Effectiveness of Microsoft Kaizala and Google Classroom towards students’ mathematical communication skill and self-efficacy in learning statistics

This study aims to describe the effectiveness and differences in the effectiveness of the learning platform with Google Classroom and Microsoft Kaizala in terms of self-efficacy and students' mathematical communication skills. The population of this research is the XII grade students of SMK Negeri 1 Giritronto Wonogiri which consists of sevent classes. From the existing population, two classes were taken randomly, namely twelfth grade of TKJ-I and twelfth grade of TKJ-II as research samples. Twelfth grade of TKJ-I was given treatment by learning using the Microsoft Kaizala platform, while twelfth grade of TKJ-II was given treatment using the Google Classroom platform. The research data were analyzed by statistical one sample t-test, MANOVA test with Hotelling's T 2 at a significant level of 0.05 and univariate test to determine which platform is more effective. The results showed that: (1) statistical learning using the Microsoft Kaizala platform was effective in terms of mathematical communication and self-efficacy, while the Google Classroom platform was effective in terms of mathematical communication but not effective in terms of self-efficacy; (2) there is a difference in effectiveness between the Microsoft Kaizala platform and Google Classroom. The Microsot Kaizala platform is more effective than Google Classroom in terms of the mathematical communication skills of class twelfth grade students of SMK Negeri 1 Giritronto Wonogiri.


Introduction
The development of science and technology facilitates communication and the acquisition of various information more quickly. To learn information about science and technology, reliable human resources are needed who have adequate capabilities and are globally competitive. Therefore, high order thinking skills (HOTS) are an important component that our human resources must have and this way of thinking can be developed through learning mathematics. The interrelated mathematical structure and consistent deductive mindset equip students who study it to be able to think logically, analytically, critically, creatively, reason effectively, efficiently, be scientific, work together, be confident, and be responsible (Ansari, 2018;Putri & Santosa, are still many students whose self-efficacy is still quite low. Based on the observations of researchers in class XI at SMK Negeri 1 Giritronto Wonogiri, some students in that class experienced a crisis of confidence and were always pessimistic/lack of confidence in solving math problems/problems given by the teacher. In the process of learning mathematics, it is still often found that there is a tendency of students who do not want to ask the teacher about how to solve mathematical problems even though students do not actually understand the material being studied. When the teacher asked which part they did not understand, the student's response was only silence, after the students finished the task, the teacher found out that there were still many students who did not know how to solve it. In addition, students' self-confidence and confidence level are still lacking when asked by teachers to solve math problems. This condition becomes increasingly critical when learning is carried out online and asynchronously. When the teacher asks students to solve problems and communicate problem solving individually in discussion forums, it turns out that students do not want to communicate the results of their solutions for fear of making mistakes and being less confident in themselves. The results of student work seem hesitant in writing the completion steps or expressing in the presentation of a mathematical symbol, even though the initial concept of completion is correct. This raises the assumption that students' self-efficacy is still low. Some other facts that researchers found during classroom observations were that the learning platform used by teachers was still limited to Whatsapp media only. The process of learning mathematics is still dominated by the teacher through giving teachers PDF files to students. Students are only objects of learning where in practice students only receive materials sent by the teacher through the Whatsapp Group created by the teacher. Teachers have not created situations and conditions so that students can play an active role in learning activities through platforms that can support the implementation of conducive learning situations and activate students. Google Classroom is one of the best platforms that provides a set of advanced features, ideal tools for students to use and improve teacher performance in saving time, keeping classes organized, and improving communication with students (Iftakhar, 2016). In addition to this platform, Microsoft Kaizala can also be used as an alternative application with various features that can make it easier for students and teachers to complete their learning needs like Whatsapp (Sari, et. al, 2021).
Based on this description, the researchers hope that the Google Classroom and Microsot Kaizala learning platforms have the opportunity to be a more effective learning platform in improving students' mathematical communication and self-efficacy compared to conventional learning in mathematics learning, especially statistical material at SMK Negeri 1 Giritronto Wonogiri. Related to this, this study aims to describe the effectiveness and differences in the effectiveness of the learning platform with Google Classroom and Microsoft Kaizala in terms of self-efficacy and students' mathematical communication skills.

