Improving Accuracy in Multiple Regression Estimates of Population Using Principles from Causal Modelling

This paper reports a mildly restricted procedure for using a theoretical causal ordering and principles from path analysis to provide a basis for modifying regression coefficients in order to improve the estimation accuracy of the ratio-correlation method of population estimation. The modification is intended to take into account temporal changes in the structure of variable relationships, a major element in determining the accuracy of post-censal estimate. The modification of coefficients is conservative in that is uses rank-ordering as a basis of change. Empirical results are reported for counties in Washington state that demonstrate the increased accuracy obtained using the proposed procedure.