[Mne_analysis] Comparison of brain responses under three different conditions?

Denis-Alexander Engemann denis.engemann at gmail.com
Sun Dec 21 11:30:01 EST 2014
Search archives:

2014-12-21 17:26 GMT+01:00 Maria Hakonen <maria.hakonen at gmail.com>:

> Hi Denis,
>
> I have used Newman-Keuls test in Statistica, but maybe paired t-test would
> work as well if I correct the alpha.
> Did you mean that I could simply use stats.ttest_rel and then correct the
> values with, for example, fdr_correction?
>
>
Yes, or just use our 1samp ttest after computing a paired contrast


> -Maria
>
> 2014-12-21 17:49 GMT+02:00 Denis-Alexander Engemann <
> denis.engemann at gmail.com>:
>
>> Hi Maria,
>>
>> 2014-12-21 16:35 GMT+01:00 Maria Hakonen <maria.hakonen at gmail.com>:
>>
>>> Hi Denis,
>>>
>>> I tried that and got  the following p values: 0.02993846,  0.47414992
>>> The corresponding Greenhouse-Geisser corrected values computed with
>>> Statistica were: 0.026775, 0.462423.
>>> Huynh-Feldt correction gives 0.023894, 0.473539 and lower-bound
>>> correction 0.036458, 0.427778
>>>
>>> The uncorrected values calculated with Statistica and Python are equal
>>> but also the corrected values are almost equal. So, maybe f_twoway_rm
>>> works correctly now?
>>>
>>>
>> There might be some minor implementation-related difference, would be
>> good to investigate.
>> Generally the ANOVA function is tested against R and SPSS, event for the
>> correction:
>>
>>
>> https://github.com/mne-tools/mne-python/blob/master/mne/stats/tests/test_parametric.py#L10
>>
>> I would therefore trust the results.
>>
>>
>>> It would also be interesting to do post-hoc comparisons between
>>> conditions but I think that Python doesn't have function for post-hoc test,
>>> or at least I didn't found it?
>>>
>>>
>> What do you mean? You simply compute paired t-tests and correct the alpha.
>>
>> -Denis
>>
>>
>>> -Maria
>>>
>>> 2014-12-21 16:49 GMT+02:00 Denis-Alexander Engemann <
>>> denis.engemann at gmail.com>:
>>>
>>>> Should be Greenhouse-Geisser,
>>>> also see source code:
>>>>
>>>>
>>>> https://github.com/mne-tools/mne-python/blob/master/mne/stats/parametric.py#L249
>>>>
>>>> 2014-12-21 15:47 GMT+01:00 Denis-Alexander Engemann <
>>>> denis.engemann at gmail.com>:
>>>>
>>>>> [f,p]=f_twoway_rm(data,factor_levels=[2,3], correction=True)
>>>>>
>>>>>
>>>>> 2014-12-21 15:45 GMT+01:00 Maria Hakonen <maria.hakonen at gmail.com>:
>>>>>
>>>>>> Hi Denis,
>>>>>>
>>>>>> Are you sure that the sphericity correction is implemented by
>>>>>> f_twoway_rm?
>>>>>> I tested it with data where the assumption of sphericity is violated.
>>>>>> I calculated p values with Statistica and corrected them using the
>>>>>> lower-bound estimate, Greenhouse-Geisser correction and the Huynh-Feldt
>>>>>> correction. However, the p-values given by f_twoway_rm were equal with the
>>>>>> uncorrected values I calculated with Statistica. I used f_twoway_rm as:
>>>>>>
>>>>>> [f,p]=f_twoway_rm(data,factor_levels=[2,3])
>>>>>>
>>>>>> where data is hemisphere*sesion ANOVA table.