Method
This research is classified as semi-experiment research (quasi-experiment) using two equivalent experiment classes. The research design used was a non-equivalent pretest-posttest group design. This research was conducted at SMK Negeri 1 Giritronto Wonogiri, from April to September 2021. The population of this study was all students of class XI at SMK Negeri 1 Giritronto Wonogiri who were spread out in five parallel classes whose relative abilities did not differ significantly. Thus, two classes were taken randomly from five existing classes, so that class XI-A was obtained as the first experiment group and XI-B as the second experiment group. Furthermore, class XI-A was 10.12928/bamme.v2i1.5523 given treatment with the Google Classroom interactive learning platform and XI-B was given treatment with the Microsoft Kaizala learning platform.
This research involves two variables, namely the independent variable and the dependent variable. The independent variable is the learning platform which consists of the Google Classroom and Microsoft Kaizala learning platforms. Meanwhile, the dependent variable is selfefficacy and mathematical communication skills.
The data collection technique in this research was carried out by giving tests and questionnaires both before and after the researchers gave treatment to the two experimental classes. The procedures used in this study are as follows: (a) making research instruments in the form of lesson plans, grid of pretest and posttest questions related to mathematical communication skills, scoring rubrics according to the variables to be studied, and selfquestionnaires. efficacy; (b) validate the research instruments made to several expert lecturers; (c) testing the instrument that has been validated on 60 students of class XI SMK Negeri 1 Giritronto Wonogiri; (d) provide self-efficacy questionnaires to students to be filled out before the pretest is carried out; (e) perform a mathematical communication ability pretest before treatment; (f) giving self-efficacy questionnaires to students to be filled out before the posttest; (g) provide a posttest of mathematical communication skills after treatment.
Data collection for non-test instrument regarding students' self-efficacy towards mathematics was obtained by using a questionnaire in the form of a checklist with a Likert scale within a time limit of 90 minutes. After obtaining self-efficacy measurement data, the total selfefficacy score is categorized based on predetermined criteria. For each statement, the respondent will be given a score according to the scale value of the answer category given based on the category of self-efficacy level that has been adjusted to the specified scale. The classification of student self-efficacy criteria according to Azwar's (2011) classification guidelines is shown in Table 1. Table 1. Student Self-Efficacy Criteria Interval Criteria + 1,5 < ≤ + 3 Very High (VH) + 0,5 < ≤ + 1,5 The test instrument regarding mathematical communication skills in this study consisted of pretest and posttest questions in the form of descriptions used to measure mathematical communication skills before and after treatment.
Descriptive analysis is used to describe the characteristics of the research data and answer descriptive problems. The descriptive analysis used in the study for data on mathematical communication skills and self-efficacy were the mean, variance, standard deviation, maximum score and minimum score. The research data analyzed were the data from the pretest and posttest on the aspect of mathematical communication skills and the results of filling out a self-efficacy questionnaire.
To determine the effectiveness of the two learning models applied in terms of each aspect, namely aspects of self-efficacy and students' mathematical communication skills, one sample ttest was used statistical test. The data analyzed for the one sample t-test is posttest data. The assumption test that must be met is the normality test of the self-efficacy questionnaire data and posttest data of students' mathematical communication skills after treatment in both groups, using the Kolmogorov-Smirnov. The data criteria are normally distributed if the significance is greater than 0.05.
In general, the statistical hypotheses tested are: To find out the difference in conditions before and after treatment from the two experimental groups in terms of self-efficacy and mathematical communication skills, the Multivariate of Variance (MANOVA) test was used with the help of SPSS 16 software for windows. The data analyzed by the MANOVA test were pretest data and posttest data from each variable.
The assumption test that must be met is the normality test and homogeneity test on the data from the pretest and posttest results of mathematical communication skills and self-efficacy questionnaire data before and after treatment in both groups. Multivariate normality test was carried out using the mahalanobis distance , -. If the sum of , which is less than . -(0,5is about 50%, then the data comes from a multivariate normal population. The statistical calculation of the MANOVA test according to Stevens is formulated as follows (2002): After obtaining the Hottelling's trace value, then it is transformed to obtain the F distribution value using the formula: Using p=3 is the number of dependent variables. The test criteria is that H ! is rejected if F "#$%& > F (!,!2;3,% ! 4% " *3*+) or the resulting significance number is less than 0.05. If in the MANOVA test the results of the similarity test of the mean scores in each group are significantly different, then the next hypothesis test is the univariate test with independent sample t-test.
