Working without a net

Prompted by a few comments on the last post--

Of the many difficult things you learn in graduate school, one of the scariest is that you, and you alone, are truly responsible for what you publish. The logic, accuracy, and weight of what you write are yours to furnish--or not.

I do not mean to say that your PI is not also responsible for your work. His/her name is on the paper too, and if it tanks, the blame will if anything fall more strongly on your advisor, as the senior author.

But as first author you are the one with direct interaction with the actual data. You collect it. You make little tweaks to your protocols to improve your data collection. Are any of those important? And did you do what you thought you did--or did you write one thing in your lab book and pipette another into your eppendorf?

You borrow a colleague's reagent, or code. Does it work as promised? And what makes you sure?

As you're collecting data, you make some snap judgments about experimental success or lack thereof. Is the problem that you put in the wrong concentration of a reagent? Or is the data real, but anomalous, unexpected? Should you trust it?

You analyze the data, and you are the first one to make decisions about what data can and should be excluded; your PI will likely weigh in, but you are on the front line. Are you excluding that data point for legitimate reasons, or because it doesn't fit your hypothesis?

You put the data into graphs. Do those graphs show your data as honestly as possible, or are they covering up something that makes you uncomfortable? When you show your data at lab meeting and someone questions your results, who is right? Is their objection valid, or do you have a better counter-argument?

Some of these questions will be resolved by talking it over with your PI, your lab colleagues, your thesis committee, or even by your reviewers. But in your development as a scientist, at some point you cannot rely on others to catch your mistakes. You will get conflicting advice from the people you talk with. You have to weigh up this advice and eventually reach your own conclusions about the best answer.

This is just plain scary. The fear does not arise out of nowhere. You do actually have to take responsibility for results that you may have gotten wrong. It would be strange if that fear did not haunt you from time to time. (I think impostor syndrome, at least in part, arises when the fear becomes continual and uncontrolled.)

When I was an early-stage graduate student, fretting over these concerns, I asked a visiting speaker during a student lunch, "How do you gain confidence in your own data?" Her advice was along the lines of, "Repetition; experience; and talking to all the smart people you can." She was right, but the best part of the interchange was hearing a highly respected scientist acknowledge that she, too, had faced that struggle--and in some ways still did, as she grappled with understanding and shaping her trainees' experiments.

Sending out my first manuscript was transformative in this respect. When we got page proofs from the publisher, it was suddenly clear to me that the assertions and arguments that I had written (and then had heavily re-written by my PI) might be read by someone, some day. Perhaps other people realize this rather obvious fact earlier in the game, but it took the formatted page proofs to stun me into it. My advisor gave lots of feedback and shaped the paper in diverse ways, and the reviewers nudged it as well, but it was now clear to me that in essence, I was working without a net.

I don't think I've been the same since.