ELSEVIER International Journal of Bio-Medical Computing 36 (1994) 3 I I-318 m GTRACK: A PC program for computing Goldstein’s growth constancy index and an alternative measure of tracking Amy M. Fureya, Charles J. Kowalskib, Emet D. Schneiderman*“, Stephen M. Willis” zyxwvutsrq ‘Department of Biostatistics, bDepariment of Biologic and Materials Sciences, University of Michigan, Ann Arbor, MI 48109, USA ‘Department of Oral and Maxillofacial Surgery, Baylor College of Dentistry, 3302 Gaston Ave., Dallas, TX 75246, USA (Received 16 September 1993; accepted IS October 1993) zyxwvutsrqponmlkjihgfedcbaZYXWVU Abstract This paper reviews Goldstein’s ‘growth constancy index,’ .$, a measure of tracking which can be used to determine whether or not individuals maintain their relative positions in the distribution of a given measurement as that distribu- tion changes over time. We suggest that F is an appropriate measure of tracking when the (standardized) measurements arise in the context of a Model I ANOVA, but that the intraclass correlation coefficient, r,, may be preferred when a Model II ANOVA is applicable. We also describe - and make available - a PC program which allows the user to choose between Model I and Model II, and computes the appropriate tracking index and confidence intervals for the corresponding parameter. Keywords: Longitudinal studies; Tracking; Repeated measurements; Growth stability 1. Introduction Goldstein’s [l] growth constancy index is a mea- sure of tracking, the tendency of individuals to maintain their relative positions in a response dis- tribution as that distribution changes over time. A number of approaches to the quantification of this phenomenon have been taken [2-41 and a good overview of these is available [5]. In addition, pro- grams implementing several of these methods have been written and made available to the biomedical * Corresponding author. research community [6-lo]. The purposes of the present paper are to describe Goldstein’s index, which we call [; to note that it is equal to a perhaps more familiar quantity, t2, the correlation ratio; to suggest that while 4 is appropriate in the context of a Model I ANOVA, the intraclass correlation coefficient, rh may be preferred in the more fre- quently encountered (in the context of tracking) case of a Model II ANOVA; and to describe and illustrate a stand-alone, menu-driven PC program, written in GAUSS zyxwvutsrqponmlkjihgfedcbaZYXW [l 11, which can be used to com- pute either or both of these quantities and confi- dence intervals for the corresponding parameters. We begin by outlining the development of [. 0020-7101/94/$07.00 0 1994 Elsevier Science Ireland Ltd. All rights reserved SSDI 0020-7101(93)00953-F