Multinomial Logistic Regression Analysis in Determining Factors That Affect Social Media Selection Among Students of Faculty of Mathematics and Natural Science, Universitas Brawijaya

Logistic regression is a statistical analysis that focuses on explaining the relationship between categorical scale response variables that have two or more categories and one or more predictor variables. Multinomial logistic regression is a logistic regression analysis where the response variable has more than two categories. The multinomial logistic regression method does not require a classic assumption test but only requires a non-multicollinearity assumption test. The aim of this research is to form a multinomial logistic regression model and identify factors that influence social media selection among students at the Faculty of Mathematics and Natural Sciences, Universitas Brawijaya. The sampling technique used is proportional stratified random sampling and the sample size was 100 undergraduate students of Faculty of Mathematics and Natural Sciences, Universitas Brawijaya. The response variable consists of three categories, namely X (Twitter), Instagram, and TikTok, with category X (Twitter) as a reference. The results of partial parameter testing show that four variables give a significant effect in choosing Instagram over X (Twitter), namely gender, purpose of use, frequency of use, and influence of friends; meanwhile, three variables give a significant effect in choosing TikTok over X (Twitter), namely purpose of use, ease of use, and type of content. The model classification accuracy based on confusion matrix and 100% - APER value is 67%.