Weighting national survey data for Bangladeshi population / Ferdous Hakim [and three others].
National-level data need to be treated for representation of population. Weighting the data yields the sample more representable. The authors share the experience from weighting of data of several nationally representative health surveys from the same frame and following similar sampling procedure....
Saved in:
| Online Access: |
Full Text (via SAGE) |
|---|---|
| Main Author: | |
| Format: | Electronic eBook |
| Language: | English |
| Published: |
London :
SAGE Publications Ltd.,
2020.
|
| Series: | SAGE Research Methods Cases : Medicine and Health.
|
| Subjects: |
| Summary: | National-level data need to be treated for representation of population. Weighting the data yields the sample more representable. The authors share the experience from weighting of data of several nationally representative health surveys from the same frame and following similar sampling procedure. A step-by-step method described here for weighting is followed in several national surveys in Bangladesh. The aim of the procedure is to compute a "final weight." For surveys having 90% or more overall response rate adjustments for the nonresponse and if the census data are not available, population calibration adjustments are not applied. In this situation, base weight suffices as the "final weight." The probabilities of selection for primary sampling units, households, sex randomization, and individuals are calculated initially. Product of the inverse probabilities yields the "base weight." In the National Mental Health Survey 2019, sum of the base weights was 79,422,101. We computed the nonresponse weight and the population calibration and adjusted these with the base weight. All of these calculations are domain specific. After the two adjustments, the final weight stands at 102,948,678. The projected population of Bangladesh was 102,161,911. The weighting procedure is a step-by-step technique which is an extensive, time-consuming, and meticulous job. It requires thorough understanding of sampling probabilities and nonresponses. Weighting of national-level surveys yield a nationally representative and comparable data at the cost of losing little precision. |
|---|---|
| Physical Description: | 1 online resource : illustrations. |
| Bibliography: | Includes bibliographical references and index. |
| ISBN: | 9781529743623 1529743621 |
| Source of Description, Etc. Note: | Description based on XML content. |