Exploring the Inﬂuence of Social Media Usage for Academic Purposes Using a Partial Least Squares Approach

968-2672-0101-307 Abstract: Social media applications have been increasingly gaining signiﬁcant attention from online education and training platforms. Social networking tools provide multiple advantages for communicating, exchanging opinions, and discussing speciﬁc issues. Social media also helps to improve the processes of teaching and learning through sharing educational programs. In this study, we used a quantitative research technique based on the partial least-squares (PLS) linear regression method to determine the inﬂuence of using social media as an online discussion and communication platform for academic purposes by assessing the relationships among the skills obtained through social media, the usage of social media, and the purpose of social media. A total of 200 students participated in this study (88% female and 12% males), and a purposive sampling technique was used to select a suitable population for the study. The results show that 61.5% of the participants use the web daily for more than ﬁve hours, mainly for social communication (meaningful dialog and discussion skills) and entertainment. The students agreed that social media develops their creative thinking, but it has no positive impact on their academic performance.


Introduction
Social media (SM) applications have gained significant attention in academia and practice; according to the annual reports of social media management platforms, social media and mobile apps achieved significant growth in usage in 2018.Throughout the Middle East, about 250 million people use 304.5 million mobile phones [1].Social media has increasingly become a necessary tool for maintaining connections and transferring experiences globally.Emmert-Streib and Matthias examined incorporating big social data with social networks that needed to build accurate prediction models [2].Emmert-Streib et al. described the importance of data from social media in a general context, which provides excellent opportunities for extensive mining amounts of text, image, and videobased data [3].Statista has reported a steady increase in social media usage globally [4] and has recorded 2.82 billion users worldwide, which is expected to increase to 3.09 billion users, as shown in Figure 1.
Social networking tools provide multiple advantages for communicating, exchanging opinions, and discussing specific issues.Social media helps to improve the processes of teaching and learning by sharing educational programs.It helps to improve learning and distance learning methods, especially due to the divergence experienced during the COVID 19 pandemic.It also provides a platform for gaining experiences and analyzing and discussing different viewpoints on academic issues.It gives the possibility of creating unique rooms for training and exchanging study materials and useful links.
ing unique rooms for training and exchanging study materials and useful links.
The distribution of social media users worldwide shows considerable variation in the number of users from one place to another, depending on the types of social sites and their features and support for different natural languages.Figure 2 illustrates that Facebook usage dominates in most Western countries, with 2.3 billion users, followed by YouTube with 1.9 billion users and WhatsApp with 1.5 billion users.Globally distributed number of users using social media networks [4].
The recent reports indicate an increase of 11% in the number of social media users in Oman in 2021, reaching 4.14 million users [5].Uddin M. found that addiction to social media affected mental health, and psychological disturbances could lead to anxiety, sleeping disorders, depression, and poor academic performances [6].Al Rahmi and Othman concluded that social media facilitated students' educational experiences; however, if not controlled and time managed, it could negatively affect students' academic performances [7].Nasrullah and Khan conducted a study in Saudi Arabia and revealed that social media did not support student learning.Students enjoyed meeting new friends online using social media, thereby spending too much of their time on such due to their online addiction [8].El Khatib and Khan claimed that most Omani students used social networking (SN) for social purposes rather than academic purposes.The students The distribution of social media users worldwide shows considerable variation in the number of users from one place to another, depending on the types of social sites and their features and support for different natural languages.Figure 2 illustrates that Facebook usage dominates in most Western countries, with 2.3 billion users, followed by YouTube with 1.9 billion users and WhatsApp with 1.5 billion users.ing unique rooms for training and exchanging study materials and useful links.The distribution of social media users worldwide shows considerable variation in the number of users from one place to another, depending on the types of social sites and their features and support for different natural languages.Figure 2 illustrates that Facebook usage dominates in most Western countries, with 2.3 billion users, followed by YouTube with 1.9 billion users and WhatsApp with 1.5 billion users.

