In our previous works, we reported that compensation of the geomagnetic field to a level less than 0.4 μT ("zero magnetic field," or ZMF) affected human cognitive processes. ZMF exposure increased the number of errors and the task processing time by 2.4% in average. However, in the array of the magnetic effects calculated from the experimental data, some readings have been found to deviate from the mean magnetic effect by more than three standard deviations. This finding could give rise to doubt as to whether the magnetic effect observed was a mere sequence of the presence of such unlikely data values. In the present work we examine the results of the unlikely data elimination and show that the corrected magnetic effect in tested humans remains statistically significant, though at a reduced magnitude 1.5%.