Exactly. You don't have clue which variable is responsible for the outcome you get. Change should be done one thing at time. I can understand fixing multiple bugs at once but I agree with your reasoning regarding adjustment of multiple variables. You just turn your experimental data into garbage by doing that.