Multilevel Regression Analysis on Graduate Student Grade Point Average
DOI:
https://doi.org/10.24256/jpmipa.v12i1.3969Keywords:
Longitudinal Data, Multilevel Logistic Regression, Multilevel Regression, Natural Logarithm.Abstract
Abstract
Multilevel regression is one of the methods used to analyze hierarchical data structures. One case of data with a hierarchical structure is the cumulative grade point average (GPA) data for students each semester (level one) which is nested within students (level two), and nested within faculties (level three). This study produced the three best three-level regression models: the multilevel regression model, the multilevel regression model with natural logarithmic transformation, and the multilevel binary logistic regression model. The multilevel regression model and the multilevel regression model with natural logarithmic transformation at a significant level of 5%, have the same variables that affect student GPA scores, including semesters, credits, gender, scholarships, and marital status with the same interaction effect, namely semester interactions with scholarships. In addition, the ICC values by the two models are also the same which explains that 91% of the total diversity of student GPA comes from the student level and 8% comes from the faculty level. For the multilevel binary logistic regression model, all explanatory variables affect GPA without involving interaction between levels.
Abstrak
Regresi multilevel merupakan salah satu metode yang digunakan untuk menganalisis struktur data hirarkhi. Salah satu kasus data dengan struktur hirarki adalah data indeks prestasi kumulatif (IPK) mahasiswa tiap semester (level satu) yang tersarang dalam mahasiswa (level dua), tersarang dalam fakultas (level tiga). Dalam penelitian ini menghasilkan tiga model regresi tiga level terbaik yaitu model regresi multilevel, model regresi multilevel dengan transformasi logaritma natural, dan model regresi logistik biner multlevel. Model regresi multilevel dan model regresi multilevel dengan transformasi logaritma natural pada taraf nyata 5%, memiliki peubah sama yang berpengaruh terhadap nilai IPK mahasiswa antara lain semester, SKS, jenis kelamin, beasiswa, dan status nikah dengan pengaruh interaksi yang sama yaitu interaksi semester dengan beasiswa. Selain itu, nilai ICC oleh kedua model tersebut juga sama yang menjelaskan bahwa 91% total keragaman IPK mahasiswa berasal dari level mahasiswa dan 8% berasal dari level fakultas. Untuk model regresi logistik biner multilevel semua peubah penjelas berpengaruh terhadap IPK tetapi tanpa melibatkan interaksi antar level.
References
Aliyudin F, Budyanra B. 2016. Faktor-Faktor yang Memengaruhi Komplikasi Persalinan Wanita Usia Subur di Indonesia Menggunakan Data SDKI 2012 (Aplikasi Analisis Regresi Logistik Biner Multilevel). Jurnal Aplikasi Statistika & Komputasi Statistik. 8(2):14–14. doi:10.34123/JURNALASKS.V8I2.53.
Azen R, Walker CM. 2011. Categorical Data Analysis for the Behavioral and Social Science. New York: Routledge.
Halim M. 2009. Identifikasi Faktor-Faktor yang Berperan Terhadap Pencapaian Indeks Prestasi Kumulatif Mahasiswa Departemen Statistika IPB. 2009: Institut Pertanian Bogor . [diakses 2023 Mar 8]. http://repository.ipb.ac.id/handle/123456789/60158.
Hosmer DW, Lemeshow S. 2000. Applied Logistic Regression. second edition. Hoboken, NJ, USA: John Wiley & Sons, Inc.
Hox J. 2002. Multilevel analysis techniques and applications. Quantitative methodology series. Mahwah, NJ, US: Lawrence Erlbaum Associates Publishers.
Hox JJ. 2010 Apr 26. Multilevel analysis: Techniques and applications: Second edition. Multilevel Analysis: Techniques and Applications: Second Edition., siap terbit.
Muspawi M, Pratama R, Sarlles M. 2020. Kontribusi Praktek Kerja Industri Dan Kemampuan Akademis Terhadap Persepsi Siswa Tentang Kesiapan Memasuki Dunia Kerja. Jurnal Kependidikan: Jurnal Hasil Penelitian dan Kajian Kepustakaan di Bidang Pendidikan, Pengajaran dan Pembelajaran. 6(3):490. doi:10.33394/jk.v6i3.2772.
Nalim N, Dewi HL, Safii MA. 2021. Analisis Faktor-Faktor yang Mempengaruhi Keberhasilan Studi Mahasiswa di PTKIN Provinsi Jawa Tengah. Jurnal Kependidikan: Jurnal Hasil Penelitian dan Kajian Kepustakaan di Bidang Pendidikan, Pengajaran dan Pembelajaran. 7(4):1003–1013. doi:10.33394/jk.v7i4.3430.
Ohyver M. 2013. Penerapan Metode Transformasi Logaritma Natural dan Partial Least Squares Untuk Memperoleh Model Bebas Multikolinier dan Outlier. Jurnal Mat Stat. 13 Januari:42–51.
Susanto CE. 2015. IPK Tinggi Mudah Kerja. [diakses 2023 Mar 7]. https://mediaindonesia.com/humaniora/5232/ipk-tinggi-mudah-kerja.
Yuniar D, Mulyati H, Cahyadi ER. 2019. Faktor-faktor yang mempengaruhi penyelesaian masa studi program pascasarjana di Institut Pertanian Bogor. Jurnal Akuntabilitas Manajemen Pendidikan. 7(2). doi:10.21831/amp.v7i2.25084.
Published
How to Cite
Issue
Section
Citation Check
License
Copyright (c) 2024 Al-Khwarizmi : Jurnal Pendidikan Matematika dan Ilmu Pengetahuan Alam
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Authors who wish to publish and disseminate their papers with the Al-Khwarizmi : Jurnal Pendidikan Matematika dan Ilmu Pengetahuan Alam, shall agree to the publishing rights set by Al-Khwarizmi : Jurnal Pendidikan Matematika dan Ilmu Pengetahuan Alam. Authors understand that they shall assign publication rights as part of the process upon acceptance for publication. The authors agreed that they would transfer certain copyrights to Al-Khwarizmi : Jurnal Pendidikan Matematika dan Ilmu Pengetahuan Alam. Consecutively, authors still retain some rights to use and share their own published articles without written permission from Al-Khwarizmi : Jurnal Pendidikan Matematika dan Ilmu Pengetahuan Alam.
Authors granted Al-Khwarizmi : Jurnal Pendidikan Matematika dan Ilmu Pengetahuan Alam the following rights; (1) the right to publish and provide the manuscripts in all forms and media for publication and dissemination, (2) the authority to enforce the rights in the manuscript, for example in the case of plagiarism or in copyright infringement.
Al-Khwarizmi : Jurnal Pendidikan Matematika dan Ilmu Pengetahuan Alam will follow COPE's Code of Conduct and Best Practice Guidelines for Journal Editors to protect the research results and take allegations of any infringements, plagiarisms, ethical issues, and frauds should those issues arise. The manuscript is attributed as the authors' work, and is properly identified.
