Back to the main page.
Bug 412 - correcttail option in montecarlo has incorrect default
| Status | CLOSED FIXED |
| Reported | 2011-01-18 15:16:00 +0100 |
| Modified | 2011-07-13 14:46:36 +0200 |
| Product: | FieldTrip |
| Component: | core |
| Version: | unspecified |
| Hardware: | PC |
| Operating System: | Mac OS |
| Importance: | P1 normal |
| Assigned to: | Saskia Haegens |
| URL: | |
| Tags: | |
| Depends on: | |
| Blocks: | |
| See also: |
Robert Oostenveld - 2011-01-18 15:16:39 +0100
at the moment the correcttail option has the default 'no', which means that the prob and mask are incorrect if the user specifies a alpha of 0.05. This was just found out based on the eventrelatedstatistics tutorial. I added the option cfg.correcttail = 'prob' to the eventrelatedstatistics totorial.
Saskia Haegens - 2011-01-19 14:51:18 +0100
actually, cfg.correcttail = no & cfg.tail = 0 & cfg.alpha = 0.05 (all defaults) leads to a warning
Roemer van der Meij - 2011-01-20 10:15:37 +0100
Didn't we decide to do a year a go to do it like this? Provide a warning for those who were unaware of the issue of the different conventions and corrections, but to let the defaults be as close to how they were for a very long time? As p-values can be a sensitive issue. I thought this was the reason why by default the p-values or the alpha are not corrected, as they never were so in the past.
Jan-Mathijs Schoffelen - 2011-03-01 10:44:25 +0100
perhaps we should discuss this in the FT-meeting
Saskia Haegens - 2011-06-23 18:14:34 +0200
additional documentation is now available here: http://fieldtrip.fcdonders.nl/faq/why_should_i_use_the_cfg.correcttail_option_when_using_statistics_montecarlo i will leave the default as is (cfg.correcttail = 'no', warning when using this in combination with tail = 0 and alpha = 0.05)