Joint Calibration Estimator for Dual Frame Surveys.

Dual Frame Estimation;Calibration;Social Sciences (General);Statistics and Numeric Data;Social Sciences;Survey Methodology

Elkasabi, Mahmoud AhmedLee, Sunghee ;

University of Michigan

关键词: Dual Frame Estimation;    Calibration;    Social Sciences (General);    Statistics and Numeric Data;    Social Sciences;    Survey Methodology;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/99941/mkasabi_1.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Dual frame surveys are becoming more common in survey practice due to rapid changes in the cost of survey data collection, as well as changes in population coverage patterns and sample unit accessibility. Many dual frame estimators have been proposed in the literature. Some of these estimators are theoretically optimal but hard to be applied in practice, whereas the rest are applicable but not as optimal as the first group. All the standard dual frame estimators require accurate information about the design domain membership. In this dissertation, a set of desirable properties for the dual frame estimators is specified. These properties are used as criteria to evaluate the standard dual frame estimators. At the same time, the Joint Calibration Estimator (JCE) is proposed as a new dual frame estimator that is simple to apply and meets most of the desirable properties for dual frame estimators. In Chapter 2, the JCE is introduced as an approximately unbiased dual frame estimator, with a degree of unbiasedness depending on the relationship between study variables and auxiliary variables. The JCE achieves better performance when the auxiliary variables can fully explain the variability in the study variables of interest or at least when the auxiliary variables are strong correlates of the study variables. The JCE for point estimates can be applied by standard survey software and can easily be extended to multiple frame survey estimation. In Chapter 3, the JCE properties are explored in the presence of the nonresponse error. Theoretically and empirically, the JCE proves to be robust to nonresponse error as long as a strong set of auxiliary variables is used. This strong set should predict both the response mechanism and the main study variables. Finally, the effect of domain misclassification on the dual frame estimators is discussed in Chapter 4. Since the JCE does not require domain membership information, it tends to be robust against domain misclassification even if domain totals are included in the calibration auxiliary variables.