Comparison of Somatic, Auditory, Visual, and Intellectual Learning Models and Snowballing Against Numerical Ability of Students

Numerical ability in learning mathematics is an important thing for students to facilitate teaching and learning process. SAVI and SNAWBALLING are learning models that can facilitate the development of numerical abilities of students. The purpose of this study is to determine the effect of SAVI and SNOWBALLING learning models on the numerical ability of students. This research is a type of research Quasy Experimental Design. The sampling technique used was class random technique with row and series material. The instrument used to collect data was a numerical ability test in the form of a description item. The data analysis technique of this study used the T-Test analysis technique. The results of this study are that the two models do not have the same numerical ability of students, meaning that there are differences in the impact given by the SAVI learning model and the SNOWBALLING learning model on numerical ability. The numerical ability of students using the SAVI learning model is better than using the SNOWBALLING learning model.

*Correspondence: E-mail: m.zainirwanto@umi.ac.id Doi: 10.24042/djm.v3i1.5690 Numerical ability in learning mathematics is an important thing for students to facilitate teaching and learning process. SAVI and Snowballing are learning models that can facilitate the development of numerical abilities of students. The purpose of this study is to determine the effect of SAVI and Snowballing learning models on the numerical ability of students. This research is a type of research Quasy Experimental Design. The sampling technique used was class random technique with row and series material. The instrument used to collect data was a numerical ability test in the form of a description item. The data analysis technique of this study used the T-Test analysis technique. The results of this study are that the two models do not have the same numerical ability of students, meaning that there are differences in the impact given by the SAVI learning model and the Snowballing learning model on numerical ability. The numerical ability of students using the SAVI learning model is better than using the Snowballing learning model.

INTRODUCTION
Given the importance of mathematics in science and technology, mathematics must be understood and mastered by students (Indrawati 2015). Factors that influence the success of students in mastering mathematics include numerical ability (Zikriah 2018). Some research results suggest that children who have high numerical ability will be able to develop new thinking by integrating a variety of basic knowledge as supporters to think more broadly, because in these students there is perseverance, tenacity, hardness of heart, interest, curiosity and creativity (Kurniyanthi , Suarni, and Gunamantha 2019). Therefore numerical ability is very needed in mathematics (A. Hamid, S. Afriza 2016).
Learning models that can be used to improve students' numerical abilities such as the SAVI learning model and the SNOWBALLING Learning Model (Tanjung and Nababan 2018). The SAVI learning model trains participants to learn by listening, speaking, and thinking (Sulakasana, Margunaasa, and Authority 2018). Thus, between physical activity and intelligence go hand in hand. Based on research on students' numerical abilities using the SCRAMBEL, TSTS, COOPERATIVE TIPESTAD, OSBORN, PAIR CHECK models (Melani, Candiasa, and Hartawan 2019;Sunilawati, Dantes, and Candiasa 2013;Lanya and Aini 2019;Irawaan 2014). The SNOWBALLING learning model is a learning model where students are trained to be responsible for discussions in stages to get the right answer, so students can exchange opinions so as to produce a maximum result and be approved (Ramadhani, R., Umam, R., Abdurrahman, A., & Syazali, M. 2019), so that this learning model can improve the numerical ability of students (Kuncoro , Purnami, and Key 2018). Based on previous research that has been done about numerical abilities using learning models SAVI, TSTS, SCRAMBL, OSBORN (Setyani, Nizardin, and Utami 2019;Melani, Candiasa, and Hartawan 2019;Sunilawati, Dantes, and Candiasa 2013;Lanya and Aini 2019) . Previous studies also included SNOWBALLING learning models for learning motivation, learning outcomes, critical thinking skills (Srisusilaningtyas 2018;Putranto, Susatyo, and Siadi 2013;Maasawet 2010 None of the previous studies discussed the two methods of learning, so this study, researchers were interested in the two learning models by focusing on the SAVI learning model and the SNOWBALLING learning model on students' numerical abilities. So this study aims to see which learning model is more effective to be used in improving numerical abilities of students.

METHOD
This research is included in quantitative research with the type of research used is quasy experimental design (Putri and Silalahi 2018). The population to be used in this study is 60 students consisting of 2 classes with 30 students each in each class selected by cluster random sampling technique (Anwar, Munzil, and Arif Hidayat 2017). The two classes were given different treatments, namely the class that was applied to the SAVI learning model and the class that was applied to the SNOWBALLING learning model. students, then the delivery of new learning material that involves the five senses and is suitable for all learning styles, then the training conducted by the teacher to help students understand and absorb new knowledge and skills in various ways and lastly is the appearance of the students (Novitasari 2017).

