Measures of inequalities in sampling from finite population
Project responsable Yves Tillé
Team member Matti Langel
Abstract Main Goals of the project: Measuring inequality is an important issue to apprehend economical, social or political questions. Inequality indexes are complex statistics which are most often calculated using samples and therefore, require specific methods to estimate their variances. The main goals of this project is to provide new insights to this question, and more generally, to propose solutions in order to improve the quality of inequality measurements in the sampling from finite population context. The project in its context: In 2004, Eurostat launched a new instrument for measuring the Social Income and Life Condition (EU-SILC). Switzerland joined this program in 2006 whose aim is to give comparable national estimates for Laeken poverty and inequality indicators. Laeken indicators of social inclusion are a set of 18 measures of inequality, poverty and social inclusion selected at the Laeken European Council in 2001. The Laeken indicators were proposed for their simple interpretation. However, from a statistical point of view, it seems important to evaluate the adequacy of this choice and to eventually propose several alternatives. Our project can lead to important results, allowing for the estimation of inequality measures by taking every precaution recommended by statistical deontology. The determination of confidence intervals for measures of inequality will truly help to determine levels of inequality and detect changes in these levels. Improving accuracy in the measurement of inequalities through survey sampling is of major importance, especially in official statistics. Methodology and scientific context: It can be observed that only a small fraction of the vast literature on inequality measures concerns estimation by means of a sample selected from a finite population. In most cases, measures of inequality are however estimated by using complex sampling and estimation methods. To provide credible statistical answers to questions related to inequality (for example : Is the level of inequality increasing?), the estimation of variance and the construction of confidence intervals are imperative. Therefore, the estimation of variance will be a major focus of this study. Other perspectives will also give shape to the predominant aim of the project: - Detailed analysis of a large variety of measures of inequality in the sampling from finite population context; - Computation of the variance of the measures, critical evaluation of these indexes, proposal of new robust measures; - Development of new calibration techniques to improve estimations; - Development of specific sampling designs for the estimation of particular indexes.
Keywords Sampling theory, Measures of inequalities, sampling, inequality, Gini, quantiles, variance estimation, Laeken, linearization, influence function
Type of project Fundamental research project
Research area Mathématiques
Method of financing FNS - Encouragement de projets (Div. I-III)
Status Completed
Start of project 1-3-2009
End of project 30-6-2011
Overall budget 99'962.00
Additional info http://p3.snf.ch/projects-121604#
Contact Yves Tillé