statistic analysis test

[Emilio Manuel Serrano]

First, thank you for sharing this great online tool with us.

 

I have been using this online tool since I saw your publication. I have not=

iced that the test we use to calculate the p-value has changed. Before, he =

used the Mann-Whitney test and now, he uses the permutation test to evaluat=

e the p-value. Why does the analysis test change? Should I re-analyze my da=

ta with this new test?

 

Thanks for your help,

 

Best regards

 

Emilio M

PhD student

[Adam Claridge-Chang]

Hi Emilio,

Glad to hear you're enjoying the app.

We originally used Mann-Whitney as a conservative non-parametric p value, but have since implemented permutation p values instead. We thought that they would be more suitable, since they are a close cousin to bootstrap distributions of the mean difference. If you search this group, I believe there is earlier discussion about this.

However, we don't think anyone should be using significance testing or p values, and include them here only because we know misinformed supervisors, reviewers and editors will ask for them anyway. We recommend reporting and discussing ∆ + CI, and either leaving p values out entirely, or including them without discussion.

best,

Adam