The reliability of the growth rate parameter estimator first presented
by J. B. Willet for a linear growth model is derived and critically e
xamined. The effects on the reliability of the growth rate of the inte
rvals between the longitudinal measurements and the covariances betwee
n the longitudinal measurements are distinguished. It is demonstrated
that, mathematically, the latter and not the former functionally deter
mines the reliability of the growth rate, but it is also recognized th
at, practically, the time intervals may affect the covariances.