<![CDATA[General Linear Mixed Model (GLMM) Bi-respon was an alternative solution for longitudinal data with bi-responses which joining fixed effects, random effects and vector of realization of bi-responses process into single statistical model. GLMM can overcome the correlation between observations in longitudinal data for the response in the form of continous data. In each formation GLMM model beginning with the determination of a tentative model through exploration of data. Exploration data covering several aspects of the individual profile, average structure, variance structure, and correlation structure. Building GLMM was done by selecting fixed effects under using Maximum Likelihood (ML) method, and the selection of variance components (the number of random effects) using Restricted Maximum Likelihood (REML) method. Based on the comparison of AIC value, Diabetes Mellitus Type 2 disease data was better to be modeled using GLMM with one response. Cross correlations matrix elements were about 0.3 to 0.6 and produced unstructured covariance. Correlation coefficient between two responses was 0.5526 and produced unstructured covariance.]]>
Copyrights © 2011
