The key to "good data work" is, to the best of your ability, avoid imposing your bias on the process. I propose that this isn't possible given that we need to make choices about the data at almost every step.
This is especially true with analysis! We have a groovy set of data - but why do we have it? What are we supposed to do? We know that scientists are interested in the possibility of life under the ice of Enceladus, but what data supports this idea or invalidates it?
To do our jobs, we need to understand the data better and hold tightly to the mantra that the story is in the data. It's not in our heads or in anyone's opinion - we need to let the data speak to us.