MEASURING UNCERTAINTY IN POPULATION PROJECTIONS GENERATED BY THE HAMILTON-PERRY VARIANT OF THE COHORT-COMPONENT METHOD

In an ex post facto test of a modification of the Root Mean Square Error approach to developing 66% confidence intervals for population forecasts made using the Hamilton-Perry Method, we find that in a sample of nine states (one from each of the nine census division), that the intervals contain the 2010 population in 126 of 162 observations (77.8%) and similar results for an ex post facto test against 2000 data. We find the results encouraging. The paper describes the Root Mean Square Error approach as well as the data and the Hamilton-Perry Method.