Premium
A two‐step method for understanding and fitting growth curve models
Author(s) -
Stanek Edward J.
Publication year - 1990
Publication title -
statistics in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.996
H-Index - 183
eISSN - 1097-0258
pISSN - 0277-6715
DOI - 10.1002/sim.4780090714
Subject(s) - univariate , growth curve (statistics) , computer science , transformation (genetics) , extension (predicate logic) , multivariate statistics , curve fitting , set (abstract data type) , simple (philosophy) , econometrics , statistics , mathematics , biochemistry , chemistry , philosophy , epistemology , gene , programming language
With data from a repeated measures design, analysts often overlook growth curve analysis based on the Potthoff‐Roy model. The reluctance to perform a growth analysis may result partially from unfamiliarity with the technique, and partially from the lack of readily available computational programs to fit the models. Although growth curve analysis has traditionally appeared as a special technique in multivariate analysis, one can fit the models with the use of two simple ideas: transformation of variables, and weighted least squares (WLS). When presented in this manner, growth models are a natural extension of univariate repeated measures models. Growth curve analyses share with univariate analyses the common objective to limit estimation to a parsimonious set of parameters that characterize the response profiles. One can readily fit models with commonly available software. The two‐step methodology also allows the fit of more general growth models that are appropriate when groups have different shape profiles. Two examples illustrate this approach to growth curve analysis.