The area ratio method of Sjöstrand is generally considered one of the most reliable reactivity determination methods and thus is a major candidate for off-line calibration purposes in future accelerator-driven systems for high-level waste incineration. In this work, the Sjöstrand area ratio method has been evaluated experimentally under thorough conditions in the strongly heterogeneous subcritical facility YALINA-Booster. Both strengths and weaknesses of the method have been identified. Most surprisingly, it has been found that the area ratio reactivity estimates may differ a factor of 2 depending on detector position. It is also shown that this strong spatial dependence can be explained based on a simple two-region point-kinetics model and corrected by means of correction factors obtained through Monte Carlo simulations. A new Monte Carlo correction method is proposed that includes, at the same time, the spatial disturbance and the effective delayed neutron fraction. In that way, the value of the effective multiplication factor is obtained from the measured dollar reactivity without the need of calculating the effective delayed neutron fraction explicitly, and thereby, the delayed neutron transport is performed only once. Further, it has been found that the Sjöstrand area ratio method is not sensitive to perturbations of the source multiplication factor.