Number of users (billion)
Social media type Figure 2. Globally distributed number of users using social media networks [4].
The recent reports indicate an increase of 11% in the number of social media users in Oman in 2021, reaching 4.14 million users [5].Uddin M. found that addiction to social media affected mental health, and psychological disturbances could lead to anxiety, sleeping disorders, depression, and poor academic performances [6].Al Rahmi and Othman concluded that social media facilitated students' educational experiences; however, if not controlled and time managed, it could negatively affect students' academic performances [7].Nasrullah and Khan conducted a study in Saudi Arabia and revealed that social media did not support student learning.Students enjoyed meeting new friends online using social media, thereby spending too much of their time on such due to their online addiction [8].El Khatib and Khan claimed that most Omani students used social networking (SN) for social purposes rather than academic purposes.The students The recent reports indicate an increase of 11% in the number of social media users in Oman in 2021, reaching 4.14 million users [5].Uddin M. found that addiction to social media affected mental health, and psychological disturbances could lead to anxiety, sleeping disorders, depression, and poor academic performances [6].Al Rahmi and Othman concluded that social media facilitated students' educational experiences; however, if not controlled and time managed, it could negatively affect students' academic performances [7].Nasrullah and Khan conducted a study in Saudi Arabia and revealed that social media did not support student learning.Students enjoyed meeting new friends online using social media, thereby spending too much of their time on such due to their online addiction [8].El Khatib and Khan claimed that most Omani students used social networking (SN) for social purposes rather than academic purposes.The students diverted their attention to social communications, which caused lower academic achievements [9].Therefore, a compelling reason for this study was to investigate using social networks.In this study, we deployed a quantitative research technique based on the linear regres-sion method to determine the influence of using social media as a communication and discussion channel for academic purposes by assessing the relationships among the skills obtained through social media, the usage of social media, and the purpose of social media.

Literature Review
The first part of the literature review focuses on the positive usage of social media.Chytas D. confirmed that most students used Facebook and YouTube channels for communicating and studying [10].Hashim et al. found that most publications were available through social media networking, and most manuscripts focused on applying social media [11].Al-Rahmi confirmed that using social media positively improved most students' performances [12].Price et al. showed that using social media positively encouraged students to discuss and exchange information [13].Bagarukayo E. confirmed that Facebook had a positive impact on students as a learning tool for assignment activities [14].Alhaddad M. showed that students preferred social media platforms as a source for information [15].Duke confirmed the possibility of using social media to discuss academic issues [16].Bal E.B. claimed that social media contributed to increased understanding and ease of sharing information [17].Tang confirmed that students used Twitter for sending subject-related materials and formative assessment activities, and therefore, Twitter in education promoted student interactions and learning content [18].Rueda L. showed that there were highperforming students using social media but claimed that teachers using traditional teaching methods engaged their students more and, therefore, enhanced their performance [19].Ali et al. showed a strong relationship between student learning performance and the satisfaction rate in information system courses using social media.The use of social media improved their academic performance, while traditional teaching methods allow instructors to engage more with students [20].Benetoli et al. performed a comparison study, which indicated that 62% of participants agreed that social media positively improved their performance [21].Kitching et al. claimed that students' communication skills improved by using healthcare programs with blogs, Twitter, and YouTube [22].Van R. indicated that social media has made teaching more accessible and enjoyable and that students also supported social media technologies [23].Eger L. stated that students felt that using Facebook was important for learning and identifying false information [24].Dunn L. indicated that students believed that social media networks were very helpful and could increase their learning experience [25].Junco et al. proved that social networking sites, such as Twitter, Wiki, and blogs, effectively enhanced students' academic performances.Junco claimed that students and academic staff were both interested in the learning process through Twitter instead of traditional learning.Therefore, it promoted social media as a suitable method for engaging students in their academic development [26].Moran et al. indicated that social media sites had a significant impact on teaching and collaborative learning [27].George and Dellasega showed that students preferred teachers who used social media as electronic resources in education, as it seemed to be very helpful in the medical curriculum [28].Alshdefait et al. confirmed sustained use of social media by students as a helpful medium for the education and learning process at Jordanian public universities [29].Sobaih et al. showed that social media had a significant impact on academic-related issues and, in particular, on teaching and learning [30].Vollum M. analyzed social media usage in physical and health education and showed that social interaction positively influenced educational outcomes [31].
The other part of the literature review focuses on the requirements of using social media in education platforms.Wang et al. confirmed that only a very few students used social media to complete their assignments and experienced improvements [32].Ghani and Hamid highlighted state-of-the-art social media big data analytical techniques and described the challenge of research problems in the field [33].
Willis and Exley proposed that using social media in the curriculum increased parents' engagement in student learning activities.However, teachers should have the essential substantive knowledge of language to inquire and contribute to social networking spaces online [34].Carranza et al. suggested that there were advantages of using mobile learning to streamline the training process [35].
Table 1 presents the comparative study results of the literature survey for the period between 2011 and 2019.Moran, 2011 [27] X X X X X X 40% of faculty members asked students to use social media networks to read and solve assignments George, 2011 [28] X X X X X Students rated the quality of teaching through social media as 4.8 out of 5; the rate of using social media as electronic resources in medical schools was 4.7 out of 5 Wang, 2011 [32] X X X X X X 20% of students used social media to solve their assignments, which contributed to their achievements Computation 2021, 9, 64 5 of 13

