<![CDATA[Penelitian pengembangan ini bertujuan menghasilkan perangkat pembelajaran STEM berbasis budaya Perahu Arumbae pada materi Teorema Pythagoras yang valid, praktis, dan efektif untuk meningkatkan kemampuan berpikir komputasional siswa. Model ADDIE (Analysis, Design, Development, Implementation, Evaluation) digunakan dengan melibatkan 44 siswa kelas VIII di sebuah SMP di Maluku Tengah sebagai sampel uji coba. Instrumen penelitian meliputi lembar validasi, lembar observasi keterlaksanaan, dan tes berpikir komputasional. Hasil validasi ahli menunjukkan perangkat sangat valid (skor 3,59/4,00). Pada uji lapangan, perangkat terbukti sangat praktis (keterlaksanaan 92,1%; respons siswa 4,35/5,00). Analisis N-Gain dan uji-t independen membuktikan peningkatan kemampuan CT kelompok eksperimen signifikan lebih tinggi (N-Gain=0,695; kategori sedang) dibanding kelompok kontrol (N-Gain=0,27; kategori rendah) yang dengan yang dikonfirmasi lebih lanjut dengan hasil uji paired sample t-test yang signifikan (sig. <0,001). Disimpulkan bahwa perangkat yang dikembangkan valid, praktis, dan efektif untuk meningkatkan kemampuan CT siswa, sekaligus menawarkan model pembelajaran kontekstual yang inovatif bagi guru. This development research aims to produce a culturally-based STEM learning device centered on the Arumbae Boat for the Pythagorean Theorem topic that is valid, practical, and effective in enhancing students' computational thinking skills. The ADDIE model (Analysis, Design, Development, Implementation, Evaluation) was employed, involving 44 eighth-grade students from a junior high school in Central Maluku as the trial sample. Research instruments included validation sheets, observation sheets for learning implementation, and computational thinking tests. Expert validation results indicated that the device is highly valid (score 3.59/4.00). In the field trial, the device proved to be highly practical (implementation rate 92.1%; student response score 4.35/5.00). N-Gain analysis and independent t-tests demonstrated a significantly higher improvement in computational thinking skills for the experimental group (N-Gain=0.695; medium category) compared to the control group (N-Gain=0.27; low category), which was further confirmed by a significant paired sample t-test result (sig. <0.001). It is concluded that the developed learning device is valid, practical, and effective for enhancing students' computational thinking abilities, while also offering an innovative, contextual learning model for teachers.]]>
