|
Activity Number:
|
342
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 6, 2003 : 9:00 AM to 10:50 AM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #301338 |
|
Title:
|
Effects of Masked Survey Design Variables on SUDAAN and SURVEYMEANS Variances
|
|
Author(s):
|
Esther Hing*+ and Sarah W. Gousen and Catharine W. Burt and Iris M. Shimizu
|
|
Companies:
|
National Center for Health Statistics and and National Center for Health Statistics and Center for Disease Control and Prevention
|
|
Address:
|
6525 Belcrest Rd., Hyattsville, MD, 20782-2003,
|
|
Keywords:
|
complex survey design ; variances ; public use files
|
|
Abstract:
|
Complex multistage sample designs are associated with higher design effects and variances than simple random sample surveys. Lack of survey design information has hampered researchers using the National Ambulatory Medical Care Survey (NAMCS) and National Hospital Ambulatory Medical Care Survey (NHAMCS) in the past. The NAMCS is an annual survey of visits to office-based physicians, while the NHAMCS annually surveys visits to hospital outpatient and emergency departments. In 2002, masked survey design variables were released with the 2000 public use NAMCS and NHAMCS files. This study compares the ratio of alternative variances derived using masked survey design variables with SUDAAN variances based on unmasked survey design variables reflecting all stages of sampling (4-stage). Three alternative masked variance estimates are compared: 4-stage Taylor series estimates (SUDAAN) and ultimate cluster estimates (1-stage SUDAAN and SAS's SURVEYMEANS procedure). Ratios were computed for samples of 130-140 estimates in each of the three settings included in the two surveys. On average, masked alternative standard errors for visits were 1-6 percent higher than unmasked estimates.
|