Data
Table 2 presents the results of respondents' descriptive demographic information.The data for this study were collected through a printed questionnaire that was provided to individual undergraduate students at different universities in Oman.The data collection period was two weeks.The students permitted us to use their responses in this study provided we did not publish personal information such as names, emails, or mobile numbers.The participants were given a training session to explain the study's aims and provide instructions on filling out the questionnaire.A purposive sampling technique was implemented to select a suitable population for the study.Therefore, 250 questionnaires were distributed, and only 200 students (176 females and 24 males) were selected.Each student answered the questions honestly according to his/her opinion.A total of 85 students were between the age of 16 and 19, and 114 students were between 20 and 29.One student was aged more than 30 years old.The social media sites included in this study were Instagram, Facebook, WhatsApp, Twitter, Snapchat, Pinterest, YouTube, LinkedIn, Google+, and Tumblr.

Research Methodology
In this study, we used a quantitative research technique based on partial least-squares (PLS) linear regression to determine the influence of using social media as a communication and discussion channel for academic purposes by assessing the relationships among the skills obtained through social media, the usage of social media, and the purpose of social media.A purposive sampling technique was used to select suitable participants for the study.
The research hypotheses were as follows: Hypothesis 1 (H1).The purpose of social media influences the skills obtained through social media.

Hypothesis 2 (H2).
The purpose of social media influences the usage of social media.

Partial Least Squares Method (PLS)
Partial least-squares (PLS) is a statistical method that finds a linear regression model by projecting the predicted variables and the explanatory variables to a new space.Therefore, it is sometimes called a "projection to latent structures".PLS requires that standardize latent variable scores have a mean value of 0 and a standard deviation value of 1.The latent variable approach is used to model the covariance structures in the two spaces (predicted and experimental variables).The main advantage of PLS is the ability to model multiple dependent and independent variables.The PLS algorithm was proposed by Lohmöller, in 1989, including three stages [36,37], as presented in Algorithm 1.
Input: standardize latent variable Output: latent constructs score, loading and path coefficients, location parameters Step 1: Estimation of latent variable scores Iteratively (1.1-1.4) until convergence is achieved or the maximum number of iterations is reached.Then go to Step 2.
1.1 outer approximation of the latent variable scores, 1.2 estimation of the inner weights, 1.3 inner approximation of the latent variable scores, and 1.4 estimation of the outer weights.
Step 2: Estimation of outer weights/loading and path coefficients.
Step 3: Estimation of location parameters.

Results and Discussion
Table 2 presents the respondents' descriptive statistical, demographic information.The current study considers factors such as skills obtained through social media, the usage of social media, and the purpose of social media.Details of the factors (latent variables) and subfactors (apparent variables) are presented in Table 3.The measurement model is used to test the latent variables and the apparent variables, as shown in the conceptual model in Figure 3.The variables used include the skills obtained through social media, the usage of social media, and the purpose of social media.

Measurement Model
The relationships were shown among the skills obtained through social media, the usage of social media, and the purpose of social media.The reliability of the model was tested based on validating the discriminant and convergent results [38]. Figure 4 shows the initial path model coefficients and the values of the variables (the skills obtained through social media, the usage of social media, and the purpose of social media).

