Estimasi Multilevel Structural Equation Modeling (Multilevel Sem) Dengan Pendekatan Em-Algorithm (Studi Kasus: Remunerasi Tenaga Kependidikan Di Lingkungan Its Surabaya Tahun 2015)

Susiani, Farisca (2016) Estimasi Multilevel Structural Equation Modeling (Multilevel Sem) Dengan Pendekatan Em-Algorithm (Studi Kasus: Remunerasi Tenaga Kependidikan Di Lingkungan Its Surabaya Tahun 2015). Masters thesis, Institut Teknologi Sepuluh Nopember Surabaya.

[img]
Preview
Text
1314201029-master-theses-.pdf - Published Version

Download (1MB) | Preview

Abstract

Multilevel Structural Equation Modeling (Multilevel SEM) adalah suatu metode yang menggabungkan antara SEM dan model Multilevel secara simultan. Penggunaan metode Maximum Likelihood Estimation (MLE) merupakan metode yang umum dipakai dalam estimasi parameter Multilevel SEM namun mengalami kendala ketika diterapkan pada kasus data unbalance. Sehingga algoritma Ekspektasi-Maksimalisasi (EM) diperlukan untuk mengestimasi data unbalance. Tujuan penelitian ini adalah mendeskripsikan algoritma EM pada Multilevel SEM dengan pendekatan MLE. Selanjutnya diimplementasikan pada studi kasus Remunerasi Tenaga Kependidikan di Lingkungan ITS Surabaya. Penerapan remunerasi di kalangan PTN merupakan fenomena baru yang akan direspons secara positif maupun negatif oleh para Tendik. Efektivitas sistem pemberian remunerasi itu sendiri dipengaruhi oleh kinerja Tendik dan Kinerja Tendik dipengaruhi oleh Motivasi, Lingkungan kerja, dan Pelatihan, dimana variabel tersebut merupakan variabel yang tidak dapat diukur secara langsung. Dikarenakan perbedaan kondisi lingkungan kerja dan beban kerja setiap unit kerja maka perlu memperhatikan konteks individu dan unit kerja dalam penelitian. Populasi Tendik di ITS yang menerima remunerasi sebanyak 698 orang kemudian diambil 100 orang (14,33%) dari 10 unit kerja sebagai responden menggunakan simple random sampling dan alokasi sampel proposional. Algoritma EM dalam mencari estimator Multilevel SEM terdiri dari merekonstruksi fungsi likelihood untuk complete data, membentuk fungsi log-likelihood untuk complete data, tahap perhitungan ekspektasi dari fungsi log-likelihood dengan memperhatikan missing data¸ dan tahap maksimalisasi untuk mencari penaksir parameter yang meminimumkan fungsi log-likelihood. Pernyataan responden menggunakan skala likert dengan 5 kategori. Hasil analisis menunjukkan bahwa ketika tenaga kependidikan memiliki persepsi motivasi berprestasi yang tinggi dan karakteristik lingkungan kerja yang nyaman maka mereka cenderung lebih terpuaskan dengan pekerjaan-pekerjaannya (kinerjanya). Tenaga kependidikan juga mempersepsikan bahwa pemberian remunerasi yang diterapkan di ITS telah berbasis kinerja atau telah efektif. Selain itu tenaga kependidikan dengan semua tingkat pendidikan dan golongan memberikan tanggapan yang positif terhadap efektifitas pemberian remunerasi di ITS. =========================================================== Multilevel Structural Equation Modeling (Multilevel SEM) is a method which combines SEM and multilevel models simultaneously. Maximum Likelihood Estimation (MLE) is a commonly method used for estimating the parameters in Multilevel SEM but it encountered problems when applied to the case of data unbalance. So the algorithm Expectation-Maximization (EM) is required to estimate the unbalance data. This research will describe the EM algorithm on MLE Multilevel SEM approach. Furthermore, It will be implemented on a case study in Remuneration of Educational Staff in ITS Surabaya. Application of remuneration among collage is a new phenomenon that will respond positively or negatively by the educational staff. The effectiveness of the remuneration system itself is influenced by the performance of educational staff and the performance is influenced by motivation, work environment, and training, where the variable is a variable that can not be measured directly. Due to differences in environmental conditions of work and the workload of each unit it is necessary to consider the context of individuals and work units in the study. The data population consisted of 698 administrative staff who get a remuneration while the respondents were 100 (14,33%) from 10 subject area using simple random sampling and proportional sample allocation. To find estimator Multilevel SEM, EM algorithm consists of reconstructing the likelihood function to the complete data, forming the log-likelihood function for the complete data, the calculation step expectation of the loglikelihood function with due regard to missing data and maximization step to find the parameter estimator which minimizes log-likelihood function. The opinion was expressed on a five ordered likert scale. The results showed that when educational staff have a perception of high achievement motivation and characteristics of the work environment was comfortable then they tend to be more satisfied with their performance. Educational staff also perceive that the remuneration applied in ITS have a performance-based or have been effective. Besides educational staff with all levels of education and classes provide positive feedback on the effectiveness of remuneration that has been applied in ITS.

Item Type: Thesis (Masters)
Additional Information: RTSt 519.53 Sus e
Uncontrolled Keywords: Multilevel SEM, Maximum Likelihood, Ekspektasi-Maksimalisasi, Remunerasi, Tenaga Kependidikan.
Subjects: Q Science > QA Mathematics
Q Science > QA Mathematics > QA278 Cluster Analysis. Multivariate analysis. Correspondence analysis (Statistics)
Divisions: Faculty of Mathematics, Computation, and Data Science > Statistics > 49101-(S2) Master Thesis
Depositing User: EKO BUDI RAHARJO
Date Deposited: 17 Dec 2019 02:33
Last Modified: 17 Dec 2019 02:33
URI: http://repository.its.ac.id/id/eprint/72385

Actions (login required)

View Item View Item