Setting Bounds and Using Old Results to Produce New Solutions in Age-Period-Cohort Multiple Classification Models

• In Event: 4150 - Topics in Sociological Methodology, Session 1

Abstract

The best fitting solutions to age-period-cohort multiple classification (APCMC) models lie on a line in multidimensional solution space. This means that there are an infinite number of best fitting solutions for an APCMC model. If a researcher has one of these best fitting solutions, then they can generate any of the infinity of best fitting solutions. This is possible because the line of solutions is a function of any one of the solution plus the product of a scalar times the null vector associated with the matrix of independent variables. By manipulating the scalar researchers can obtain any of the best fitting solutions. I show how this technique can be used with solutions from past research to produce solutions that refine and/or challenge the results of past research; how it can be used to produce upper and lower bounds estimates for the coefficients of ages, periods, and cohorts in APCMC models; and how it can be used to investigate the effects of parameterizations on solutions that are perpendicular to the null vector.

Author

• Robert M. O'Brien, University of Oregon