Using spectopo() shows a significant difference in spectral power among 3 groups by analyzing different bands, but there is no significant difference in ersp of three groups. Please advise.
You can think of the ersps as spectral powers in time, while spectral power would be the average (or sum) of all these times. Meaning that while the overall SP might differ, you are making many more comparisons when analyzing ersps and breaking down the data into smaller comparable chunks, where distinct patterns do not emerge as easily as when more data is averaged together.
In short, a difference detected with the use of SP does not neccesitate there is a difference in ersps as well.
Thank you Jakob Sajovic . Actually I am working on Mindfulness-based stress reduction (MBSR) data. I found that using spectopo() for PSD spectral power in delta, theta and alpha is increasing after MBSR. But during time frequency analysis using newtimef() of eeglab ERSP power in delta band is decreasing. so there is difference in the results of PSD and time frequency analysis.
Is the difference significant? If not you cannot claim there is an actual difference. Moreover outliers can skew averages a lot, which can make your graph seem like a power is smaller, when in fact this is an effect of an outlier. For ERSP you also need to designate events. If you have an event someone has to react to (even if this is listening to guided meditatio, i.e.), you would expect at least the delta power to drop, but theta likely as well. Without a more detailed explanation of an experiment it is difficult to explain and explore any further though.
- Badges
- Science topic