Interdisciplinary thinking is critical for equipping students to apply scientific knowledge and tackle societal challenges across various disciplines, which has been recognized as a key objective of twenty-first century science education. However, research on effective interdisciplinary assessment in secondary school science education is still limited. The purpose of this study was to develop and validate an instrument for evaluating seventh graders' interdisciplinary thinking within lower-secondary science contexts. A four-dimensional framework for evaluating interdisciplinary thinking was proposed, leading to the development of an assessment instrument. Participants were 316 seventh--grade students randomly selected from a lower-secondary school in Jiangsu, China. The multidimensional random coefficients multinomial logit (MRCML) model was employed to examine the reliability and validity of the instrument. The results indicated that the four--dimensional conceptual framework of interdisciplinary thinking was appropriate, and the multidimensional partial credit model (PCM) fitted the data reasonably well. The differential item functioning (DIF) analysis also revealed the absence of gender bias, confirming the measurement invariance statistics reached. In addition, disparities in student performance across the four dimensions were identified, highlighting the need for tailored instructional practices to foster their ability in lower-secondary school science education.