Getting Honest About Data
I've been happily plodding along, hopeful and unaware, towards what I thought would be my first published paper. It's nothing big, nothing earth-shattering, but it's mine (and a few other people's), and I've been working at it for so goddamn long (relative to the rest of my life). But it recently occurred to me, halfway through the process of responding to reviewer comments, that my data may be have been misrepresented by the statistical analyses I used. Misrepresented. I'm going to let you sit with that for a minute because it took me over an hour to move past this fact. My interpretations and conclusions in my already-peer-reviewed paper might be bullshit, and this alters the message of the paper I've already submitted.
I sat in my office nearly the entire day, hardly moving, rapidly typing, deleting and retyping R code to try and figure out where things went wrong. By the end of the day I was exhausted, hella cranky, and still wanting for answers. More than anything, I was worried about what the senior scientist with whom I'm collaborating would think. We were 3 days away from our resubmission deadline, my conclusions had suddenly seemed as valid as a snowball on the Senate floor, and I was unable to decide whether to press on or retract the questionable data.
Though the other two problems still exist, the last one is no longer an issue. It took me a few minutes of hand-wringing and light whining to realize that I can't publish something that I doubt. I'm disappointed with the current situation, but I'm coming away with a lesson in science ethics. You just don't put something out unless you're data say it's true. As for my paper, she will eventually see the light of day, but I need to do a bit more tinkering before her debut. I'm trying to think of it as progress, not failure.
To document interesting ideas about science and nature and reflect on the experience of being a scientist from the margins.