To find out whether learning using the Google Classroom learning platform is more effective than the Microsoft Kaizala learning platform in terms of students' self-efficacy and mathematical communication skills, univariate test statistics (independent samples t-test) were used. The assumption test that must be met is the homogeneity test and the normality test on the students' self-efficacy and mathematical communication skills after treatment in both groups. Homogeneity test using Levene test with homogeneous data criteria is if the significance value is greater than 0.05. Meanwhile, the normality test uses the Kolmogorov-Smirnov with the criteria that the data is normally distributed if the significance value is greater than 0.05.
The criterion used is the Benferroni criterion where the significance level is α/p with p being the number of dependent variables (Stevens 2002). The test criteria are 678)9 < 9:;<= , then ! is rejected. The univariate test formula used according to Stevens (2002). Information:

Results and Discussion
All teaching and learning activities in this study took place in seven meetings, then students were given tests and questionnaires in both experimental classes. To describe the conditions before and after the treatment of the questionnaire and test results from each aspect, namely the aspects of self-efficacy and students' mathematical communication skills, the data on the results of the selfefficacy questionnaire are presented in Table 2. Based on the results of the descriptive statistical data analysis in Table 2, it shows that in the Microsoft Kaizala platform group, there was an increase in student self-efficacy score before and after treatment. This shows that the mean self-efficacy of students who take part in the learning process using the Microsoft Kaizala platform is better than the Google Classroom platform. The frequency distribution of student self-efficacy questionnaire data for the two experimental groups is shown in Table 3.
Based on Table 3, it can be seen that in the Microsoft Kaizala platform group after treatment cumulatively 80% of students have very high and high self-efficacy categories, while in the Google Classroom platform group 56,7% of students who have high self-efficacy criteria and very high, so it can be said that there is an increase in student self-efficacy after treatment.
The test results of students' mathematical communication skills on the Google Classroom platform group and the Microsoft Kaizala platform are presented in Table 4. Based on the results of the descriptive statistical data analysis in Table 4, it can be seen that the mean results of the mathematical communication ability test (posttest) in the two groups after treatment resulted in an increase in mathematical communication skills both in the Google Classroom platform group and the Microsoft Kaizala platform with different increase ranges. In the Google Classroom platform group, the mean score increased by 6.27, from an initial score of 10.12928/bamme.v2i1.5523 76.33 to 82.60. Meanwhile, in the Microsoft Kaizala platform group, the increase in score was 14.56, from the initial score of 74.94 to 89.50. Thus, it can be concluded that there is an increase in students' mathematical communication skills in both groups of models, and shows that the mean mathematical communication ability of students who take part in the learning process using the Microsoft Kaizala platform is better than the Google Classroom platform.
The research data was then analyzed to determine the effectiveness of each group of learning models, so a one sample t-test was tested. Hypothesis testing using the one sample t-test can be done if the assumption of normality is met. Based on the results of the normality test using the Kolmogorov-Smirnov test, the results are shown in Table 5.  Table 5 shows that all significance values are greater than 0.05. This shows that all data are normally distributed. Because the data is normally distributed, the one sample t-test can be done.
There are six hypothesis tests tested in this section, namely: !+ : ++ ≤ 75 (Learning mathematics with the Microsoft Kaizala platform is not effective in terms of students' self-efficacy).
!+ : ++ ≤ 75 (Learning mathematics with the Microsoft Kaizala platform is effective in terms of students' self-efficacy).
!-: -+ ≤ 70 (Learning mathematics with Microsoft Kaizala platform is not effective in terms of students' mathematical communication skills).
!-: -+ ≤ 70 (Learning mathematics with Microsoft Kaizala platform is effective in terms of students' mathematical communication skills).
!> : +-≤ 75 (Learning mathematics with the Google Classroom platform is not effective in terms of student self-efficacy).
!> : +-≤ 75 (Learning mathematics with the Google Classroom platform is effective in terms of student self-efficacy).
!? : --≤ 70 (Learning mathematics with the Google Classroom platform is not effective in terms of students' mathematical communication skills).
!? : --≤ 70 (Learning mathematics with the Google Classroom platform is effective in terms of students' mathematical communication skills). Based on Table 6, the test results show that the 678)9 value obtained for the Microsoft Kaizala platform group on the self-efficacy variable is 5,786 and the student's mathematical communication skills variable is 8,328, more than 9:;<= namely t(0,05; 29) = 2.045 so it can be concluded that !+ , !-is rejected, which means that learning mathematics with the Microsoft Kaizala platform is effective in terms of self-efficacy and students' mathematical communication skills and is significant because the acquisition of a significant value of 0,000 < 0,05.
Meanwhile, for the Google Classroom platform, the t "#$%& value for the self-efficacy variable is 1,998 where the value is less than 9:;<= =2,045, it is concluded that H !> is accepted, meaning that learning mathematics with the Google Classroom platform is not effective in terms of selfefficacy. This can also be seen from the significance value, which is 0.0053 > 0.05. Furthermore, the acquisition of the t "#$%& value on the student's mathematical communication ability variable is 4.978> 9:;<= = 2.045, it is concluded that !? is rejected, which means that learning mathematics with the Google Classroom platform is effective in terms of students' mathematical communication skills. This can also be seen from the significance value of 0.003 <0.05.
To find out which learning model is more effective between the Microsoft Kaizala platform and the Google Classroom platform, a univariate test (independent samples t-test) was conducted. Before being analyzed using the independent samples t-test, the mean difference test was first tested for the score data before treatment using the MANOVA test with T -Hotelling′s criteria. If the results conclude that the two classes are not different, then the score data analyzed to compare the effectiveness of learning with each learning model is the score data after treatment. Furthermore, the MANOVA test with T -Hotelling′s criteria can be performed if the assumption test is met. The assumption test that must be met is the normality test and the home genetics test.
The normality test of the data before and after the treatment used was multivariate normality. Based on the results calculations done manually with the help of office excel software. It is obtained that the mahalanobis path between each observation of the mean sector density after being sorted as shown in Table 7. 10.12928/bamme.v2i1.5523  Table 7 shows the mahalanobis distance between each observation and the mean vector after being sorted has a range that is not far from 50%, it can be concluded that the data before and after treatment on the Microsoft Kaizala platform group and the Google Classroom platform have a multivariate normal distribution.
Furthermore, the homogeneity test was carried out using the Box's M test, the results obtained as shown in Table 8.  Table 8, the significant value is greater than 0,05, it can be concluded that the covariance variance matrix of the two populations is homogeneous. The assumption test of score data before and after treatment is fulfilled, then the MANOVA test is continued.
The hypotheses that will be tested in this section are: ! : There is no difference in the mean score between the Microsoft Kaizala platform and the Google Classroom platform in terms of self-efficacy and mathematical communication skills : : There is difference in the mean score between the Microsoft Kaizala platform and the Google Classroom platform in terms of self-efficacy and mathematical communication skills Statistically, the hypothesis can be symbolized: The results of the multivariate test of pretest and posttest data can be presented in Table 9. Based on the results of the multivariate test in Table 9, it was obtained that the significance for Hotelling's Trace > 0,05, which was 0.886, indicating that ! accepted so that it is concluded that the initial conditions before treatment there is no difference in the mean self-efficacy and mathematical communication skills of students between the Microsoft Kaizala platform and the Google Classroom platform. Furthermore, because there is no difference in the mean in the pretest data, it is not necessary to carry out further tests on the pretest data. While the results of the multivariate test on the posttest data show the F-value in Hotelling's Trace is 7,320 : with a significance of 0,003 <0.05, so that at the 0,05 level of significance ! is rejected, meaning that in the final condition after treatment there is a difference in the mean self-efficacy and students' mathematical communication skills in the two experimental groups or it can be concluded that there is a difference in effectiveness between the Microsoft Kaizala platform group and the Google Classroom platform in terms of self-efficacy and students' mathematical communication skills.
The results of the multivariate test on the posttest data concluded that there was a mean difference between the Microsoft Kaizala platform groups and the Google Classroom platform in terms of students' self-efficacy and mathematical communication skills, so further tests were conducted to see which variables contributed to these differences. Next, a further independent sample t-test will be conducted with the Bonferroni kriteria criteria The hypotheses tested for the t-test are: ! : ++ ≤ +-(Microsoft Kaizala platform is no more effective than Google Classroom platform in terms of student self-efficacy).
: : ++ > +-(Microsoft Kaizala platform is more effective than Google Classroom platform in terms of student self-efficacy).
! : -+ ≤ --(Microsoft Kaizala platform is no more effective than Google Classroom platform in terms of students' mathematical communication skills).
: : -+ > --(Microsoft Kaizala platform is more effective than Google Classroom platform in terms of students' mathematical communication skills).
Univariate test results with Bonferroni criteria can be briefly presented in Table 10. Based on the results of the calculations in Table 10, it is obtained that the 678)9 for the student's self-efficacy variable is 3, 213 > 9:;<= = 3,182 so that ! is rejected. Therefore, it can be concluded that the Microsoft Kaizala platform is more effective than the Google Classroom