The Effect of the Deep Learning Approach on Eighth-Grade Students' Mathematical Problem-Solving Ability

Source abstract

Abstrak Kemampuan pemecahan masalah matematika merupakan salah satu kompetensi penting yang harus dimiliki siswa dalam pembelajaran matematika. Namun, siswa masih mengalami kesulitan dalam memahami masalah kontekstual, menyusun model matematika, dan menentukan strategi penyelesaian yang tepat. Penelitian ini bertujuan untuk mengetahui pengaruh pendekatan deep learning terhadap kemampuan pemecahan masalah matematika siswa. Penelitian menggunakan pendekatan kuantitatif dengan metode quasi experiment dan desain Posttest-Only Non-Equivalent Control Group Design. Populasi penelitian adalah siswa kelas VIII SMP Negeri 20 Gresik tahun ajaran 2025/2026. Sampel penelitian berjumlah 64 siswa yang terdiri atas kelas eksperimen dan kelas kontrol. Kelas eksperimen memperoleh pembelajaran menggunakan pendekatan deep learning melalui model Problem Based Learning (PBL), sedangkan kelas kontrol menggunakan pendekatan surface learning melalui model ekspositori. Data dikumpulkan menggunakan tes uraian berdasarkan indikator pemecahan masalah polya. Analisis data dilakukan melalui uji normalitas, homogenitas, dan uji Mann-Whitney. Hasil penelitian menunjukkan bahwa rata-rata nilai posttest kelas eksperimen lebih tinggi dibandingkan kelas kontrol. Hasil uji Mann-Whitney memperoleh nilai Asmp. Sig. (2-tailed) sebesar 0,000 < 0,05 yang menunjukkan bahwa pendekatan deep learning berpengaruh signifikan terhadap kemampuan pemecahan masalah matematika siswa. Nilai effect size sebesar 0,60 menunjukkan kategori pengaruh besar. Dengan demikian, pendekatan deep learning mampu meningkatkan kemampuan pemecahan masalah matematika siswa secara signifikan. Kata Kunci: Pendekatan Deep Learning, Kemampuan Pemecahan Masalah Matematika, Problem Based Learning (PBL). Abstract Mathematical problem-solving ability is a fundamental skill that students should develop in mathematics learning. Nevertheless, many students continue to experience difficulties in interpreting contextual problems, formulating mathematical models, and selecting appropriate solution strategies. This study investigated the effect of the deep learning approach on students’ mathematical problem-solving ability. A quantitative quasi-experimental method with a Posttest-Only Non-Equivalent Control Group Design was employed. The participants were 64 eighth-grade students of SMP Negeri 20 Gresik in the 2025/2026 academic year, divided equally into an experimental group and a control group. The experimental group was taught using the deep learning approach integrated with the Problem-Based Learning (PBL) model, whereas the control group received surface learning through the expository model. Data were obtained from essay tests developed according to Polya’s problem-solving indicators. Following normality and homogeneity testing, the Mann–Whitney U test was used to analyze the data. The findings indicated that the experimental group achieved a higher mean posttest score (87.82) than the control group (74.47). The Mann–Whitney U test produced an Asymp. Sig. (2-tailed) value of 0.000 (< 0.05), while the effect size of 0.60 indicated a large effect, demonstrating that the deep learning approach significantly enhanced students’ mathematical problem-solving ability. Keywords: Deep Learning Approach, Mathematical Problem-Solving Ability, Problem-Based Learning (PBL).