<![CDATA[This research aims to compare the effectiveness of two learning models in improving students' research skills: the STEM-based science learning model with a Research Skill Development (RSD) framework and the conventional learning model. This research uses a quasi-experimental method with two groups: an experimental class taught using an RSD-based STEM approach and a control class following traditional learning. The research instrument is a research skills test given in the form of a pretest and posttest, where each test evaluates students' abilities in all stages of the research process, starting from problem formulation to data analysis. The results showed that the experimental group experienced a significant increase in research skills compared to the control group. This improvement includes aspects of students' abilities in formulating hypotheses, collecting data, analyzing results, and concluding findings based on empirical data. In contrast, students in the control class showed more minor improvements, with most still at basic skill levels. These findings provide a new contribution to applying STEM-based learning models with the RSD framework at the science education university program, especially in the context of science teaching. This research confirms that the RSD-based STEM approach is superior in developing students' research skills and is relevant for preparing them to face academic and professional challenges that require substantial research skills.]]>
