The iterative proportional fitting algorithm and the NM-method: solutions for two different sets of problems

Abstract: In this paper, we identify two different sets of problems. The first covers the problems that the iterative proportional fitting (IPF) algorithm was developed to solve. These concern completing a population table by using a sample. The other set concerns constructing a counterfactual population table with the purpose of comparing two populations. The IPF is commonly applied by social scientists to solve problems not only in the first set, but also in the second one. We show that while it is legitimate to use the IPF for the first set of problems, it is not the right tool to address the problems of the second kind. We promote an alternative of the IPF, the NM-method, for solving problems in the second set. We provide both theoretical and empirical comparisons of these methods.