>>>>>>
>>>>>> -Maria
>>>>>>
>>>>>> 2014-12-21 15:08 GMT+02:00 Denis-Alexander Engemann <
>>>>>> denis.engemann at gmail.com>:
>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> 2014-12-21 14:00 GMT+01:00 Maria Hakonen <maria.hakonen at gmail.com>:
>>>>>>>
>>>>>>>> Hi,
>>>>>>>>
>>>>>>>> Is it that when I have hemisphere*session ANOVAs, I can use
>>>>>>>> f_twoway_rm, and this gives the corrected values?
>>>>>>>>
>>>>>>>
>>>>>>> with correction=True
>>>>>>>
>>>>>>>
>>>>>>>> Should one-way ANOVA be used in the cases where I am not interested
>>>>>>>> in hemisperic asymmetry? Is there any function for one way ANOVA that gives
>>>>>>>> sphericity corrected values?
>>>>>>>>
>>>>>>>>
>>>>>>> A one-way ANOVA is currently not implemented. But you can just look
>>>>>>> at the relevant main effect. Should be equivalent.
>>>>>>>
>>>>>>>
>>>>>>>> -Maria
>>>>>>>>
>>>>>>>> 2014-12-21 14:21 GMT+02:00 Denis A. Engemann <
>>>>>>>> denis.engemann at gmail.com>:
>>>>>>>>
>>>>>>>>> Hi Maria,
>>>>>>>>>
>>>>>>>>> On 21 Dec 2014, at 12:51, Maria Hakonen <maria.hakonen at gmail.com>
>>>>>>>>> wrote:
>>>>>>>>>
>>>>>>>>> Hi Denis,
>>>>>>>>>
>>>>>>>>> Should I use two-way ANOVA if I also want to compare hemispheres,
>>>>>>>>> i.e. have a 2*3 (hemisphere*session) ANOVA table?
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> Sounds like the repeated measures anova is what you want to do
>>>>>>>>> here indeed. I just realized the sphericity correction is actually
>>>>>>>>> implemented by our code. So that should fly. Checkout the correction
>>>>>>>>> parameter.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> I would need to do group level analysis.
>>>>>>>>>
>>>>>>>>> -Maria
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> 2014-12-21 12:38 GMT+02:00 Denis-Alexander Engemann <
>>>>>>>>> denis.engemann at gmail.com>:
>>>>>>>>>
>>>>>>>>>> Hi Maria,
>>>>>>>>>>
>>>>>>>>>> 2014-12-21 10:18 GMT+01:00 Maria Hakonen <maria.hakonen at gmail.com
>>>>>>>>>> >:
>>>>>>>>>>
>>>>>>>>>>> Hi all,
>>>>>>>>>>>
>>>>>>>>>>> I would need to find the time intervals where there are
>>>>>>>>>>> statistically significant differences between MEG responses measured under
>>>>>>>>>>> three different conditions.  I study the responses by comparing the mean
>>>>>>>>>>> values in steps of 200 ms between the range of 400-2400 ms. I would like to
>>>>>>>>>>> do this for responses from several brain areas. What could be the best way
>>>>>>>>>>> to compare the mean values of the tree responses?
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>> Single subject or group level?
>>>>>>>>>> For single subject a standard F-test would be appropriate. For
>>>>>>>>>> groups a oneway repeated measures ANOVA, if I understand your design
>>>>>>>>>> correctly (simply 3 different conditions).
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>> I have planned to try f_twoway_rm in a loop. However, when
>>>>>>>>>>> rmANOVA is used, sphericity test is needed, and if the assumption of
>>>>>>>>>>> sphericity is violated, p values need to be corrected. Is there any
>>>>>>>>>>> function for this in Python (e.g. Mauchly's sphericity test and
>>>>>>>>>>> Greenhouse-Geisser test)?
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>> Note that this only needed if you have more than 2 factors with
>>>>>>>>>> more than two levels. Currently this is not implemented in our stats code.
>>>>>>>>>> The other questions is why you need a 2-way test for 3 conditions. I
>>>>>>>>>> probably did not fully understand your design.
>>>>>>>>>>
>>>>>>>>>> Denis
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>> Many thanks already in advance!
