If the intensity discrepancy is indeed just a fluke, then the directional information from the same data would be expected to show the correct generic behavior.