@PsyPAG recently asked, via twitter “What advice would you give to yourself at the beginning of your postgraduate studies?” I thought that this question deserved way more air time than what is usually given to the pit of despair that twitter often descends into. It certainly got me thinking about the things I wish I knew at the start of my DPhil. My personal list is pretty long, but, I do have two-step piece of advice I’d give to anyone starting their postgrad studies.
1) Invest in open and reproducible science, and do it now.
The PsyPAG survival guide (p. 85) has a pretty stellar introduction to open science that will get you going (Crüwell et al. 2019 is also a nice place to start, as is the Improving your statistical inferences MOOC). These actions include; preregistering your analysis, helping to avoid many questionable research practices; share data, as an additional product of your research; publish open access, so that everyone can read your work; make your analyses reproducible, so that others can check your working and learn from your code (a bit like secondary school maths). The list goes on. But, any efforts to engage in these actions helps improve the quality and value of your work. The final statement made a strong impression on me “Remember, a study with one of these [open science] actions is better than a study with none”.
Now that I can more or less assume you know a little about open and reproducible science we can get to the crux of my advice.
2) Ignore your fears that open and reproducible research practices might harm your career prospects.
Research is hard. Starting and continuing a career in research is hard. It makes sense that postgrads are terrified at doing (or not doing) something that might harm their career prospects. We compare ourselves to that first year PhD that somehow has 10 publications already and worry that we aren’t moving fast enough to be competitive (note to self; this is not an impostor syndrome post). When it comes to enacting these new ‘open science’ practices, the worry is that we might be ‘shooting ourselves in the foot’ essentially by taking more time to make our research process better and more rigorous.
I think that this worry is misplaced. I also think that this worry has dangerous consequences; both to the quality of research overall, as well as our own wellbeing. We need to shift towards considering the time spent on openness and reproducibility as an investment in research best practices, not as time lost on something ‘extra’. Think of it as doing the best work possible, rather than maybe, possibly, unknowably, being less competitive in the ‘game’ of science.
But, I’m worried that I will publish less if I do [Insert open/reproducible research practice of choice]
This captures the broad worry of most PhD students interested in pursuing a research career. We want that post-doc, that tenure-track position, and we need papers to get that post (a quick aside, notice that “I need papers” already steps outside of the scope of “I want to do good research”). Publishing is not guaranteed, except perhaps the accepted in principle following stage 1 review of a registered report. So, a sub-piece of advice: don’t think of each study you want to run as a guaranteed paper – I made this mistake, and have backlogs of work that I may never have time to publish, if it’s of sound enough quality to publish in the first place. Running four studies instead of two does not mean that you will have more publications. Chances are it means that you’ll just have more unpublished studies at the end of your PhD.
What you do have control of is the quality of work you produce. You can choose to invest time in learning to program and then sharing your analysis scripts and data alongside a pre-print of your paper. Now, entirely outside all the fun of trying to publish in prestigious journal X, you have given more research output than most published papers. I don’t see this as publishing less; I see this as publishing more. The research paper is just an advertisement for the research you’ve done (apologies to whoever said this first), but the; protocol, data, and code are the actual research you have completed.
Change is coming, fast. Openness and reproducibility are not only valuable to the quality of research out there, they are marketable skillsets. Tech-unsavvy senior PIs will have to get involved in openness and reproducibility soon – funders will be mandating it. It’s also becoming more evident that many PIs don’t actually view lots of publications as a good thing; as it should be – quality over quantity. The PhD is support to be a time of skill acquisition, not publish or perish. You aren’t expected to have a CV the length of a small book. You are supposed to have skills that will be valuable to the research process, and to your future employer. Openness and reproducibility are awesome skills to showcase.
I’ve written before about how unpleasant it is to look back on your PhD work having learned about open and reproducible research near the end of that process. I know this is speaking with hindsight. I know that I’m being preachy. I’m also aware that learning from others experiences is valuable. Please learn from mine; invest in openness and reproducibility asap and consider how you will benefit from these practices, rather than worry about possibly not running an extra study.
If you are at the beginning of your PhD, you have an entire research career ahead of you. You can adopt best practices from the very beginning. You’ll thank yourself for it.
Dr Sam Parsons is a postdoctoral research associate at the University of Oxford.