跳过导航
Why Most Published Research Findings Are False
Article

Why Most Published Research Findings Are False

Akademien der Wissenschaften Schweiz, 2005 更多详情


Editorial Rating

8

Qualities

  • Innovative

Recommendation

“It can be proven that most claimed research findings are false.” Those are fighting words, but if anyone has the right to say them, John Ioannidis does. He’s co-director of the Meta-Research Innovation Center at Stanford and one of the researchers other scientists cite most often. This is one of his best-known papers. getAbstract recommends his insights to researchers, research readers and anyone who would like to be more scientifically literate. Readers interested in Ioannidis’s framework for assessing the proportion of false positive findings should refer to the original paper.

Summary

When various study types are subjected to simulations that assess power, ratio of true to untrue relationships, and bias very few have a positive predictive value of more than 50%. Indicators of substandard scientific studies include:

  • Small sample sizes – When everything else is equal, studies with a large sample size are more likely to garner true results. Large samples mean more power.
  • Small effect sizes – A scientific field with a small effect size (for example, a certain nutrient’s minimal effect on a complicated disease) ...

About the Author

John P. A. Ioannidis is a professor at Stanford University. He directs the Stanford Prevention Research Center and co-directs the Meta-Research Innovation Center (METRICS) at Stanford.


Comment on this summary or 开始讨论

More on this topic

Related Channels