Estimation and inference of domain means subject to shape constraints

TL;DR

This paper introduces a new design-based estimator for population domain means that incorporates shape constraints, improving estimation accuracy especially for small domains, with theoretical guarantees and practical validation.

Contribution

It proposes a constrained estimator for domain means that respects shape restrictions, extending existing methods with asymptotic properties and simulation validation.

Findings

Constrained estimator is consistent and asymptotically normal.

Improves estimation accuracy over unconstrained methods, especially for small domains.

Validated with application to U.S. National Survey of College Graduates.

Abstract

Population domain means are frequently expected to respect shape or order constraints that arise naturally with survey data. For example, given a job category, mean salaries in big cities might be expected to be higher than those in small cities, but no order might be available to be imposed within big or small cities. A design-based estimator of domain means that imposes constraints on the most common survey estimators is proposed. Inequality restrictions that can be expressed with irreducible matrices are considered, as these cover a broad class of shapes and partial orderings. The constrained estimator is shown to be consistent and asymptotically normally distributed under mild conditions, given that the shape is a reasonable assumption for the population. Further, simulation experiments demonstrate that both estimation and variability of domain means are improved by the constrained estimator in comparison with usual unconstrained estimators, especially for small domains. An application of the proposed estimator to the 2015 U.S. National Survey of College Graduates is shown.