There is tremendous risk in repeatedly analyzing data as it comes in…

For example, in the absence of any effect, the p-values of repeated t-tests on a growing dataset will randomly walk between 1 and 0!

Thanks to @lakens for inspiring this simulation!