<![CDATA[This study aimed to analyse differences in 21st-century skills (4C) and physics learning outcomes between students receiving Science, Technology, Engineering, and Mathematics (STEM)-based instruction and those receiving conventional instruction. A quantitative approach with a quasi-experimental, nonequivalent control group design involved 60 Grade XI students at SMA Negeri 7 Halmahera Selatan in the 2026/2027 academic year, comprising two intact classes of 30 students each selected through purposive sampling. Critical thinking, creativity, communication, and collaboration were observed by two observers over three instructional sessions using a structured observation instrument, and learning outcomes on static fluids were measured with a pretest and a posttest. Data were analysed using descriptive statistics, independent-samples t-tests, normalised gain (N-gain), and analysis of covariance (ANCOVA). The experimental group achieved a higher mean overall 4C score (M = 89.15, SD = 2.52) than the control group (M = 73.82, SD = 2.48), t(58) = 23.760, p < .001, and a higher posttest mean (M = 84.73, SD = 3.99) than the control group (M = 70.07, SD = 3.10). The difference in posttest scores remained significant after controlling for pretest scores, F(1, 57) = 249.77, p < .001. N-gain values of 0.70 and 0.401, respectively, indicated greater improvement in the experimental group. These findings suggest that STEM-based physics instruction was associated with higher 4C scores and better learning outcomes, although the quasi-experimental design limits causal inference.]]>
