Nidhomuddin¹, N Chamidah2 and A Kurniawan2

 

¹PhD Student, Faculty of Science and Technology, Airlangga University, Indonesia

²Department of Mathematics, Faculty of Science and Technology, Airlangga University, Indonesia

Corresponding author:  nur-c@fst.unair.ac.id

Abstract. Longitudinal data is obtained based on observations for n independent subjects, where each subjects is observed repeatedly in different periods and correlated each other. Longitudinal data are important for studying individual transitions and especially for causal studies on individual behaviour changes. In this study, we use two variable response, scholastic aptitude test and islamic test score observed from 2016 until 2018. Two response variables have significant correlations that they are influenced not only percentage of public senior high school but also percentage of students that accepted in the first choice and the percentage of students who graduated in the same year. Therefore, for modeling of scholastic aptitude test and islamic test score, we use biresponse multipredictor model for longitudinal data based on local linear estimator. The result show that locally modeling approach is more suitable because the pattern of the scholastic aptitude test score and islamic test score based on three predictor variables of state islamic colleges in western and eastern Indonesia is different. The result of this study can be used by the ministry of religious affairs to produce policies so that the quality of state islamic colleges in Indonesia will be better by increassing scholastic aptitude test and islamic test.