|
Abstract:
|
The relative concentration index (RCI) is a widely used and attractive measure for measuring socioeconomic inequalities in health. First proposed by economist Nanak Kakwani in 1977, this index has its sampling distribution derived in 1997 also by Kakwani based on the assumption that the data comes from simple random samples. Recently, health surveys are increasingly recognized as valuable sources for assessing disparity for a wide range of health behaviors. However, estimators of RCI that incorporate complex samplings features, such as stratification, clustering, or unequal probability sampling, are not available for use in making valid inference about the direction and magnitude of relative disparity across socioeconomic groups. In this presentation, we proposed a linearization approximation approach to derive its variance estimator under complex survey designs. We also proposed an approximate estimator which only requires aggregate data for calculating the variance. Both variance estimators are evaluated using simulations studies. We illustrated its use on estimating seriocomic disparities on obesity using data drawn from the National Health and Nutrition Examination.
|
