Are you in the process of reviewing a paper? do you need to evaluate the stats or experimental design in a manuscript but could do with some additional guidance? You need look no further, as eLife Ambassador Savio Chan has pulled together a list of articles (below) that can help you to identify and address common problems with published articles in the basic biomedical sciences.

  1. Why you shouldn’t use bar graphs to show continuous data (and what to do instead)
  1. The problem with underpowered studies (Low power is a problem even when you find a significant difference)
  1. Why it’s important to report all excluded observations & the reasons for exclusion
  1. P-values are often reported incorrectly: Why it’s important to present the information needed to verify the test result
  1. Unblinded studies find larger effects
  1. Animal research: Follow the ARRIVE guidelines to improve transparency and reproducibility
  1. Animal research: Multi-lab studies may improve reproducibility
  1. Check for clusters of non-independent data (replicates, mice from the same litter, etc.): Did the authors account for non-independence in their analysis?
  1. Beware of image manipulation

 

Educational Articles for Peer Reviewers

  1. Common misconceptions about data analysis and statistics
  1. Small samples are more likely to give spurious results
    • Wainer, H. The Most Dangerous Equation: Ignorance of how sample size affects statistical variation has created havoc for nearly a millennium. American Scientist 95, 249-256 (2007). http://www.jstor.org/stable/27858964
  1. Transparency is the key to quality
  1. Research methods: Know when your numbers are significant