>>>>>>>>>>>
>>>>>>>>>>> Regards,
>>>>>>>>>>> Maria
>>>>>>>>>>>
>>>>>>>>>>> _______________________________________________
>>>>>>>>>>> Mne_analysis mailing list
>>>>>>>>>>> Mne_analysis at nmr.mgh.harvard.edu
>>>>>>>>>>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> The information in this e-mail is intended only for the person
>>>>>>>>>>> to whom it is
>>>>>>>>>>> addressed. If you believe this e-mail was sent to you in error
>>>>>>>>>>> and the e-mail
>>>>>>>>>>> contains patient information, please contact the Partners
>>>>>>>>>>> Compliance HelpLine at
>>>>>>>>>>> http://www.partners.org/complianceline . If the e-mail was sent
>>>>>>>>>>> to you in error
>>>>>>>>>>> but does not contain patient information, please contact the
>>>>>>>>>>> sender and properly
>>>>>>>>>>> dispose of the e-mail.
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> _______________________________________________
>>>>>>>>>> Mne_analysis mailing list
>>>>>>>>>> Mne_analysis at nmr.mgh.harvard.edu
>>>>>>>>>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> The information in this e-mail is intended only for the person to
>>>>>>>>>> whom it is
>>>>>>>>>> addressed. If you believe this e-mail was sent to you in error
>>>>>>>>>> and the e-mail
>>>>>>>>>> contains patient information, please contact the Partners
>>>>>>>>>> Compliance HelpLine at
>>>>>>>>>> http://www.partners.org/complianceline . If the e-mail was sent
>>>>>>>>>> to you in error
>>>>>>>>>> but does not contain patient information, please contact the
>>>>>>>>>> sender and properly
>>>>>>>>>> dispose of the e-mail.
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>> _______________________________________________
>>>>>>>>> Mne_analysis mailing list
>>>>>>>>> Mne_analysis at nmr.mgh.harvard.edu
>>>>>>>>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> The information in this e-mail is intended only for the person to
>>>>>>>>> whom it is
>>>>>>>>> addressed. If you believe this e-mail was sent to you in error and
>>>>>>>>> the e-mail
>>>>>>>>> contains patient information, please contact the Partners
>>>>>>>>> Compliance HelpLine at
>>>>>>>>> http://www.partners.org/complianceline . If the e-mail was sent
>>>>>>>>> to you in error
>>>>>>>>> but does not contain patient information, please contact the
>>>>>>>>> sender and properly
>>>>>>>>> dispose of the e-mail.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> _______________________________________________
>>>>>>>>> Mne_analysis mailing list
>>>>>>>>> Mne_analysis at nmr.mgh.harvard.edu
>>>>>>>>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> The information in this e-mail is intended only for the person to
>>>>>>>>> whom it is
>>>>>>>>> addressed. If you believe this e-mail was sent to you in error and
>>>>>>>>> the e-mail
>>>>>>>>> contains patient information, please contact the Partners
>>>>>>>>> Compliance HelpLine at
>>>>>>>>> http://www.partners.org/complianceline . If the e-mail was sent
>>>>>>>>> to you in error
>>>>>>>>> but does not contain patient information, please contact the
>>>>>>>>> sender and properly
>>>>>>>>> dispose of the e-mail.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> Mne_analysis mailing list
>>>>>>>> Mne_analysis at nmr.mgh.harvard.edu
>>>>>>>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>>>>>>>>
>>>>>>>>
>>>>>>>> The information in this e-mail is intended only for the person to
>>>>>>>> whom it is
>>>>>>>> addressed. If you believe this e-mail was sent to you in error and
>>>>>>>> the e-mail
>>>>>>>> contains patient information, please contact the Partners
>>>>>>>> Compliance HelpLine at
>>>>>>>> http://www.partners.org/complianceline . If the e-mail was sent to
>>>>>>>> you in error
>>>>>>>> but does not contain patient information, please contact the sender
>>>>>>>> and properly
>>>>>>>> dispose of the e-mail.
