Back to the main page.
Bug 1779 - ERF tutorial fails to find significant effect said to be in data
| Status | CLOSED INVALID |
| Reported | 2012-10-12 16:26:00 +0200 |
| Modified | 2012-10-12 17:50:39 +0200 |
| Product: | FieldTrip |
| Component: | documentation |
| Version: | unspecified |
| Hardware: | PC |
| Operating System: | Windows |
| Importance: | P3 normal |
| Assigned to: | |
| URL: | |
| Tags: | |
| Depends on: | |
| Blocks: | |
| See also: |
Roemer van der Meij - 2012-10-12 16:26:28 +0200
In: Parametric and non-parametric statistics on event-related fields Last section of tutorial: http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics The call to ft_clusterplot (the last function call in the tutorial), says it cannot find any cluster plot-able at the current alpha of 0.05. A few lines above, it is stated in the text: "Now a channel-time cluster is found from 0.33 s untill 0.52 s in which p < 0.05."
Lilla Magyari - 2012-10-12 17:46:30 +0200
hi Roemer, this issue came up this week when the test_tutorial_eventrelatedstatistics failed exactly because of the same reason. If I remember well, Jorn fixed it. The problem was that the p value is close to 0.05, so it sometimes significant, sometimes not. Depends on the randomization. So, if you run the tutorial script again (or a few times), it will be significant again ;). If it won't, then there is another problem. Lilla