碩士 / 淡江大學 / 統計學系碩士班 / 98 / In epidemiology studies, the logistic regression model is used popularly for inferring the relation of risk factors and a binary response variable. For rare diseases, we usually take case-control studies to improve sampling efficiency. When some major confounding variables are difficult to quantity, a matched case-control design can be adopted to eliminate biased comparisons between cases and controls. Breslow & Day (1980) use the conditional likelihood to eliminate the too many intercept terms. Relying on convenient statistical software, the logistic regression models are easily and rapidly implemented. It is, however, often misplace to examine whether a logistic model fit the data well. It should be recognized that before constructing the final model, a goodness-of-fit test of the model is still an important issue. We generalized the idea of Cheng & Chen (2004), and propose a Score type test for the logistic regression model under matched case-control data. The Score test converges weakly to a centered Gaussian process by the result of Arbogast & Lin (2004).We assess the performance of the proposed test through simulation studies in finite sample and illustrate the score test by a low birth weight study.
| Identifer | oai:union.ndltd.org:TW/098TKU05337001 |
| Date | January 2010 |
| Creators | Ching-ju Tseng, 曾靖茹 |
| Contributors | Li-Ching Chen, 陳麗菁 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 40 |
Page generated in 0.0116 seconds