ABSTRACT
Given the properties of the intrinsic estimator, including robust trend estimation with finite samples and consistent estimation with diverging samples, the intrinsic estimator has the potential to analyze age-period-cohort (APC) data and meet the expectation of the investigators. The author demonstrates the intrinsic estimator method with the data sets of the lung cancer mortality among the US males, and the Human Immunodeficiency Virus (HIV) mortality data among the US males and females. He then demonstrates the constraint estimator using the lung cancer mortality data among the US females, and compares the results between the intrinsic estimator and the constrained estimators, using the equality constraints that are commonly used in the literature and the non-contrast constraints that are recently discovered in Fu. The APC models use more degrees of freedom than the AP and AC models, and are expected to achieve a better goodness-of-fit. The author examines the goodness-of-fit of the two models, the linear model and the loglinear model.
