Abstract
The influence of finite-size effects on the determination of the critical exponent of the magnetization β is critically discussed. We show that the usual fitting procedure, of fitting a power law to experimental data, is not correct for finite-size disturbed data and only accidentally results in nearly correct values for the critical exponent. The critical temperature resulting from a fit procedure is always too high, because the fitting range includes finite-size disturbed data. A new method is proposed, which circumvents the problems arising in fitting a power law to the experimental magnetization data. This method uses an effective critical exponent and allows the determination of β and Tc even from finite-size disturbed data. The method will be reliable for evaluating experimental data in the currently interesting field of confined magnetic films.