Markovian Transition Probabilities: Estimation And Test Procedures
The outcome variables generated from longitudinal studies are generally correlated and do pose a formidable challenge to model the repeated measures data. It is worth noting that the relationship between dependent variables as well as between dependent and explanatory variables can reveal very usefu...
Saved in:
| 主要作者: | |
|---|---|
| 格式: | Thesis |
| 语言: | English |
| 出版: |
2013
|
| 主题: | |
| 在线阅读: | http://eprints.usm.my/43381/1/Mahboobeh%20Zangeneh%20Sirdari24.pdf http://eprints.usm.my/43381/ |
| 标签: |
添加标签
没有标签, 成为第一个标记此记录!
|
| 总结: | The outcome variables generated from longitudinal studies are generally correlated and do pose a formidable challenge to model the repeated measures data. It is worth noting that the relationship between dependent variables as well as between dependent and explanatory variables can reveal very useful information for the policy makers and researchers in various fields. In modeling for data with dependence in outcomes as well as when the outcomes are associated with the potential explanatory variables, marginal or conditional models can be used. Most of the works found in the literature are based on marginal models and only relatively few works employ the conditional models in addressing the dependence. In a different class of models, known as the quadratic exponential form, models are developed in order to take account of underlying associations among the outcome variables. The conditional models, on the other hand, are mostly based on the Markovian assumptions. These models are gaining importance increasingly due to the underlying nature of relationships among the outcome variables generated from longitudinal data. |
|---|