Propensity Score Dengan Pemodelan Structural Equation Model-Partial Least Square (SEM-PLS) Pada Kasus HIV/AIDS

Ningsih, Setia (2016) Propensity Score Dengan Pemodelan Structural Equation Model-Partial Least Square (SEM-PLS) Pada Kasus HIV/AIDS. Masters thesis, Institut Technology Sepuluh Nopember.

[img] Text
1314201018- Master_Thesis.pdf - Accepted Version
Restricted to Repository staff only

Download (2MB) | Request a copy

Abstract

Propensity score merupakan salah satu metode statistika yang dapat digunakan untuk menganalisis data observasional dimana Randomized Controlled Trial (RCT) tidak dapat untuk dilakukan. Metode ini digunakan untuk mengurangi bias dalam estimasi dampak dari perlakuan pada data yang bersifat observasi karena adanya faktor confounding. Dalam penelitian observasional, kovariat biasanya tidak seimbang antara kelompok perlakuan dan kelompok kontrol. Metode yang digunakan adalah propensity score marginal mean weighting through stratification untuk menggetahui effek confounding. Jika variabel yang digunakan adalah variabel laten, maka pendekatannya menggunakan structural equation model (SEM). Namun penerapan SEM berbasis covarian memiliki asumsi-asumsi yang harus terpenuhi, seperti data harus berdistribusi multivariate normal dan memerlukan jumlah sampel yang relatif cukup besar. Untuk mengatasi permasalahan dalam pemenuhan asumsi tersebut, sebagai alternatif maka dikembangkan SEM berbasis varian atau Partial least square (SEM-PLS). Pada penelitian ini, data yang digunakan yaitu data sekunder hasil rekam medis penderita HIV/AIDS. Kepatuhan terapi ARV sebagai confounding, sedangkan variabel respon yaitu infeksi oportunistik. Estimasi propensity score menggunakan pemodelan SEM-PLS. Selanjutnya, memeriksa keseimbangan kovariat antara kelompok yang patuh terapi ARV dan kelompok yang tidak patuh terapi ARV. Jika kelompok yang patuh dan tidak patuh seimbang (balance), maka dilakukan pengujian effect treatment. Hasil propensity score marginal mean weighting through stratification pada kasus HIV/AIDS diperoleh bahwa pada 3 strata effect treatment sebesar 0.0275. =========================================================================================================== Propensity score is a statistical method that can be used to analyze the observational data where Randomized Controlled Trial (RCT) is not able to do. This method is used to reduce bias in the estimation as the impact of treatment on observational data because of their confounding factors. In the observational study, covariates usually not balanced between the treatment group and the control group. The method used is propensity score marginal mean weighting through stratification to recognize the confounding effects. If the variable used is the latent variables, the approach will be using structural equation modeling (SEM). But the application of covariance-based SEM has the assumptions that must be met, such data must be normally multivariate distributed and the number of samples requiring relatively large. To solve the problems in the fulfillment of these assumptions, as an alternative solution, it is developed SEM-based variant or Partial Least Square (PLS-SEM). In this study, the data used is secondary data from medical records of patients with HIV / AIDS. ARV therapy adherence as confounding, while the response variable is an opportunistic infection. Propensity score estimation used is SEM-PLS approach. Furthermore, examine the balance between the adherent covariates ARV therapy and those who do not comply ARV therapy. If the group of adherent and non-adherent balance, then tested with treatment effect will be conducted. The results of propensity score with marginal mean weighting through stratification in the case of HIV / AIDS found that in three strata there is a different effect treatment between those obey adherent antiretroviral therapy and who do not obey ARV therapy at 0.0275.

Item Type: Thesis (Masters)
Additional Information: RTSt 519.53 Nin p
Uncontrolled Keywords: SEM-PLS, Propensity Score, MMW-S, HIV/AIDS
Subjects: Q Science > QA Mathematics > QA278.3 Structural equation modeling.
Divisions: Faculty of Mathematics and Science > Statistics > 49101-(S2) Master Thesis
Depositing User: Mr. Tondo Indra Nyata
Date Deposited: 20 Jan 2020 03:12
Last Modified: 20 Jan 2020 03:12
URI: http://repository.its.ac.id/id/eprint/72752

Actions (login required)

View Item View Item