An Analysis of Social Media Activities and Student Performance using Topic Modeling and Correlation

Social Media is famous among teenagers and students. Social media changes the way we communicate to society and views in life. In this study, we explore and analyze the social media activities of students and their performance in school. Qualitative analysis was performed using topic modeling algorithm (Latent Dirichlet Allocation). The following narratives of students with good grades: {Trending Memes, Birthday Parties and Events, School Related, Election Related and Education / Learning Related}. The following narratives of students with low grades: {Trending Memes, Political Propaganda, Showbiz News}. The common narratives between the two corpora are Trending Memes and Politics. The coherence scores of the topic models show a positive result of 0.36 and 0.34 respectively. The narratives was discovered using open coding technique using the topic models as qualitative codes. Correlation analysis was conducted to analyze the relationship of the number of Facebook posts and the student grades. The result shows a negative correlation of 71.2%. The result can be interpreted as the number of social media activities increases, their math grades decreases. The Facebook posts presented in this study is not translated to English to avoid possible bias with the translation.