Effectiveness of Problem-Based Learning Model in Science Learning: A Meta-Analysis Study

This study aims to determine the effect and impact of the overall research on the Problem-based learning model in science learning. This type of research is a meta-analysis. The data sources in this study are 17 national and international journals published from 2017-2022. The process of searching for data sources through Google Scholar, ScienceDirect, Wiley, ProQuest, and Eric Journal. Inclusion criteria are research on problem-based learning models with experimental or quasi-experimental methods and measurement of learning outcomes to evaluate Problem-Based Learning learning models. The results showed that the average effect size value of all studies (ES = 1.40) was very high. This finding explains that the Problem-Based Learning model provides a very high positive impact on science learning. In addition, the Problem-Based Learning learning model is effective to be applied to students' science learning at school. Effect measurement in this study is influenced by the level of education, year of publication, learning outcomes, and sample size. Overall, the Problem-Based Learning model is very useful in increasing students' potential in facing the 21st century. the control class is 70.31.


INTRODUCTION
Science learning is a compulsory subject that must be mastered by students in facing the 21st century (Ulfa et al., 2017;Elfira et al., 2023;Razak et al.., 2021;Fradila et al., 2021;Oktarina et al., 2021).Science learning is very important for students to stimulate critical thinking skills (Rahman et al., 2023;Suharyat et al., 2022;Zulkifli et al., 2022).According to Rahayu et al., (2012) Science learning is a subject that studies phenomena about living and nonliving things.In addition, science learning leads students to study themselves with the surrounding nature (Listyawati, 2016).In learning science, students must be able to develop scientific and systematic thinking skills in solving a problem (Suendarti & Virgana, 2022;Rahman et al., 2023;Suharyat et al., 2023;Santosa et al., 2021), However, students experience a lot of control in science learning.The results of the 2018 PISA survey in (Sofianora et al., 2023;Ichsan, et al., 2022);Supriyadi et al., 2023;Karim, 2023) shows that Indonesian students' science learning is still low, only obtaining a score of 396, ranked 71 out of 78 members.The process of learning science at school still tends to be passive (Kusnandar, 2019;Aydede, 2022), so that students are less interested in learning (Ferdyan et al., 2021).Taupik & Fitria (2021) stated that the teaching and learning process has not included students to be active in learning.Students find it difficult to understand the science learning material provided by the teacher (Putri et al., 2018;Santosa et al., 2021;Ichsan et al., 2022;Santosa & Yulianti, 2020), The learning model is still conventional and the learning process has not directed students to think scientifically (Ejin, 2017;Zulyusri et al., 2022;Santosa, 2021).Therefore, there is a need for a learning model that can improve the student learning process.Problem-based learning is a learning model that involves students more actively in learning to solve a problem (Bayram & Deveci, 2022;Şenyiğit & Yüzüncü, 2021;Suharya et al., 2022;Putra et al., 2023;Murdiyah et al., 2020;Alfares, 2021).Problem-based learning model is a learning model that presents students in a problem then directed to students to solve the problem (Paradina et al., 2019;Mustofa & Hidayah, 2020;Suhirman & Yusuf, 2019).Problem-based learning model can increase students' independence in learning (Dewi et al., 2013;Zulyu sri et al., 2022;Theabthueng et al., 2022).Research results Kasuga et al., (2022) stated that the problem -based learning model encourages students to think critically and creatively in learning.In addition, problem -based learning students not only understand concepts but focus on solving a problem ( Phasa, 2020;Çeliker & Dere, 2022;Marthaliakirana et al., 2022).Previous research by Yulianingtias et al., (2016) stated that the problem-based learning model can improve learning outcomes and problem-solving skills.Research results Amin et al., (2020) The problem-based learning model can improve students' critical thinking skills and motivation in learning.Students who have critical thinking skills will find it easier to understand concepts and learning content (Rahman et al., 2023).Research results Montejo, ( 2019) problem based learning dapat mendorong kecerdasan emosional siswa dalam belajar sehingga siswa lebih percaya diri.Research oleh Saputro et al., (2020) Problem based learning can increase stude nts' confidence and critical thinking.therefore, the problem based learning model is one of the solutions to improve the quality of student learning.Based on the above problems, this study aims to determine the effect and impact of the whole research of Problem based learning model in science learning.

