Mark Andrews. Twitter: @xmjandrews
- Why most published research findings are false.
- How p-hacking can make almost anything significant.
- How optional stopping can inflate Type I error rates.
- How lower powered studies can inflate the value of true effects.
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/')