The Feynman-α method is an in core experiment, aimed to measure the decay coefficient of a sub-critical ZPR. The basic theory of the Feynman-α, corresponds between the variance to mean ratio of the neutron count distribution (as a function of the count gate T) and an exponentially decaying function determined by the decay coefficient α. However, the true (theoretic) value of both the mean and variance are not tractable, and implementation of the Feynman-α method requires sampling of the mean and variance of the neutron count distribution in detection gates of duration T, for a range of values of T. Sampling the variance and mean is clearly vulnerable to statistical uncertainty, but also to small perturbations in the detection signal, due to imperfections in the recording system. Such imperfections might cause small perturbation that are very hard to detect, but still large enough to affect the outcome of the experiment. The outline of the present study is to propose an alternative method for sampling the variance to mean ratio, based on the sampled absolute deviation from the sampled mean, rather the the sampled quadratic deviation. The theoretical correspondence between the Mean Average Deviation (MAD) and the variance of the count distribution was previously introduced, and the study focuses on the advantages of the MAD, mainly it's robustness and moderate response to momentary failures of the detection system.