Growth Curve Modeling: An Alternative Approach to Estimating Patterns of Population Change

Curtis, Katherine, and Jerald R. Herting
Working paper no. 2004-18

Abstract

Working from the formal model of population growth, Pt = P0ert, we use population change in the U.S. Great Plains between 1900 and 1930 to demonstrate the efficiency of an alternative approach to estimating population change, r: growth curve modeling (GCM). Through this method, the researcher treats processes of change as a trajectory rather than a series of panels, resulting in a more efficacious and parsimonious analysis. Three general aspects of population growth can be addressed through linear and non-linear GCM techniques, namely (1) establishing a mean growth rate that represents the general pattern of change across all units, (2) estimating the degree of local level variation around the average rate, and (3) analyzing the relative contribution of factors influencing this variation. Our findings reveal that the non-linear cubic model most appropriately suits the early years of development on the Great Plains, indicating that the rate of population change varied over the period with rapid growth at the outset followed by a decreasing rate of growth in the middle decade and an accelerating rate during the later portion. In addition, we observed considerable variation around the average growth rate, and attribute much of this to county settlement date, urban proximity, and economic base, each to a varying extent across the different models.