It’s the nightmare scenario: you look back at an old bit of code and realize you’ve made a mistake and, to make matters worse, the paper has already been published. This year I lived that nightmare scenario. I had shared my code only to discover that a variable that should have been reverse scored (which boils down to multiplying the number by -1), wasn’t. It was a minor oversight that I’d made as a 1st year PhD student learning new statistics, I hadn’t caught the mistake until now, and, worse still, the code had been used in two papers I wrote simultaneously. I considered changing my name and hiding but as I had a postdoc and my mother claims to like me, I figured it was better to keep my current identity.
‘…the right decisions don’t come without risk….’
Reaching out to the senior author we knew there was only one solution: We had to redo the statistics and submit corrections. As an early career researcher, I was panicked. What if the results were drastically different, was a retraction (possibly two) in my future? Fear aside, a mistake was made, we had to own it, and if we were going to believe in scientific integrity then we had to show ours. It’s been my experience that the most difficult decisions, the ones that I’m truly afraid to make – those are the decisions I know to be right. But the right decisions don’t come without risk and I can’t pretend that I wasn’t, and continue to be, worried that not everyone would see this as a minor mistake. Science is competitive and the feeling of having to be flawless, particularly at this phase of my career, is a weight. As a woman in science I already have to fight to be taken seriously, to be seen as competent, and I had committed a sin, I had made an honest mistake that had been published, twice. Before I could find out the results of my mistake on my career, I had to find out their impact on my papers.
‘As a woman in science I already have to fight to be taken seriously, to be seen as competent…’
I somehow survived three painful hours while I waited to finish work at my postdoc and could get back to where I kept the study data. Upon sitting at my desk (liquid courage in hand) I redid the stats, anxious to find the results. Now look, I’m no slouch with numbers, I know what multiplying by -1 does to them, but panic overrode sense in that moment and I needed to see to believe. First paper: Flipped the direction of effect on a non-significant variable that remained that way. Okay, fairly minor, just requires that the journal update the tables. Second paper: Again, the only thing that changed was the direction of effect, though this variable had been and still was significant, means we had to adjust the numbers, a line in the abstract, and three sentences in the results. Not great, but as variables go it hadn’t even rated being mentioned in the discussion.
Okay, okay, okay (deep breaths, bit more whisky), this could be so much worse I told myself. I screwed up but hey, everyone makes mistakes, I was learning something new, I should’ve have caught it earlier, but it was caught now. Onto the next step, making the corrections, contacting coauthors, and letting the journals know. Time to really live by our ideals. But first! Another moment of panic while I wondered if I had made the same mistake in my two newest papers. Opening code, reading through, and…no, I hadn’t made the mistake again. Somewhere along the way I had clearly learned how to do these statistics correctly, I just hadn’t caught it while I was working on these two papers and had copy-pasted the code across them. Good news, I am in fact capable of doing things correctly.
‘I had lived my nightmare and it felt, as least in this moment…completely survivable…’
Writing the email to my coauthors wasn’t something that I was particularly looking forward to. “Oh hey fellow researchers that I respect and admire, I screwed up and am going to let the journals and the world know. PS, please don’t think less of me and hate me. Okay, thanks.” While that’s not what I wrote, that’s what it felt like. An admission of imperfection, shame, guilt, a desire to live under a rock. However, I’ve been blessed with caring and understanding collaborators, each of whom was extremely supportive. Next, I sent an email to the journals explaining the mistake and requesting corrections be published. Each journal was understanding and helped us write and publish corrections and that was it, it was done. I had lived my nightmare and it felt, as least in this moment…completely survivable. I had imagined anxiety and panic and battling my own shame and guilt. This…this was a feeling of stillness that I was not prepared for.
Prior to contacting the journals and writing this blog, I asked myself how much this would hurt my career. Would a small mistake cost me my reputation, respect, and future in the science I’d already sacrificed so much for? Would writing this blog and openly speaking to the fact that I had made a mistake only further the potential damage to career and respect? Would a single mistake, done at the beginning of my PhD and not since repeated, mean that others didn’t trust my science and statistics, not want to work with me? Would I trust my own skills, and more importantly, myself, again? There was so much uncertainty and so little information available on this experience, yet mistakes like this must happen more than we think, they just go unspoken.
‘…genuine mistakes? We have to make those acceptable to acknowledge, correct, even retract, and speak about, to learn and move on from.’
This, this is the crux of a problem in science, there are unknown consequences of acknowledging and speaking openly about our mistakes and, by failing to do so, we only further increase the chance that mistakes go uncorrected. Let’s hold those that perform purposeful scientific misconduct accountable, but genuine mistakes? We have to make those acceptable to acknowledge, correct, even retract, and speak about, to learn and move on from them. Those who don’t learn from their mistakes? Well, they may be doomed to face the consequences. As a note, if we’re going to move towards openness and transparency in science then we need to be particularly careful that those in underrepresented groups aren’t unfairly punished or scrutinized for admitting and speaking about mistakes as these groups are already under a microscope and face unique and frustrating challenges. We cannot allow openness and transparency to be used as one more excuse for someone to tell us no, not if science is to diversify and progress.
‘What kind of person and scientist do I want to be?’
Of all the questions I asked myself, deciding to write this post came down to one: What kind of person and scientist do I want to be? As an animal welfare scientist, I have long believed in being transparent and open in science, I realized that’s who I am as a person as well. Living by my ideals meant not only correcting my mistake but also talking openly and frankly about it. These choices, challenging as they may have been, are the right ones. To err is human and luckily for me I have divine friends, mentors, and colleagues that forgive me my mistakes and sins. I believe that we should all be so lucky and that mistakes should be openly and transparently discussed. For now, I live to science another day and look forward to the challenges, mistakes (which I intend to catch prior to publication), and learning that come with it.
For those interested in working with me (imperfections and all) when my current postdoc ends this January, feel free to get in touch via ResearchGate (https://www.researchgate.net/profile/Lauren_Robinson7) or Twitter (https://twitter.com/Laurenmrobin).
Links to published corrections: