Chemistry is one branch of natural science. Chemistry is vital in education and everyday life, meaning that chemistry is fascinating to learn (Ahmad et al., 2021)(Redhana & Merta, 2017)(Yuliastini et al., 2018). However, most learners consider learning chemistry less interesting and confusing (Yuliastini et al., 2018). Many students have a negative stigma towards studying chemistry (Almubarak et al., 2023). According to (Rahayu et al., 2020), one of the causes of students' chemistry learning difficulties is saturation and low motivation towards chemistry learning. It is because students have difficulty understanding complex chemical concepts (Ertikanto et al., 2018)(Ijirana & Supriadi, 2018)(Redhana & Merta, 2017) besides, chemical materials are also microscopic, so they cannot be imagined by students (Dewi et al., 2019)(Fitriyana et al., 2020). Current technological advances should be able to overcome the problem of saturation and low motivation among learning learners (Cano et al., 2023)(Froehlich, 2018)(Talledo et al., 2020) because there are so many technologies that can be used to support learning (Creer, 2018). However, there are still many students who have not received learning media that are packaged following technological developments (Avalos et al., 2019)(Froehlich, 2018)(Talledo et al., 2020), Thus resulting in saturation and low motivation to learn students which ultimately affects the cognitive learning outcomes of students. It is what causes students' low cognitive learning outcomes in chemistry subjects (Rahayu et al., 2020). Therefore, innovation is needed to overcome these problems. One of the innovations that can be done to improve students' cognitive learning outcomes is the use of learning media (Widarti et al., 2020).

Learning media must be packaged following technological developments and involve and utilize information and communication technology as facilities to facilitate the learning process (Ernawati et al., 2022)(Haristiani & Rifa'i, 2020)(Miskiah et al., 2019). It is following the demands on education in the 21st century (Fitriyana et al., 2020)(Soegoto et al., 2022)(Wardani et al., 2017). Science, especially in Information, Communication, and Technology (ICT), is developing rapidly and is sophisticated (Elina et al., 2016)(Saripudin et al., 2020)(Wardani et al., 2017). Therefore, the use of electronic learning media (e-learning) that can be used anywhere and anytime is expected to increase students' interest in learning, as well as increase learning effectiveness and efficiency (Atmojo et al., 2020)(Sangsawang, 2020)(Wardani et al., 2017).

E-learning media is a learning media that follows technological developments and answers the needs of students for learning that can adapt to the times (Ana, 2020)(Suhandiah et al., 2022). However, the use of e-learning media alone is not enough to improve students' cognitive learning outcomes if the e-learning media does not follow the learning styles of students (Gigzaw & Hirpa, 2020)(Mufdalifah et al., 2018). It is because students do not have an optimal understanding of the material. Therefore, it is necessary to have e-learning media tailored to learners' learning styles (Hariadi, 2015).

Based on research by (Hajrhma et al., 2022), not all students know their learning style, so they are not optimal in understanding the material (Audu, 2020). It means that it is essential that teachers direct students to know the type of learning style they have. Each student can have their own learning style, so differences in learning styles affect the process of receiving student information in learning (Banjarnahor et al., 2017). According to (Hariadi, 2015), there are three types of learning styles: visual learning style, auditory learning style, and kinesthetic learning style. Visual learning is a learning style that involves seeing, so the sense of sight plays an important role in this type of learning style. Auditory learning is a learning style that is based on listening, so the sense of hearing plays an important role in this type of learning style. Kinesthetic learning is a learning style through experience, movement, and touch. Based on these three learning styles, there must be three types of e-learning media under the three types of learning styles.

According to researchers, learning media in accordance with the three types of learning styles and current technological developments are e-posters, podcasts, and interactive videos. The three learning media are electronic-based. It is in line with the development of the millennial generation's needs, namely the use of digital/electronic media (Ernawati et al., 2022). It is also in accordance with the current development of the world of education, namely differentiated learning (Rafi & Pourdana, 2023).

Based on the description above, it is necessary to innovate learning media that is packaged according to electronic-based technological developments. In addition, to maximize the process of students understanding the material, the learning media must also be adjusted to the type of learning style of students. This study aims to analyze the effectiveness of e-learning media based on learning style in the form of eposters, podcasts, and interactive videos on the cognitive learning outcomes of learners with visual, auditory, and kinesthetic learning styles. Therefore, the e-learning media based on the learning style developed in this study can be applied to other chemical materials or subjects.

