A recently introduced resampling method for determining the contribution of
measurement errors to the standard error in bilinear and trilinear model e
stimates is critically examined. Using Monte Carlo simulations, it is shown
that this method does not work as intended. A noise addition method is pro
posed as a generally applicable alternative. The principles involved in app
lied work are illustrated on real UV-vis data taken from the literature. Co
pyright (C) 2001 John Wiley & Sons, Ltd.