Penelitian ini dilatarbelakangi oleh rendahnya keterampilan proses sains siswa pada muatan IPAS yang disebabkan penggunaan model pembelajaran yang kurang bervariasi yang tidak melibatkan siswa secara aktif. Tujuan penelitian ini adalah untuk mendeskripsikan perbedaan dan peningkatan keterampilan proses sains dengan siswa yang menggunakan model discovery learning berbasis kearifan lokal di kelas eksprimen dengan model direct instruction di kelas kontrol. Penelitian ini menggunakan pendekatan kuantitatif dengan metode quasi eksprimen dan desain nonequivalent control group pretest-posttest. Subjek penelitian adalah siswa kelas IV SDN 1 Gandasoli yang terdiri dari 21 siswa kelas eksprimen dan 21 siswa kelas kontrol. Variabel bebas dalam penelitian ini adalah model discovery learning berbasis kearifan lokal dan variabel terikatnya adalah keterampilan proses sains. Teknik pengumpulan data menggunakan tes tulis berupa pretest dan posttest berbentuk essay. Teknik analisis data yang digunakan dalam penelitian ini adalah uji normalitas, uji homogenitas, uji t, dan N-Gain. Hasil penelitian ini menunjukkan bahwa nilai thitung sebsar 5,82 dan ttabel 1,68 yang berarti H1 diterima Ho ditolak. Hal ini terdapat perbedaan signifikan antara kelas eksprimen yang menggunakan model discovery learning berbasis kearifan lokal dengan kelas kontrol menggunakan model direct instruction. Selain itu, nilai N-Gain kelas eksprimen sebesar 0,76 dan kelas kontrol sebesar 0,29, yang menunjukkan bahwa model discovery learning berbasis kearifan lokal lebih efektif dalam meningkatkan keterampilan proses sains siswa dibandingkan model direct instruction.

This study was motivated by the low science process skill of students in the IPAS subject, which was caused by the use of less varied learning models that did not actively involve students. The aim of this research was to describe the differences and improvements in science process skills between students taught using the discovery learning model based on local wisdom in the experimental class and those taught using the direct instruction model in the control class. This study employed a quantitative approach with a quasi-experimental method and a nonequivalenti control group pretest-posttest design. The research subject were fourth-grade students at SDN 1 Gandasoli, consisting of 21 students in the experimental class and 21 students in the control class. The independent variable in this study was the discovery learning model based on local wisdom, while the dependent variable was science process skills. Data were collected using written tests in the form of pretests and posttests in essay format. The data analysis techniques used in this study included normality tests, homogeneity tests, t-tests, and N-Gain analysis. The results showed that the t-count value was 5,82 and the t-table value was 1,68, indicating that H1 was accepted and H0 was rejected. This meant there was a significant difference between the experimental class using the discovery learning model based on local wisdom and the control class usng the direct instruction model. Moreover, the N-Gain score in the experimental class was 0,76 while in the control class it was 0,29, indicating that the discovery learning model based on local wisdom was more effective in improving students’ science process skilss compared to the direct instruction model.