>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> Mne_analysis mailing list
>>>>>>> Mne_analysis at nmr.mgh.harvard.edu
>>>>>>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>>>>>>>
>>>>>>>
>>>>>>> The information in this e-mail is intended only for the person to
>>>>>>> whom it is
>>>>>>> addressed. If you believe this e-mail was sent to you in error and
>>>>>>> the e-mail
>>>>>>> contains patient information, please contact the Partners Compliance
>>>>>>> HelpLine at
>>>>>>> http://www.partners.org/complianceline . If the e-mail was sent to
>>>>>>> you in error
>>>>>>> but does not contain patient information, please contact the sender
>>>>>>> and properly
>>>>>>> dispose of the e-mail.
>>>>>>>
>>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> Mne_analysis mailing list
>>>>>> Mne_analysis at nmr.mgh.harvard.edu
>>>>>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>>>>>>
>>>>>>
>>>>>> The information in this e-mail is intended only for the person to
>>>>>> whom it is
>>>>>> addressed. If you believe this e-mail was sent to you in error and
>>>>>> the e-mail
>>>>>> contains patient information, please contact the Partners Compliance
>>>>>> HelpLine at
>>>>>> http://www.partners.org/complianceline . If the e-mail was sent to
>>>>>> you in error
>>>>>> but does not contain patient information, please contact the sender
>>>>>> and properly
>>>>>> dispose of the e-mail.
>>>>>>
>>>>>>
>>>>>
>>>>
>>>> _______________________________________________
>>>> Mne_analysis mailing list
>>>> Mne_analysis at nmr.mgh.harvard.edu
>>>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>>>>
>>>>
>>>> The information in this e-mail is intended only for the person to whom
>>>> it is
>>>> addressed. If you believe this e-mail was sent to you in error and the
>>>> e-mail
>>>> contains patient information, please contact the Partners Compliance
>>>> HelpLine at
>>>> http://www.partners.org/complianceline . If the e-mail was sent to you
>>>> in error
>>>> but does not contain patient information, please contact the sender and
>>>> properly
>>>> dispose of the e-mail.
>>>>
>>>>
>>>
>>> _______________________________________________
>>> Mne_analysis mailing list
>>> Mne_analysis at nmr.mgh.harvard.edu
>>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>>>
>>>
>>> The information in this e-mail is intended only for the person to whom
>>> it is
>>> addressed. If you believe this e-mail was sent to you in error and the
>>> e-mail
>>> contains patient information, please contact the Partners Compliance
>>> HelpLine at
>>> http://www.partners.org/complianceline . If the e-mail was sent to you
>>> in error
>>> but does not contain patient information, please contact the sender and
>>> properly
>>> dispose of the e-mail.
>>>
>>>
>>
>> _______________________________________________
>> Mne_analysis mailing list
>> Mne_analysis at nmr.mgh.harvard.edu
>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>>
>>
>> The information in this e-mail is intended only for the person to whom it
>> is
>> addressed. If you believe this e-mail was sent to you in error and the
>> e-mail
>> contains patient information, please contact the Partners Compliance
>> HelpLine at
>> http://www.partners.org/complianceline . If the e-mail was sent to you
>> in error
>> but does not contain patient information, please contact the sender and
>> properly
>> dispose of the e-mail.
>>
>>
>
> _______________________________________________
> Mne_analysis mailing list
> Mne_analysis at nmr.mgh.harvard.edu
> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
>
>
> The information in this e-mail is intended only for the person to whom it
> is
> addressed. If you believe this e-mail was sent to you in error and the
> e-mail
> contains patient information, please contact the Partners Compliance
> HelpLine at
> http://www.partners.org/complianceline . If the e-mail was sent to you in
> error
> but does not contain patient information, please contact the sender and
> properly
> dispose of the e-mail.
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20141221/29d9b97c/attachment.html 


More information about the Mne_analysis mailing list