replication-crisis-demos

Some demos of statistical problems related to the replication crisis in psychology (and elsewhere)

View on GitHub

Some demos related to the psychology replication crisis

Author

Mark Andrews. Twitter: @xmjandrews

Demos

  1. Why most published research findings are false.
  2. How p-hacking can make almost anything significant.
  3. How optional stopping can inflate Type I error rates.
  4. How lower powered studies can inflate the value of true effects.

Source code

The demos above are all written in R and Shiny. The source code can be found at https://github.com/lawsofthought/replication-crisis-demos.

Note: To use shinyapps.io, I had to install package PKI and this lead to errors when I tried to install the usual way. This was fixed with

install.packages('PKI',,'http://www.rforge.net/')