) in Northeastern part of the USA. The results show a generally good quality of instrument through various aspects of empirical evidence, such as dimensionality, model-data-fit, and validity. Finally, student learning growth in understanding interdisciplinary science shows a sharp increase between elementary and middle school. The study sheds light on developing more reliable and valid instruments in assessing student science understanding based on the new standards and it also provides suggestions of implementation in both educational research and practice.