Meta-Analysis of Sample Survey Data

• Activity Number: 518
• Type: Contributed
• Date/Time: Wednesday, August 1, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Section on Government Statistics
• Author(s): Abera Wouhib*+ and Joe Fred Gonzalez
• Companies: National Center for Health Statistics and National Center for Health Statistics
• Keywords: meta-analysis ; Survey data

Abstract

Meta-Analysis is a method for combining summary measures from similar studies. This technique can be applied to randomized controlled experiments as well as to sample surveys; however, there are some issues when applying meta-analysis to randomized controlled experiments. Some of these issues include: applying large sample theory to small sample experiments; inappropriate claims of generalizability of results to large populations; and the existence of publication biases due to the submission of selected (significant) experimental results to scientific publications. In general, the above issues may not apply to well-designed sample surveys. For most surveys, published point estimates and confidence intervals are based on large sample theory and are generally reliable because of the inherent large sample sizes, high coverage of the target population, and high response rates, which ensure the representativeness of the sample. Hence, publication bias may not be an issue using meta-analysis with survey data. In this paper, we will demonstrate an application of meta-analysis to public health surveys.