This research uses a quantitative approach, a knowledge discovery process that uses data in the form of numbers to find information that someone wants to know (Gunawan et al., 2022). This approach is the most suitable for collecting accurate population information. Research data analyzed with a quantitative approach must be observable, measurable, and in the form of objective reality that can be broken down and studied independently. Analysis in quantitative research is centered on numerical data (numbers), which are then processed using descriptive statistical methods. Descriptive statistics are statistics used to analyze data by describing the data obtained so that through a quantitative approach, objective research results will be obtained (Hashim et al., 2022). Therefore, the quantitative approach is the approach that best suits this study.

The assumption test used in this study is a normality test. The normality test is conducted to determine whether the data obtained by researchers comes from a normally distributed population. The normality test in this study used SPSS 21 as an auxiliary tool. According to (Ranti et al., 2017), the hypotheses tested are as follows:

H0 : Normal distributed data

H1 : Data is not normally distributed

Hypothesis testing is carried out through the Kolmogorov-Smirnov test. The criteria for taking the above hypothesis, according to (Ranti et al., 2017), are as follows:

a.If sig<0.05, H0 is rejected

b.If sig>0.05, H0 is accepted

The hypothesis test used is the Wilcoxon signed-rank test. The Wilcoxon signed-rank test is a non-parametric test based on paired samples. The Wilcoxon signed rank test determines whether there are differences in learning outcomes between before and after using e-learning media based on learning style. This test is also used to determine whether there is an increase in learning motivation after using e-learning media based on learning style in learning. This test is used if the data is abnormally distributed (Bina, 2021). In this study, the hypotheses used are:

The decision-making on the Wilcoxon signed-rank test can be obtained using the Asymp sig value. The basis for decision-making is the same, namely:

Based on the Wilcoxon signed rank test analysis, data were obtained on whether e-learning media based on learning style had a significant effect on the cognitive learning outcomes of students. Then, this N-gain analysis was conducted to determine how much influence the e-learning media based on the developed learning style on students' cognitive learning outcomes (Cohen & Swerdlik, 2010). This N-gain analysis shows the effectiveness of e-learning media based on learning styles developed on students' cognitive learning outcomes. The calculation formula used is as follows:

Information:

= normalized gain

Si = skor pre-test

Sf = post-test score

Is = skor maximum ideal

The criteria used to represent the normalized N-gain value, according to Cohen & Swerdelik (2010), can be seen in Table 1.

The research instruments used in this study were learning style analysis questionnaire instruments and learning outcome test instruments. The learning style analysis questionnaire contains 30 statements adapted from (Chislett & Chapman, 2005). The learning outcome test instrument used in this study amounted to 5 description questions that had been declared valid based on the results of validation that five validators had carried out. The calculation of validation results in this study uses Aiken's V statistics. Aiken (1985) formulated Aiken's V formula for validity values based on the assessment results of validators. Here are Aiken's V statistics formulated:

As for reliability, it is calculated using the Cronbach alpha formula, which is as follows:

Information:

r11 = Cronbach Alpha reliability coefficient

k = many items/question items

Σσ2b = number / total variance per item/item of question

σ2t = number/ total variance.

Based on calculations, the coefficient of reliability of the learning outcome test instrument is 0.98, so it can be concluded that the instrument is reliable.

The population in this study was all grade X students of SMA Negeri 3 Banjarmasin. The sample used was class X PMIA 1 students, totalling 32 students. The determination of samples used in this study used purposive sampling techniques because the selection of research samples was based on several criteria. The sample in this study has been following the criteria applied by the researcher, namely students (samples) are class X semester II, students have low cognitive learning outcomes in chemistry learning known from interviews with subject teachers and pre-test results, students do not know the type of learning style they have which is known from observations made on students, And students master how to use Google Classroom well, this is because students are familiar with online learning using Google Classroom. Therefore, class X PMIA 1 can be used as a sample in this study to represent the other class X PMIA. Based on interviews with subject teachers, all class X students have the same criteria.

