Number of Factors in Growth Curve Modeling

Abstract

Basic growth curve models parameterize the mean and covariance structure of a set of repeated measures by latent factors that represent the polynomial influences of time. In practice it may be hard to choose the number of factors, i.e., the order of the polynomial. Simple calculations are proposed to estimate this order.

Keywords:

• latent curve
• mean structure
• order of polynomial

Notes

¹ In the GCM path diagram, the Intercept factor has paths of 1.0 from it to all variables; the Slope factor has paths 0, 1, 2, 3… to all variables; and so on.

² This is a model of the type , where codes time, is a parameter matrix, and expands to all observations. See, e.g., Pan and Fang (2002, p. 44).

³ Although the tiny sample size would make this a problematic endeavor.