This is an updated report on an earlier study that has been revisited after additional measurements were taken on the considered permanent sample plots. Thus, much of the material in this report replicates the original publication intended to demonstrate statistically robust methods for site-dependent height model development. The choice of form of the model is treated in this study as a secondary issue, while the main focus is on the conceptual aspect of parameter estimation for self-referencing models, such as site index models. The data considered came from the Consortium for Accelerated Pine Production Studies (CAPPS) at the Warnell School of Forest Resources. In this paper, we present two practical approaches to base-age invariant and non-biased parameter estimation and compare the results to those obtained from base-age dependent methods. Unlike the traditional methods, the base age invariant techniques do not violate regression assumptions and produce identical results regardless of the choice of base age or applied algorithm.