Figure 2. SNOWBALLING model research design
In Figure 2, we can see the steps of the SNOWBALLING learning model, which starts from conveying the topic of the lesson, asking students to solve problems in pairs, uniting one pair with another pair into a group and solving problems in the group, then uniting again group one with other groups to solve problems in groups that have been united, the group informs the results of the group to other students, finally the teacher will compare the answers from each group (Siregar, Ardiana, and Rosyidi 2019).
Data collection techniques used to evaluate and validate the data obtained is a matter of description tests, namely by giving questions that are of type description to students who will then be collected back to researchers to retrieve the test data results. The type of data used to process research data is quantitative data.
The data analysis technique used in this study was the T-Test analysis technique, with the research design used was post-only control design (Dongoran, Basri said, and Defitriani 2019). There are several stages before the test instrument is used in conducting research including, 1) arranging the material that will be used in the questions, 2) making the questions grid, 3) compiling questions and their keys and 4) conducting a cobasoaltes test (Maasawet 2010). The instrument was tested on students to find out the validity, reliability, level of difficulty, and distinguishing problems.

Unite the two couples into a group
The group completed the previous problem

Informing the results of group discussions
The teacher compares the answers of each group

RESULTS AND DISCUSSION
Based on research results obtained from research on the effect of the SAVI and SNOWBALLING models on the numerical abilities of students, the following is presented a descriptive description of the results of the SAVI and SNOWBALLING learning models on numerical abilities. Based on the results of the previous data analysis, the researcher will conduct a T test. Before the T test the researcher will conduct a normality and homogeneity test first as a prerequisite test. Normality test is used to find out whether the data is normally distributed or not, while the homogeneity test is used to find out whether the data is homogeneous or not. This prerequisite test uses a significance level α = 0.05. H0 = Data is normally distributed H1 = Data not normally distributed Conclusion if p value > α so H0 accepted and rejected H1, while if p value < α so H0 rejected and accepted H1.
Numerical ability normality test results are in Table 2. Based on Table 2, normality test results with a significant level α = 0.05 shows that the application of the SAVI learning model and the SNOWBALLING learning model to numeric abilities shows the p value of the SAVI learning model is 0.128 and the SNOWBALLING learning model is 0.115 for the Shapiro-Wilk output probability.
Because the values of p value> α, both data are normally distributed.
Homogeneity test with hypothesis H0= Homogeneous data, H1= Not homogeneous data. H0 accepted if p value > α. Homogeneity test results are in Table  3. Based on Table 3, the homogeneity test results with a significant level of 0.05 looks p value is 0.125 and greater than the significance level of 0.05, it can be concluded that the two data are homogeneous (Ramadhani, R., Huda, S., & Umam, R. 2019). After the prerequisite tests are normality tests and homogeneity tests are generated that the data is normally distributed and homogeneous, then the researcher will conduct an independent samples test. Independent Samples Test results are found in Table 4. Based on Table 4, the calculated T value (3,731)> T table is 1.72472 then H0 is rejected. So the two models do not have the same average.
In Table 1 the average for the SAVI learning model is 86.5334, while for the SNOWBALLING learning model is 80.3667. It means that there is an average difference between the SAVI learning model and the SNOWABALLING learning model. So there will also be differences in their effect on the numerical ability of students, where the SAVI learning model is superior to the SNOWBALLING learning model. This can be seen from the steps in the two learning models.  (Yulianti, Haris, and Chandra 2018;Nur Agustina, Yurniwati, and MS 2019;Setyani, Nizardin, and Utami 2019;Sulakasana , Margunaasa, and Authority 2018;Dewi, Murda, and Pudjawan 2019). Different from previous studies, our research has proven that the SAVI and SNOWBALLING learning models actually have an impact on the numerical ability of students (Munifah, Tsani, I., Yasin, M., Tortop, H. S., Palupi, E. K., & Umam, R. 2019), and the most effective is the SAVI learning model.

CONCLUSIONS AND SUGGESTIONS
Based on the results of the study concluded that there is an influence of the SAVI learning model and the SNOWBALLING learning model on the numerical abilities of students. However, according to the data that has been obtained, the SAVI learning model is more effective than the SNOWBALLING learning model, because the SAVI learning model encourages the integration of physical movements with intellectual activities that use all the senses so that the numerical abilities of students develop faster.

Numeric ability
Levene's test for equality of variances