DESIMAL: JURNAL MATEMATIKA

ABSTRACT

*Correspondence: E-mail: himatunkhoiriyah4@gmail.com Doi: 10.24042/djm.v6i2.18137This study aimed to determine the effect of using the MEA learning model on improving students' mathematical problem-solving abilities.The MEA learning model is a model that teachers can use to improve students' problem-solving abilities.This research is meta-analytic, namely by collecting data from previous researchers and then finding the effect size of all the data that has been obtained.Many of the data points found that match this research have met the criteria when searched using Google Scholar with the specified keywords.This research analysis tool uses SPSS software version 29.0 and uses the Cohen formula in data analysis.The finding of the effect size of the entire data is 1.431, which is categorized as having a considerable influence.Based on moderator variable analysis, the number of samples with  > 31 showed a larger effect size than  < 30.Furthermore, based on the level of education of junior high school students, it has an effect size that is categorized as having a large influence.This meta-analysis study shows that using the MEA learning model affects students' mathematical problem-solving abilities.

INTRODUCTION
According to the National Council of Teachers of Mathematics (Barham, 2020), one of the principles and standards for school mathematics is identifying problem-solving as one of the process standards that students need to achieve.Students learning mathematics must possess problem-solving abilities.Therefore, students' problem-solving abilities must continue to develop.Problem-solving is applying experience, knowledge, and skills to solve problems in new situations to achieve specific goals (Mariani & Susanti, 2019;Palupi, Suyitno, & Prabowo, 2016;Wahyudi & Anugraheni, 2017).
In addition, practicing problemsolving also requires specific learning methods.Difficulties in life and calculations are complicated.Solving a problem requires a thorough understanding of cause and nature, the right plan, proper execution, and the revision of the results.The ability to solve mathematical problems is a complex skill that needs to be taught carefully to students.Therefore, the learning methods used in developing mathematical problem-solving skills must also be supportive of letting students learn mathematics with the opportunity to practice problem-solving (Yapatang & Polyiem, 2022).
In MEA learning is learning using a problem-solving approach to solving a given condition.The approaches in the MEA model, according to Sahrudin (as cited in Haryanti ( 2018)), are: 1. Identify the differences between current conditions (the current state) and goals (the goal state); 2. Arrange subgoals to reduce these differences; 3. Please choose the correct operator and apply it correctly so that the subgoals that have been prepared can be achieved.

METHOD
This study uses meta-analysis to find articles that broadly and accurately relate the MEA learning model to students' mathematical problem-solving abilities.A meta-analysis combines and evaluates data from several research articles investigating and testing conceptual and hypothetical study topics.Meta-analysis is a set of statistical methods for correlating the quantitative results of multiple researchers to produce an overall summary of empirical knowledge on a given topic (Puspitasari & Airlanda, 2021).

RESULTS AND DISCUSSION
Based on search results through Google Scholar, the researcher obtained articles that were relevant to the research study sample criteria used in this metaanalysis study.The criteria in this study were that research articles had one experimental class with the MEA model and a control class using conventional or other models.The available data consisted of sample size, mean, and standard deviation.

Meet the Inclusion Criteria Processing Data
Get the Effect Size Based on Table 1, it can be seen that each study had a different effect size.The lowest effect size in this study is 0.347, which, according to Cohen, means it has a moderate effect, and the highest is 7.156, which is categorized as having a huge effect.
A heterogeneity test was carried out by examining the p-value.If  < 0.05, then the null hypothesis, which states that the effect size of each study is homogeneous, is rejected, so the estimate chosen is the random-effects model.If  > 0.05, then the null hypothesis is accepted and the fixed-effect model is used (Paloloang et al., 2020).Based on the results of the known analysis, it can be concluded that the estimation uses a random-effects model.Based on Table 2, the effect size of the entire data is 1.431 concerning Cohen, so the effect size is huge, with a standard error of 0.5253.The lowest limit of the confidence interval is 0.401, and the highest limit is 2.460, with a 95% confidence interval level.Thus, the use of the MEA learning model has a significant influence on students' mathematical problem-solving abilities.

Figure 2. Forest Plot
Another meta-analysis result is the forest plot, as shown in Figure 2. The effect size refers to the studies in this study; the smallest effect size is 0.35 (Permana, 2023), and the highest (Asih & Ramdhani, 2019) effect size is 7.16.Linguistically, the statistical results of the studies related to the effect sizes indicated that all 11 studies that made up the sample had a positive effect.In Figure 2, studies are located to the right of the no-effect line, represented by the dotted line passing through zero.All studies show benefits for the experimental group that learns with the MEA model.In Table 3, the moderator variables in this study are educational level and sample size.The results based on the educational level showed that the SMP level is the largest group, and the effect size given at this level is 0.871, so it is categorized as having a large influence.As for the SD and SMA levels, each number is one, so the effect size given cannot be ascertained.Further research is needed regarding the results of this study.
Moderator variable analysis based on sample size for sample size  < 30 is 0.991, so it is categorized as having a large influence.The sample size  > 31 is 1.780, so it is categorized as having a huge influence.
The analysis results show that the study's overall effect size is 1.431 with a huge influence category, so it can be concluded that the MEA learning model improves students' mathematical problem-solving abilities compared to conventional learning.The stages in the MEA learning model are one example of how they can hone students' abilities in solving given problems.Furthermore, if these stages continue to be applied by students, their problem-solving abilities can continue to increase (Palupi et al., 2016).
Based on an educational level, according to Paloloang et al. ( 2020) and Tamur & Juandi (2020), higher grades have higher engagement.The junior high school level significantly influences improving students' mathematical problem-solving abilities with the MEA model of learning.In contrast, further research is needed for the elementary and high school levels.
The results of the moderator variable analysis based on sample size have an effect size of  < 30, which is categorized as a significant influence.In contrast, for  > 31, it is categorized as an effect.As the results of the meta-analysis study conducted by Öksüz, Eser, & Genç (2022) and Tamur, Juandi, & Adem (2020) concluded that, "the combined effect size of the small sample group (30 or less) is significantly different from the combined effect size of the large sample group (31 or more)".At the same time, these results complement similar studies in the literature, including sample size as an analytical variable (Tumangkeng, Yusmin, & Hartoyo, 2018;Turgut & Temur, 2017).Based on these studies, the effect of a small study group on a small sample is stronger than the effect on a large sample.This is different from the findings in research of Paloloang et al. (2020); more primary studies are needed for better results in the analysis.

CONCLUSIONS AND SUGGESTIONS
The results of the analysis show that the use of the MEA learning model has an influence on improving students' mathematical problem-solving abilities and also has a positive influence on their mathematical problem-solving abilities.The results of the analysis also show that the difference in effect size is due to the sample size and education level.
When using this model, it is recommended to pay attention to the sample size and level of education so that it can work effectively.

Table 1 .
The Effect Size for Each Study

Table 1 .
Overall Result by Random-effect