<BR>We present an outline of relative distribution methods, with an<BR>application to recent changes in the U.S. wage distribution. Relative<BR>distribution methods are a nonparametric statistical framework<BR>for analyzing data in a fully distributional context. The<BR>framework combines the graphical tools of exploratory data analysis<BR>with statistical summaries, decomposition, and inference. The<BR>relative distribution is similar to a density ratio. It is technically<BR>defined as the random variable obtained by transforming a variable<BR>from a comparison group by the cumulative distribution<BR>function (CDF) of that variable for a reference group. This transformation<BR>produces a set of observations, the relative data, that<BR>represent the rank of the original comparison value in terms of<BR>the reference group’s CDF. The density and CDF of the relative<BR>data can therefore be used to fully represent and analyze distributional<BR>differences. Analysis can move beyond comparisons of<BR>means and variances to tap the detailed information inherent in<BR>distributions. The analytic framework is general and flexible, as<BR>the relative density is decomposable into the effect of location<BR>and shape differences, and into effects that represent both compositional<BR>changes in covariates, and changes in the covariateoutcome<BR>variable relationship.<BR>The authors wish to thank Annette D. Bernhardt, Marc Scott, Paul Janssen, the<BR>editor, and the anonymous reviewers for the many helpful comments that have much<BR>improved this paper.<BR>* Pennsylvania State University<BR>