Affiliations: [a] Department of Economics, University of California San Diego, La Jolla, CA, USA | [b] Department of Sociology, University of California Riverside, Riverside, CA, USA
Correspondence:
[*]
Correpsonding author: Jeff Tayman, Department of Economics, University of California San Diego, 9500 Gilman Drive, La Jolla, California, USA. Tel.: +1 858 581 6710; E-mail:[email protected]
Abstract: In this study we examine the regression-based ratio-correlation method and
suggest some new tools for assessing the magnitude and impact of coefficient
instability on population estimation errors. We use a robust sample of 904
counties from 11 states and find that: (1) coefficient instability is not a
universal source of error in regression models for population estimation and
its impact is less than commonly assumed; (2) coefficient instability is not
related to bias, but it does decrease precision and increase the allocation
error of population estimates; and (3) unstable coefficients have the
greatest impact on counties under 20,000 in population size. Our findings
suggest that information about the conditions that affect coefficient
instability and its impact on estimation error might lead to more targeted
and efficient approaches for improving population estimates developed from
regression models.
Keywords: Population estimation, ratio-correlation method, coefficient instability
