We often prefer to have consistency in our data, but data without variation in crucial measures doesn't tell us anything. For example, if everyone in your office gets the same performance score (4 out of 5), the score will have no value in terms of how it explains actual work performance. In order to understand how one thing affects something else, you must have differences in the "one thing". Meaningful differences in the main thing you are measuring are essential to understanding how it relates to the outcomes that interest you.