| Summary: | 碩士 === 淡江大學 === 數學學系碩士班 === 100 === When the measurement error and mixed effect appear in the model at the same time, we can not find much discussion on the literature. The main reason is that the marginal distribution of the integral to the random effect is no longer a generalized linear model. This paper discussed the estimated method between measurement error and mixed effect in the log-linear and logistic model. In the log-linear model, the estimation method usually included naive, regression calibration, simulation extrapolation, small measurement error approximation, and there is another estimation method "Weighted and Corrected Score Function" which is weighted, corrected and weighted again under replication situation. The logistic model in addition to use the integral to obtain the marginal distribution, it also used the moment constructed estimated equation to estimate and compared between partial calibration and without calibration under replication situation. At last, it used the computer to simulate the estimated method which was brought up in this paper.
|
|---|
