Reminds me of A/B testing on pupils. The result may teach you something about methodologies, but pupils also learn to do tests on humans.<p>If my work is measured by an arbitrary metric like LOC, I write extra lines or keep them to a low level so that I can easily improve it later.<p>But the hardest part is probably identifying the behavior as a first step. For example the ambition to change behavior on a grand scale is probably undesirable. I think in most cases there is probably an underlying condition to this.
"Sometimes the solution to a behavior problem is simply to revisit incentives and make sure they align with the desired goal." -- Use this post by Shane Parrish (his podcast is one of my favorites) to think about the type of incentives are being used in your organization, and which reinforcement schedules are applied. One must-have element in a working place is to add a clear and explicit definition (in writing!) of roles and responsibilities. People shouldn't try to guess what is considered a desired behavior.