This study used a learning style questionnaire containing 30 points of statements adapted from (Chislett & Chapman, 2005). Each statement has three answer choices: answer choice A for visual learning style, answer choice B for auditory learning style, and answer choice C for kinaesthetic learning style. This learning style questionnaire is presented in Google Forms. The results of the learning style analysis of class X PMIA 1 students are presented in Table 2.

Table 2 shows that the learning style that most learners have is the auditory learning style. The learning styles of learners are also influenced by their learning experiences. It is because one of the factors that affect learning style is external factors. In this case, external factors that play an essential role are school factors, namely learning methods. The lecture method is the most frequent and most commonly used method of learning. Thus, if it is associated with auditory learning styles, the lecture method, which is the method most often encountered by students, makes them accustomed to learning and requires them to concentrate on understanding the material delivered orally. It makes students tend to have an auditory learning style.

Data on learners' cognitive learning test results obtained from pre-test and post-test in limited trials can be seen in Table 3. Based on Table 3, it is known that at the time of the pre-test, students' knowledge is in the sufficient and good category. After being given treatment, namely learning using e-learning media based on learning style, the post-test results of students are in the good and very good categories. The lowest and highest values of pre-test and post-test are presented in Table 4.

Table 4 shows that after going through the learning process, post-test results have improved. It means that the learning process goes well. The application of learning using e-learning media based on learning style is more interactive because students can learn to use learning media that suits their learning style. Besides that, when in class, students are grouped according to their learning style, then problems are discussed, and the results of their discussions are conveyed according to their respective learning styles, namely simple posters, sound recordings, and videos. Learning by utilizing technology combined with learning styles can increase student participation and independence of learning, and make students more active during discussions (Ernawati et al., 2022). Adaptive learning media is considered to have clear, systematic, logical stages, and can be used to measure critical and creative thinking abilities (Sulistyanto et al., 2022). Students need to be trained to use their thinking skills in learning so that they not only can memorize but also can think critically (Listiaji et al., 2022). The learning atmosphere in the classroom becomes more pleasant and makes students enthusiastic to participate in classroom learning. The results of student discussions on each learning style are presented in Table 5

The average value of learning outcomes in the realm of knowledge of students in each indicator of competency achievement (GPA) can be seen in Table 6.

Table 6 shows that there was an increase in the percentage of learning outcomes in all indicators after implementing learning using e-learning media based on learning style. Of the five indicators, the indicator with the highest level of achievement or percentage is indicator 3, which groups (C2) solutions into strong electrolytes, weak electrolytes, and non-electrolytes. This is because participants already understand the concept of electrolyte and non-electrolyte solutions well, so it is easy to group solutions. In addition, this matter presents solutions not only in the laboratory but also in everyday life. The indicator with the lowest level of achievement or percentage is indicator 4, which explains (C2) the event of a person's electrocution when there is direct contact with electricity. This is because students still have difficulty deeply understanding that electrolytes in the human body can conduct electricity. In addition, this material is microscopic, so students cannot imagine it.

The average pre-test and post-test scores of learners' cognitive learning outcomes in each learning style are presented in Figure 1.

Based on the data presented in Figure 1, which shows the average pre-test and post-test scores of cognitive learning outcomes of students in each learning style, it is known that the average post-test scores of students with auditory learning styles have the highest percentage. The results of this study support the results of previous research conducted by (Hariadi, 2015), which showed that students with auditory learning styles obtain better cognitive learning outcomes compared to students who have visual and kinesthetic learning styles. The results of this study also show that the lecture method makes students tend to have an auditory learning style. When the lecture method is packaged in accordance with technological developments, which are developed in the form of a podcast, the cognitive learning outcomes of students with auditory learning styles are better than students with visual and kinesthetic learning styles. It is directly proportional to the data in Figure 2, which shows that the average N-gain value of learning outcomes on auditory learning styles obtained the highest value. That is, students with auditory learning styles experience the greatest improvement in cognitive learning outcomes among the three learning styles. However, overall, the difference between the average post-test score and the average N-gain of learning outcomes between learning styles is very small.

