The Non-uniqueness Property of the Intrinsic Estimator in APC Models

Abstract

An important property of the intrinsic estimator is discussed, which has received no attention in literature so far: the age, period, and cohort estimates of the intrinsic estimator are not unique but vary with the parameterization and reference categories chosen for these variables. We give a formal prove of the non-uniqueness property for effect coding and dummy variable coding. For data on female mortality in the U.S. over the years 1960-1999, we show that the variation in the results obtained for different parameterizations and reference categories is substantial and leads to contradictory conclusions. We conclude that the non-uniqueness property is a new argument to not routinely apply the intrinsic estimator.keywords: APC, identification problem, intrinsic estimator, contrast, reference category, effect coding B. Pelzer, corresponding author Assistant professor

Should age-period-cohort analysts accept innovation without scrutiny? A response to Reither, Masters, Yang, Powers, Zheng and Land

“…However the model does not work as a general purpose APC model; no model does. Our concerns mirror those of others regarding another the Intrinsic Estimator (Luo, 2013;Pelzer et al, 2014) and we agree with Fienberg (2013Fienberg ( :1983) that solutions to the APC identity is an endless and fruitless quest. It is surely time to move onto substantively focused considerations of the meaning of the three comp Figure 1: The DGP (black) and results (grey) from fitting the HAPC model to 100 datasets generated as in equation 1.…”

Is the United States a Counterexample to the Secularization Thesis?

“…In the past decade, the crossclassified multilevel model (Yang andLand 2006, 2013a) and the intrinsic estimator Yang et al 2008;Yang and Land 2013a) have come to be seen as the state-of-the-art tools of choice for APC analysis, but criticisms have recently been leveled against these methods (Luo 2013;Luo and Hodges 2013;Bell and Jones 2014;Pelzer et al 2015). Fortunately for our analysis, we do not need to take sides in this debate because even the main proponents of these techniques acknowledge that they are not a universal solution to the identification problem and there are situations in which they are not appropriately used (Yang andLand 2013b, p. 1971).…”

Block Constraints in Age–Period–Cohort Models with Unequal-width Intervals

“…Constrained estimators include the CGLM proposed by Mason et al (1973) and the IE introduced by Fu (2000) and Yang and associates (2008). Constrained estimators have been criticized for relying on external information to specify constraints when such information generally does not exist or cannot be verified Jones 2014a, 2014b;Glenn 1976Glenn , 2005Luo 2013a;Luo et al 2015;O'Brien 2011aO'Brien , 2011bO'Brien , 2014Pelzer et al 2015;Rodgers 1982aRodgers , 1982b. Robertson and Boyle (1986) and Boyle and Robertson (1987) proposed an alternative approach that has frequently been employed by sociologists: using individual-level data to create A, P, and C categories with unequalinterval widths.…”