An epic journey through statistics and machine learning

StatQuest: One vs Two-Tailed P-values

Here’s the deal!

2 thoughts on “StatQuest: One vs Two-Tailed P-values”

This is a great lesson. I have done some R functions for this, stacked histograms etc to simulate this 2 stage process and show where the p-values are coming from (left-tailed or 2-sided test). This type of problem pops up when there is a 2 stage procedure on the same data where the outcome of the first test determines the second procedure. e.g. performing post-hoc confidence intervals after rejecting the null in a preliminary test, so take caution as confidence levels may be a bit lower than you think, especially for small sample sizes. You might look at this in a future video.

This is a great lesson. I have done some R functions for this, stacked histograms etc to simulate this 2 stage process and show where the p-values are coming from (left-tailed or 2-sided test). This type of problem pops up when there is a 2 stage procedure on the same data where the outcome of the first test determines the second procedure. e.g. performing post-hoc confidence intervals after rejecting the null in a preliminary test, so take caution as confidence levels may be a bit lower than you think, especially for small sample sizes. You might look at this in a future video.

That’s a great idea! Thanks for sharing it with me. ðŸ™‚