At least one study was done that concluded the real effect wasn't from the change, it was the expectation that people were being measured. You could change in any direction and get the same results - repeated across different companies which in turn mean some left what another went to - and both got similar gains.