Figure 2 shows that in each learning style, there are 11 students with N-gain values of learning outcomes categorized as medium and 21 students with N-gain values of learning outcomes categorized as high. It means that there are still some students whose improved learning outcomes are in the moderate category. Figure 2 also shows that the cognitive learning outcomes of learners with auditory learning styles are higher than learners with visual and kinesthetic learning styles. It is the influence of the learning method that students get at the previous level of education, namely the lecture method, that requires them to concentrate on understanding the material delivered orally. Therefore, students with auditory learning styles are accustomed and more receptive to the lecture method, and when the lecture method is packaged in accordance with technological developments, which are developed in the form of podcasts, students with auditory learning styles will find it easier to accept the material delivered. In Table 6, it can be seen that the lowest N-Gain value is found in the kinesthetic learning style, which is 0.71, while the highest N-Gain value is found in the auditory learning style, which is 0.80, while the N-Gain value in visual learning style is 0.75. From these values, it can be seen that the difference in N-gain values between learning styles is not significant, and based on the data presented in Table 7, overall, it shows that the average N-gain value of learning outcomes in each learning style is categorized as high. Based on the N-Gain value, it can be seen that the e-learning media based on the learning style developed in this study is very influential on the cognitive learning outcomes of students. The results of this study also support previous research conducted by (Hariadi, 2015), which found that learning styles affect the cognitive learning outcomes of students.

The normality test was conducted to determine whether the cognitive learning data of students before and after the use of e-learning media based on learning styles developed in learning obtained in normally distributed research or not (Ranti et al., 2017). In conducting normality tests, researchers use SPSS 21 as a tool. The normality test results from the pre-test and post-test data can be seen in Table 8.

Based on the results of the normality test in Table 8 above, it can be seen that the sig value in the Kolmogorov-Smirnov section table, both pre-test and post-test, is sig. < 0.05. It shows that H0 is rejected and H1 is accepted, which means that the data on cognitive learning outcomes of students before and after the use of e-learning media based on learning styles in learning (pre-test and post-test) in this study are not normally distributed. Therefore, the analysis of results cannot be done with parametric tests, so the analysis of data on cognitive learning outcomes of students is carried out with non-parametric tests, namely with the Wilcoxon signed rank technique(Bina, 2021).

a. Wilcoxon Signed Rank Test

Pre-tests and post-tests in this study were given to the same learners. Therefore, the analysis of cognitive learning outcomes in this study used the Wilcoxon signed rank test. The Wilcoxon signed rank test was conducted to determine whether there were differences in students' cognitive learning outcomes before and after the use of learning style-based e-learning media in learning. In conducting the Wilcoxon signed rank test, researchers used SPSS 21 as a tool. The Wilcoxon signed rank test results from student cognitive learning outcomes data can be seen in Table 9.

Based on the statistical test of the Wilcoxon signed rank test on the cognitive learning outcomes data of students presented in Table 9, an asymp value was obtained. Sig. (2-tailed) which is 0.000 <0.05. These results show that H0 is rejected and H1 is accepted, which means that there are differences in students' cognitive learning outcomes between before (pre-test) and after (post-test) using e-learning media based on learning styles in learning in the form of e-posters, podcasts, and interactive videos.

Based on the results of the research that has been described, it is known that of the three learning styles, students with auditory learning styles with learning media in the form of podcasts experience the greatest increase in cognitive learning outcomes among students with visual and kinesthetic learning styles, with learning media in the form of e-posters and interactive videos, namely with N-Gain values on visual, auditory, kinesthetic learning styles of 0.75, 0.80, and 0.71, respectively. However, overall, based on the N-gain value and Wilcoxon signed rank test, e-learning media based on learning style in the form of eposters, podcasts, and interactive videos are effective for improving the cognitive learning outcomes of students with visual, auditory, and kinesthetic learning styles. The results of this study show that it is very important for students to know the type of learning style they have and get learning media that suits their respective learning styles so that students are more optimal in receiving the material delivered in learning. Therefore, there is a need for innovations in preparing learning media for students, which is packaged with technology and in accordance with the learning styles of students. In addition, there is a need for a learning style test to determine the type of learning style of students. Learning style tests can be conducted to all students every new school year. So that all teachers have data on student learning styles and can apply differentiated learning based on student learning styles.