Affiliations: [a] Arkansas Institute for Economic Advancement,
University of Arkansas at Little Rock, 2801 South University, Little Rock,
Arkansas 72204-1099, USA | [b] Pacific Northwest Laboratories, P.O. Box 999,
Richland, Washington 99352, USA | [c] Currently with the U.S. Department of Energy.
Note: [1] This paper came as an unintended consequence of work on the Hanford
Environmental Dose Reconstruction Project, Subcontract No. 041581-A-k1. U.S.
Department of Energy. An earlier version of this paper was presented by
David Swanson at an invited lecture organized by the Demography Division and
Communications Division, Statistics Canada, February 11, 1991, Simon A
Goldberg Conference Center, Ottawa, Ontario. The authors are deeply grateful
for comments provided by the staff of the Demography Division on this paper,
and, in particular, to Ravi B. P. Verma. Steve Boschee assisted with
computer programming and Peggy Jobe typed an early version of the
manuscript. Carolyn Farr typed the final copy.
Abstract: This paper proposes a new method for short-term county population projections. It
is based on a modification of the ratio-correlation method of population
estimation. The modified ratio-correlation method can produce projections with a
high potential for accuracy without requiring substantial data and intensive
intellectual labor inputs. Tests of accuracy are examined for the modified
ratio-correlation method and two currently available alternatives using data
from Washington state. The tests suggest that the new method performs well.
Advantages of the modified ratio-correlation method are discussed, with
particular attention given to the formal measurement of uncertainty. Forecast
intervals are constructed and examined for the projections constructed for
counties in Washington state. Given certain limitations, the forecast intervals
appear to be robust in terms of providing accurate assessments of the precision
associated with county population projections made using the modified
ratio-correlation method.
