Analisis Faktor-Faktor yang Memengaruhi Angka Partisipasi Kasar SMA/Sederajat di Indonesia Menggunakan Regresi Ridge

Abstract

One indicator of education in Indonesia is the gross enrollment ratio (GER). Based on data from Statistics Indonesia, the GER at the senior high school level in Indonesia is still low compared to the GER at primary and junior high schools. Summarizing the findings of prior studies, the factors affecting GER include the number of schools, percentage of poor population, education budget, and student-teacher ratio. Therefore, this study aims to examine the variables that affect GER in Indonesian senior high schools. Since multicollinearity between the predictor variables was identified, ridge regression was employed. This study found that the number of senior high schools, the percentage of the poor population, and the education budget simultaneously had a significant effect and contributed 71.39% to the GER at the senior high school level in Indonesia. It is remarkable to observe that, partially, the number of senior high schools, the percentage of poor people, and the education budget had no direct effect on the GER. Additionally, there was a positive correlation between the variables and GER. The number of senior high schools and the education budget have a favorable impact. In contrast, the percentage of poor people has a negative effect on GER. Meanwhile, the student-teacher ratio does not have a linear relationship with the GER at the senior high school level in Indonesia.

Metode Response Based Unit Segmentation Partial Least Square pada Model Partial Least Square Path Modeling

“…Pada saat banyak variabel yang terlibat sebagai prediktor, maka regresi berganda menjadi pilihan yang populer. Seperti pemodelan Angka Partisipasi Kasar (APK) untuk tingkat SMA [2], pengaruh motivasi dan sarana sekolah terhadap hasil belajar siswa [3], dan indeks pembangunan pendidikan pada daerah tertinggal di Indonesia [4]. Namun, pada saat melihat banyak variabel yang belum tentu dapat ditinjau secara langsung, model regresi berganda menjadi kurang tepat untuk digunakan.…”

“…Pada clustering REBUS PLS Dendogram menghasilkan unit observasi menjadi 2 kelas/segmen dan berdasarkan nilai CM segmen 1 terdapat 28 dan segmen 2 terdiri dari 6 Provinsi. REBUS PLS mampu mendeteksi adanya heterogenitas , terlihat dari nilai 𝑅 2 pada masing-masing segmen yang terbentuk yaitu local model menjadi lebih besar untuk tiap laten endogennya (kecuali nilai 𝑅 2…”