Hybrid Logistic Regression and Random-Response Model in a Matched Pairs Study
| Autor: | Mei-Chi Li, 李梅琪 |
|---|---|
| Rok vydání: | 2006 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 94 For two dependent samples, each observation in one sample pairs with an observation in the other sample. A two-way table having the same categories for both classifications summarizes such data. This proposal presents analyses of square contingency tables with matched pairs data. However, in survey sampling, persons being interviewed often refuse to answer or give correct answer to sensitive questions that may embarrass them or be harmful to them in some way. For instance, some persons may not respond truthfully to political questions such as, "Did you change your mind in the presidential election in Taiwan after the 319 gun shot?" The method was first proposed by S. L. Warner (1965) called the random-response model. Designate the people in the population who have or do not have the characteristic of interest as groups A and B, respectively. Thus each person in the population is in either group A or group B answered only one question. We present a method of estimating the proportion of people who have some characteristic of interest without obtaining direct answers from the people interviewed with matched pairs data and then test its difference. Furthermore, the baseline-category logit model is introduced as well to see if there is any substantial difference between two groups as the independent variable changes along increase its units. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|