Model Two Stay Two Stray Terintegrasi Deep Learning Terhadap Keterampilan Komunikasi Ilmiah Siswa Sekolah Dasar

Abstract

The scientific communication skills of elementary school students in Indonesia are still low, reflected in the achievement of PISA 2022 scientific literacy of 383 points and the low participation of students in discussions during observations at elementary schools in Dabin 3, Jumantono District. The Two Stay Two Stray (TSTS) model has the potential to increase interaction between students, but its independent implementation has not optimally encouraged in-depth scientific communication so it was integrated with the Deep Learning approach. This study aims to determine the effect of the TSTS model with a Deep Learning approach on the scientific communication skills of fifth-grade students. Method: true experimental research with Posttest-Only Control Design involving 84 fifth-grade students (43 experimental classes, 41 control classes) from four elementary schools in Dabin 3, Jumantono District. Data were collected using a validated performance assessment sheet (Aiken's V = 0.96) and reliable (Cronbach's Alpha = 0.714), then analyzed using an independent t-test and Cohen's d, and further examined per learning syntax using SEM-PLS with SmartPLS to identify the most influential stages. Results: The posttest average of the experimental class (69.12) was higher than the control class (63.15), with a significance of 0.010 (<0.05) and a moderate effect size (d = 0.58). Implications: The TSTS model with a Deep Learning approach has been proven to be effective in improving students' scientific communication skills and is suitable as an alternative for science learning in elementary schools, with an emphasis on the stay-stray stage and presentation of results.