Bootstrapping the LIS: Statistical Inference and Patterns of Inequality in the Global North

The problem of statistical inference has long been associated with quantitative inequality research. Within the last five years, however, significant developments have occurred in both the theory and practice of conducting formal statistical inference with common measures of inequality such as the Gini index. These new techniques involve the use of Monte Carlo, bootstrap resampling plans that seek to recover the standard error and sampling distribution of inequality estimates directly through the empirical distribution of the sample data, thereby facilitating statistical inference via confidence interval estimation and hypothesis testing. Using the income survey of the Luxembourg Income Study (LIS) project, this paper provides an analytical evaluation of the bootstrap procedure in the context of comparative inequality research, and uncovers patterns of distributional change in the global North over the last two decades. While it is now generally accepted that inequality has increased in the United States and United Kingdom during this period, the extent to which other wealthy nations have been able to avoid this trend has generated some debate. The paper presents new evidence to address this discussion, demonstrating along the way how the ability to conduct formal statistical inference with the Gini index provides an effective and important new evaluative tool. The paper provides an informative analysis of current methodological developments in inequality research, and demonstrates how they may be applied in the specific context of the LIS, but also can be used as a practical guide for handling the problems of statistical inference in more general social scientific settings.