669 – Variance Estimation in Sample Surveys
Variance Estimation for the Multidisciplinary Treatment Planning (MTP) Survey
Irene Prabhu Das
National Cancer Institute
Hyunshik James Lee
Westat
Benmei Liu
National Cancer Institute
The Multidisciplinary Treatment Planning (MTP) Survey will be conducted using an instrument jointly developed by the National Cancer Institute (NCI) and the Commission on Cancer (CoC) to survey CoC-accredited hospitals in the United States on how multidisciplinary treatment planning is offered to cancer patients for one of six cancer types (Brain/CNS, Lung, Head & Neck, Gynecologic, Gastrointestinal, and Breast cancer). This survey used a census of facilities and only one cancer type is randomly selected for each facility by an unequal probability sampling method using case volume as the basis to determine the selection probability. The sampling method poses a unique challenge for variance estimation because a single cancer type is selected independently from each facility, which can be regarded as the sampling stratum because each facility is included with certainty. It requires combining facilities into variance strata and special handling of nonresponse. This paper presents a theoretical framework for variance estimation and the results of a simulation study of the proposed variance estimator based on the jackknife technique.