Data Mining: Diamonds or Deceit?

There are a number of incentives and societal pressures to communicate study results with statistically significant treatment differences. However, overstatement of scientific findings, even in the absence of factual errors or data falsification, can have negative effects for both the communicator and his or her target audience. This case focuses on a fictional abstract in which three authors have differing ideas on how to communicate results from study investigating health benefits of unpasteurized milk. Questions presented in this case include:

  • When is it appropriate to frame a study’s title and abstract around secondary, post-hoc findings?
  • What are appropriate ways of responding to the problem that a study's funding source may appear to bias its results?
  • When is it appropriate to communicate different conclusions from the same data set?
  • Who is the audience for study abstracts?

Document Links