<![CDATA[Research needs to be conducted with the aim of significantly comparing the problem-solving skills and self-efficacy between students who follow learning with Problem Based Learning (PBL) enriched with Science, Technology, Engineering, Mathematics (STEM) and students who follow learning with a scientific lapproach to chemistry, focusing on substances and their changes. The research method is quasi-experimental quantitative research. Cluster random sampling techniques are used to determine the control group (classes applying a scientific approach) and the experimental group (classes applying the PBL-STEM model). The study uses 1 control class and 1 experimental class. Each class averages 36 students. The instruments consist of problem-solving questions and self-efficacy questionnaires validated for construct, content and empiric. Empirical validation of self-efficacy questionnaire, initial number of items = 48, Valid items = 35, Invalid items = 13. Cronbach's alpha reliability test using SPSS yielded a Cronbach's alpha coefficient of 0.779 > 0.6, indicating that the self-efficacy questionnaire is reliable. Empirical validation of problem-solving questions, initial number of items = 12, Valid items = 9, Invalid items = 3. The Cronbach's alpha coefficient is 0.644 > 0.6, indicating that the questions are reliable.The 9 prerequisite tests for MANOVA are met and followed by the MANOVA test. The research results indicate that the multivariate Hotelling's trace test with Sig 0.000 < 0.05 shows a significant difference in problem-solving skills and self-efficacy originating from different classes. The practical implication is that schools and teachers can consider adopting and integrating the PBL-STEM model into the curriculum, especially for science subjects like chemistry. Thus, teaching will not only focus on knowledge transfer but also on developing essential 21st-century skills.]]>