RESEARCH METHODS
This type of research is meta-analysis research.Meta-analysis is a type of research that reviews previous studies that can be statistically analyzed (Santosa et al., 2021;Suharyat , et al., 2022;Yang et al., 2013;Wang & Wang, 2020).The data sources in this study came from 17 national and international journals published from 2017 -2022.The process of searching for data sources through Google scholar, Eric, ScienceDirect, Wiley, and Taylor of Francis with the PRISMA method.According to Aslikhah Nurkamto ( 2019) The steps of meta-analysis in this study are 1) Determine and summarize the research topics to be studied; 2) collect research results according to the research topic; 3) Determine the effect size value of each article; 4) Draw meta -analysis conclusions.
Furthermore, the keywords for searching data sources are "problem-based learning", "science learning"; "The Effect of Problem Based Learning on Science Learning".Data analysis in this research is quantitative statistical analysis with JSAP application.In this research, meta-analysis calculates Effect Size (ES), Standard Deviation (SD) and mean value.Furthermore, the Effect Size criteria can be seen in (Table 1.).2. Shows that the average value of Effect Size (ES = 0.828) with high criteria.This explains that the problem-based learning model has a significant effect on student learning activities.The next stage, determining the effect size model by conducting a heterogeneity test.The results of the heterogeneity test can be seen in Table 3.
Based on Table 3.Showing the value of the heterogeneity test (Q = 52.20;p = 0.00 <0.05), the effect size in the study is heterogeneously distributed.These results explain the meta-analysis model used in this study is a random effect model.The average effect size value is 0.828.This finding is analyzed based on Cohen's framework in (Table .1),then the Problem Based Learning learning model has a positive impact on students' science learning activities with high criteria.Furthermore, it calculates the publication bias by using the Funnel Plot method.Funnel Plot analysis can be seen in Figure 1 Model  1. Shows the results of analysis with the funnel plot method from 12 primary studies analyzed in the meta-analysis showing symmetrical effect size data, so it has a small publication bias.Next, conduct the Rosenthal Fail-Safe N (FSN) test to determine the possibility of publication bias.The results of the Rosenthal Fail-Safe N (FSN) test can be seen in Table 4.  3. Shows that the Rosenthal Fail-Safe N (FSN) value is 238, then 238 (5.17 + 10) = 2.50 > 1 means that the research in the meta-analysis is resistant to publication bias.The next step is to calculate the p-value to test the hypothesis.This is to determine Problem Based Learning has a positive impact on students' overall science learning activities based on random effect models.The results of the overall analysis based on random effect models can be seen in Table 5.Based on Table 5.The overall effect size value (ES = 0.837) with high criteria.Furthermore, the z value = 6.327 with p-value = 0.000 <0.5, meaning that the application of Problem Based Learning is very eff ective to improve students' science learning outcomes than conventional learning classes.

Discussion
The application of the problem-based learning model has a significant impact on student learning activities.This can be seen from the average effect size value (ES = 0.828) high criteria.Problem-based learning can encourage critical thinking skills and student learning outcomes (Mulyanto et al., 2018;Saputro et al., 2020;Razak et al., 2022).The problem-based learning model makes students more active in the learning process so that they are more courageous in expressing their ideas (Seyhan & Türk, 2022;Sr & Ray, 2019;Fitriani, 2020).Menurut Aidoo & Ofori (2016) problem-based learning model students are more creative and innovative in solving a problem.
Teaching and learning activities at school teachers are guided to be able to apply learning models that encourage students to be more active.According to (Hastuti et al., 2016) Problem-based learning model can improve students' understanding of concepts in science learning.Science learning students are led to be more active and creative in learning (Kodariyati & Astuti, 2016, Janah & Widodo, 2013;Tiarini et al., 2019).Not only that, in learning a student must be able to think scientifically and critically in solving a problem (Tosun, 2013;Winarti et al., 2022;Festiyed et al. , 2022).Mokambu (2021) stated that the problem-based learning model can improve creative thinking skills in science learning.Problem-based learning model is very effective to be applied in science learning .This can be seen from the value (z = 6.327 or p-value <0.05), so the problem-based learning model really needs to be applied in students' science learning.Problem-based learning model can stimulate students' critical thinking in science learning (Devi & Bayu, 2020;Setiawan, 2013).Students who have critical thinking skills in science learning can apply the subject matter to their environment.(Supratman et al., 2021;Nwazota & Institute, 2018;Yustiana el al., 2022).So, the problem-based learning model is one of the models that can support students' thinking skills in learning science (Suhaimi et al., 2022;Ichsan et al., 2022).

Table 2 .
Overall Effect Size

Table 5 .
Overall analysis based on random effect models