Model Reliability
Composite reliability is used to validate the construct reliability results and inner consistency values, which is more suitable than Cronbach's alpha.Hair

Measurement Model
The relationships were shown among the skills obtained through social media, the usage of social media, and the purpose of social media.The reliability of the model was tested based on validating the discriminant and convergent results [38]. Figure 4 shows the initial path model coefficients and the values of the variables (the skills obtained through social media, the usage of social media, and the purpose of social media).

Measurement Model
The relationships were shown among the skills obtained through social media, the usage of social media, and the purpose of social media.The reliability of the model was tested based on validating the discriminant and convergent results [38]. Figure 4 shows the initial path model coefficients and the values of the variables (the skills obtained through social media, the usage of social media, and the purpose of social media).

Model Reliability
Composite reliability is used to validate the construct reliability results and inner consistency values, which is more suitable than Cronbach's alpha.Hair et al. determined the least score value of composite reliability should be 0.7, and the Cronbach's alpha

Model Reliability
Composite reliability is used to validate the construct reliability results and inner consistency values, which is more suitable than Cronbach's alpha.Hair et al. determined the least score value of composite reliability should be 0.7, and the Cronbach's alpha minimum score should be 0.6 [39].The measurement model reliability was validated based on subfactor and factor loading reliability.A score of 0.45 for the subfactors' loading is preferable [40].The scores of subfactor loadings are examined, and any variable with a score greater than 0.50 is accepted [41], and any subfactors with low loading scores are removed from the final model, as shown in Figure 5.In addition, Table 4 presents the latent construct scores of composite reliability, factor loadings, and Cronbach's alpha, which were achieved using the PLS algorithms.The results show the proposed model can be reliable because Cronbach's alpha score is above 0.437, and the score of composite reliability is greater than 0.621.composite reliability is greater than 0.621.

Convergent Validity
To verify the convergent validity of the proposed model, the following two conditions were required: (a) Achieve a score greater than or equal to 0.7 for the outer loadings.(b) Ensure that the score of average variance extracted (AVE) for all latent variables is more than 0.50 [42].A score of 0.4 is adequate [43] with the condition that the composite reliability score is more than 0.6 [44].Table 4 shows that the convergent va-

Convergent Validity
To verify the convergent validity of the proposed model, the following two conditions were required: (a) Achieve a score greater than or equal to 0.7 for the outer loadings.
(b) Ensure that the score of average variance extracted (AVE) for all latent variables is more than 0.50 [42].A score of 0.4 is adequate [43] with the condition that the composite reliability score is more than 0.6 [44].Table 4 shows that the convergent validity is acceptable based on the obtained variance extracted score ranging from 0.492 to 0.739.Discriminant validity should evidence that the construct (latent variable) in the PLS path model has the most robust relationship score as compared with other construct variables.Table 5 shows that the obtained results of the square root of AVE and constructs correlation score were in the accepted ranges, which indicates a satisfactory discriminant validity for constructs.

Structural Model Analysis
The PLS bootstrapping test is used to determine the R-squared results of independent and dependent variables, in addition to path coefficient values, which are shown in Table 6 and illustrates in Figure 6.

Structural Model Analysis
The PLS bootstrapping test is used to determine the R-squared results of independent and dependent variables, in addition to path coefficient values, which are shown in Table 6 and illustrates in Figure 6.The path coefficient β is tested based on the obtained structural model and hypothesis, as PLS does not require normally distributed data.However, the latent dependent variables are evaluated using R 2 and the average variance.The construct's percent variation of the model is measured using R 2 [45].The model has a predictive significance if the R 2 score is greater than zero, and a score less than 0 indicates a predictive insignificance model, as shown in Figure 7.
Computation 2021, 9, 64 11 of 14 Equation (1) determines the goodness-of-fit (GOF) value, which is the overall model fit for PLSEM as follows: Table 7 presents that the GOF of the proposed model is equal to √0.04912 × 0.618163 = 0.174253.The path coefficient β is tested based on the obtained structural model and hypothesis, as PLS does not require normally distributed data.However, the latent dependent variables are evaluated using R 2 and the average variance.The construct's percent variation of the model is measured using R 2 [45].The model has a predictive significance if the R 2 score is greater than zero, and a score less than 0 indicates a predictive insignificance model, as shown in Figure 7.

