[FieldTrip] Nonparametric Satatics Based on "Blair and Karniski 1993" Using FieldTrip

David Groppe david.m.groppe at gmail.com
Fri Sep 7 15:37:57 CEST 2018


What you would need to do Kamia is to set the neighbors matrix to the
identity matrix:

neighbours        = ft_prepare_neighbours(cfg_neighb, dataFC_LP);

That function is described here:
http://www.fieldtriptoolbox.org/reference/ft_prepare_neighbours


And I was wrong with what I initially told you. This will still cluster
variables across time within single electrodes, so it is not equivalent to
the Blair & Karniski test. I don't know if there is a way to prevent any
clustering at all in FieldTrip (perhaps set the neighbour matrix to all
zeros?).

I have some MATLAB code that implements the Blair & Karniski method:

https://www.mathworks.com/matlabcentral/fileexchange/54585-mult_comp_perm_t2-data1-data2-n_perm-tail-alpha_level-mu-t_stat-reports-seed_state?s_tid=prof_contriblnk

https://www.mathworks.com/matlabcentral/fileexchange/29782-mult_comp_perm_t1-data-n_perm-tail-alpha_level-mu-reports-seed_state?s_tid=prof_contriblnk

That might be easier to use. It is part of a toolbox for ERP analysis that
may be useful if you're working with ERPs:
https://openwetware.org/wiki/Mass_Univariate_ERP_Toolbox

cheers,
   -David



On Thu, Sep 6, 2018 at 3:35 PM Kamia Kavi <kamiakavi at gmail.com> wrote:

> Hey David,
>
>
>
> Thank you very much for your tip as well as your paper. I printed that and
> will read it tomorrow.
>
>
>
> Two follow-up questions:
>
> 1) So, to not cluster the timepoints/channels, is it enough the change the
> ‘cfg.correcm’ parameter from ‘cluster’ to ‘no’ and leave the other
> parameters unchanged?
>
>
>
> 2) if cfg.correctm = ‘no’ (and, of course, using cfg.method =
> 'montecarlo'), are significant values (i.e., ones rather than zeros) in
> stat.mask logical matrix already corrected for multiple comparisons? In
> other words, can these ‘ones’ in the matrix be considered as statically
> significant channel/timepoints between two conditions at a certain critical
> value?
>
>
>
> Best wishes,
>
>
>
> Kamia
>
> On Thu, Sep 6, 2018 at 7:35 PM David Groppe <david.m.groppe at gmail.com>
> wrote:
>
>> Hi Kamia,
>>    The cluster-based permutation test implemented by FieldTrip is similar
>> to the Blair & Karniski method:
>>
>> http://www.fieldtriptoolbox.org/tutorial/cluster_permutation_timelock
>>
>> If you set the parameters of the test such that no timepoint/electrode
>> has any neighbours (i.e., you can't form any clusters), than it is
>> equivalent to the Blair & Karniski method.
>>
>> For an explanation of how the cluster-based test differs from Blair &
>> Karniski, I cover that in this review/tutorial paper:
>>
>> https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4060794/
>>
>> cheers,
>>    -David
>>
>>
>>
>> On Thu, Sep 6, 2018 at 10:36 AM Kamia Kavi <kamiakavi at gmail.com> wrote:
>>
>>> Dear FieldTrip-ers,
>>>
>>> Is there an option in FirldTrip to perform a non-parametric statistics
>>> as described in Blair and Karniski 1993?
>>>
>>> Best wishes,
>>>
>>> Kamia
>>> _______________________________________________
>>> fieldtrip mailing list
>>> https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>>> https://doi.org/10.1371/journal.pcbi.1002202
>>>
>> _______________________________________________
>> fieldtrip mailing list
>> https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>> https://doi.org/10.1371/journal.pcbi.1002202
>>
> _______________________________________________
> fieldtrip mailing list
> https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> https://doi.org/10.1371/journal.pcbi.1002202
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20180907/5b515bc7/attachment.html>


More information about the fieldtrip mailing list