<![CDATA[Abstract This research is motivated by the low problem-solving ability of students and the dominance of low-level cognitive instruments (LOTS) compiled by teachers, which often fail to accurately measure high reasoning power. The purpose of this study is to analyze the empirical validity of problem-solving-based science assessment instruments on electrical, magnetic, and alternative energy materials. The research method used was quantitative descriptive involving 32 students in grade IX. Data were collected through multiple-choice test instruments and analyzed using the Anates 4.0.9 program and SPSS to measure validity, reliability, difficulty level (ITK), and differentiating power (IDB). The results showed that 65% of the questions were declared empirically valid with a very high reliability coefficient of 0.870. The level of difficulty was dominated by the medium category (60%), but it was found that 35% of the questions had poor discriminating power. Qualitative findings from expert validators also reveal gaps in the synchronization between indicators and image stimuli in magnetic matter. It can be concluded that this instrument meets the requirements for strong measurement regularity, but requires improvements in the synchronization aspect of contextual stimuli to measure high-level thinking skills (HOTS). Overall, 55% of the questions are suitable for immediate use, 30% need revision, and 15% must be discarded. Keywords: Empirical Validity, Science Assessment Instruments, Problem Solving, Question Item Analysis ]]>
