The heavy academic workload and its competitive funding system often leaves little time for scientists to focus on the truly important – publishing truly clear and reproducible science, how to improve scientific systems for all, how to improve the university/ academic/ lab working environment.
In the spirit of trying to focus on the important (despite terrible, ugly workloads right now), here’s a few thoughts.
- Brush up on statistics and data presentation early: in the age of big data, statistics can seem hairy and scary. Knowing what type of statistical testing should be used (i.e. Student’s t-test vs 1 way ANOVA) is critical foundation. Even big data statistics is a bit more digestible when you break it down to the fundamental questions:
- What are the authors trying to look for?
- What are they comparing between? Is it relative or directly quantitative?
- Is the probability threshold for a potentially real finding reasonable, especially compared to the total number of random findings that their technique has uncovered.
- Avoid establishing a system where a PhD is only conferred after the student has published one first author paper: Yes, the aim of a PhD is to make a new scientific contribution to your field. And yes, students should theoretically publish at least one first author paper (aka the formal demonstration of having contributed to a field) in order to merit a doctorate degree. But there will always be unfortunate circumstances where this cannot happen. In desperate circumstances, if the production of positive data is the only gateway to graduation and earning a real salary…
- Cleverly designed and rigorous experiments with negative data are just as important as those producing positive data. Imagine the wealth of negative data that could have been highly published and how important it would have been for industry scientists when finding truly effective drug treatments against diseases.
- Don’t chase after and only value high impact factor papers. Valuing only high impact factor papers, and the companion desire to only publish in them, forces you to find/package the most novel positive data, complete with a 100% definitive mechanism ever. Sometimes this is just not 100% clear or possible ever. And yet! Read a combination of big and small papers, and cite the small papers that make an important finding for your field.
- Some types of studies inherently produce more variability than others. Okay, I firmly believe in repeating experiments until you know the breadth of variation in your data points (i.e. giving you the greatest confidence of knowing when an observation is truly different versus just different in appearance by chance). Know when you should repeat an uncertain observation, and when you should be moving on. Don’t automatically try to make a P <0.05 P-value by manipulating data points around.
- You can publish and contribute your best to science. And peacefully and happily move on, if needed. A recent senior scientist said to me, ‘[Doing] research can kill you.’ As gatekeepers of the most probable and closer approximation to truth, ask yourself: what is the value of staying if your job system prevents you from contributing what you think you ought to be contributing?