Pairwise comparison is a common technique in multi-criteria decision analysis (MCDA), with simplified methods like the RANking COMparison (RANCOM) aiming to reduce cognitive load by using only directional or equality judgments. This study tested the perceptual validity of such comparisons through an experiment with 81 participants evaluating which of two grids had more black squares, or if they were equal, at varying difficulty levels. Accuracy exceeded 97% for differences above 10% but dropped to 54.7% for differences below 5%. A logistic psychometric model confirmed this pattern and allowed estimation of Just Noticeable Differences (JND) at various confidence levels. The findings support simplified pairwise methods under moderate and large differences, while highlighting reduced reliability in nearly equal cases.