Author

Reni Hiola

Subject

- Pendidikan

Abstract

In research of epidemiology, structural equation modeling (SEM) has been become very popular, especially for latent variables. In SEM there are assumptions that must include the data should be normally distributed multivariate and a used large of data. For overcome these problems required the alternative approach of SEM based variance or partial least square (PLS). SEM-PLS does not require an assumption that a lot. In health sector randomization is not possible, because it concerns the lives of humans. So that assumptions independent can’t be achieved. This can lead to imbalances covariates and selection bias. Therefore, to overcome these problems applied propensity score (PS). This method is a statistical analysis that can be used to analyze study design Non-Experimental where can’t do randomization to treatment groups. Furthermore, as suggested new methods for handling selection bias is a marginal meanweighting through stratification (MMW-S). The analysis result obtained when using MMW-S is powerful because MMWS show strong reduction in of selection bias. The author uses an innovative method by using empirical data HIV/AIDS. Briefly using MMW-S with a predisposition, clinical manifestations, and opportunistic infection. And adherence to antiretroviral (ARV) as a confounding variable. The results showed that the method of MMW-S can removed bias more than 93.5%.

Publisher

Faculty of Mathematics and Science Yogyakarta State University

Contributor

-

Publish

2016

Material Type

PROSIDING

Right

-