Conclusions
Many studies have examined the purpose of using social media for teaching and training in academic platforms and have shown that social media does not support students' learning and that its use should be controlled and time managed to prevent adverse effects on students' academic performance.Therefore, in this study, we investigated the influence of using social media as a communication and discussion channel for educational purposes by assessing the relationship between the skills obtained through

Conclusions
Many studies have examined the purpose of using social media for teaching and training in academic platforms and have shown that social media does not support students' learning and that its use should be controlled and time managed to prevent adverse effects on students' academic performance.Therefore, in this study, we investigated the influence of using social media as a communication and discussion channel for educational purposes by assessing the relationship between the skills obtained through social media and using social media and the purpose of social media.We used a quantitative research technique based on the linear regression method to examine and investigate the influence of using social media as a discussion and communication platform for academic purposes.A total of 200 participants, including 195 undergraduate students from various Omani universities, participated in this study (88% female and 12% males).61.5% of the student respondents reported using the web daily for more than 5 h and mainly for social dialog communication and entertainment purposes.The students agreed that social media develops their creative thinking, but it has no positive impact on their academic performance.The conceptual model reliability was validated based on subfactor loading with a score of 0.45.The result of the mathematical PLS model based on Cronbach's alpha score (0.437) was valid, and the composite reliability was greater than 0.621.This means that the students were more focused on the skills they obtained through social media than the purpose of social media as a communication and discussion channel, and they did not focus on enhancing academic performance.
The following are some suggestions for increasing students' engagement and interests: (a) Encourage students to use social media for academic purposes and practice; (b) Encourage teachers to create and monitor special classes for engaging students in discussions and presentations; (c) Create free space for dialogue between students and exchanging views regarding solving school assignments; (d) Closely monitor and direct the students to use social media positively as flipped classrooms; (e) Restrict the students from wasting their time indulging in unproductive chat purposes; (f) Raise awareness of the risks involved in using social media platforms (due to security concerns) rather than curtailing them.

Figure 2 .
Figure 2.Globally distributed number of users using social media networks[4].

Figure 3 .
Figure 3. Conceptual model for assessing the relationships among the skills obtained through social media, the usage of social media, and the purpose of social media.

Figure 4 .
Figure 4. Initial path model coefficients and the values of the variables (s1-s5 skills obtained through social media; u1-u6 the usage of social media; p1-p4 the purpose of social media).
et al. determined the least score value of composite reliability should be 0.7, and the Cronbach's alpha

Figure 3 .
Figure 3. Conceptual model for assessing the relationships among the skills obtained through social media, the usage of social media, and the purpose of social media.

, 9 , 64 8 of 14 Figure 3 .
Figure 3. Conceptual model for assessing the relationships among the skills obtained through social media, the usage of social media, and the purpose of social media.

Figure 4 .
Figure 4. Initial path model coefficients and the values of the variables (s1-s5 skills obtained through social media; u1-u6 the usage of social media; p1-p4 the purpose of social media).

Figure 4 .
Figure 4. Initial path model coefficients and the values of the variables (s1-s5 skills obtained through social media; u1-u6 the usage of social media; p1-p4 the purpose of social media).

Figure 5 .
Figure 5. Final path model after removing the least significant variables.

Figure 5 .
Figure 5. Final path model after removing the least significant variables.

Figure 6 .
Figure 6.Values of selected variables after implement the PLS bootstrap method.

Figure 7 .
Figure 7. Blindfolding evaluation procedure for validating the predictive accuracy of the model.

Figure 7 .
Figure 7. Blindfolding evaluation procedure for validating the predictive accuracy of the model.

Table 3 .
Details of the latent and apparent variables.

Table 4 .
Indicators of latent constructs score.

Table 4 .
Indicators of latent constructs score.

Table 6 .
Bootstrap values and T-values of path coefficients.Discriminant validity should evidence that the construct (latent variable) in the PLS path model has the most robust relationship score as compared with other construct variables.Table5shows that the obtained results of the square root of AVE and constructs correlation score were in the accepted ranges, which indicates a satisfactory discriminant validity for constructs.

Table 6 .
Bootstrap values and T-values of path coefficients.
Figure 6.Values of selected variables after implement the PLS bootstrap method.

Table 7 .
Model evaluation results.

Table 7 .
Model evaluation results.