It's because of the robust part, Robust averaging is of course not the
same as just averaging.
Vladimir
On 24 Apr 2012, at 17:25, Deborah Talmi wrote:
> Hi, yes I merged then robust-averaged. there must be a difference
> between averaging over trials/subjects becuase now I used grandmean
> and it is perfect.
> thanks for your suggestion and prompt help
> d
>
> ------------------------
> Deborah Talmi, Ph.D.
> School of Psychological Sciences, University of Manchester
> (0)161 275 1968
>
> ________________________________________
> From: Vladimir Litvak [[log in to unmask]]
> Sent: Tuesday, April 24, 2012 5:19 PM
> To: Deborah Talmi
> Subject: Re: eeg question
>
> Did you merge all the subjects and then averaged? Try using the grand
> mean.
>
> Vladimir
> On 24 Apr 2012, at 17:14, Deborah Talmi wrote:
>
>> Hi,
>> All files are in the same order. I didn't use spm_eeg_grandmean, I
>> averaged across subjects using spm_eeg_average with the exact same
>> options selected. But perhaps this was wrong - will using the
>> grandmean function give different results?
>> Cheers,
>> d
>>
>> ------------------------
>> Deborah Talmi, Ph.D.
>> School of Psychological Sciences, University of Manchester
>> (0)161 275 1968
>>
>> ________________________________________
>> From: Vladimir Litvak [[log in to unmask]]
>> Sent: Tuesday, April 24, 2012 1:25 PM
>> To: Deborah Talmi
>> Cc: [log in to unmask]
>> Subject: Re: eeg question
>>
>> Hi Debbie,
>>
>> The order in the grand average file might be affected by the order
>> specified by D.condlist (that you can change via Other/Sort
>> conditions). If D.condlist was the same for all individual files then
>> the grand average will have the same order, but if not the order in
>> the grand average will be the same as in the first subject which
>> might
>> be different from what you assumed in your code. The best thing to do
>> I think is to make sure that all the individuals have the same
>> D.condlist in the epoched files before computing the averages. Then
>> the averages will be in this order and it will be also the physical
>> order of trials in the data files and will be carried over to the
>> grand average.
>>
>> Once you sort that out the other thing to pay attention to is the
>> weighted/unweighted option when computing the grand mean. What you
>> did
>> should be the same as 'unweighted' but you might have specified
>> 'weighted' when using spm_eeg_grandmean.
>>
>> If those two things don't explain it then ask me again and show me
>> an example.
>>
>> Best,
>>
>> Vladimir
>>
>> On Tue, Apr 24, 2012 at 12:37 PM, Deborah Talmi
>> <[log in to unmask]> wrote:
>>> Hi Vladimir,
>>> Please let me know if you prefer for me to post this on the list...
>>>
>>> I've extracted data from individual subject baseline corrected,
>>> average
>>> files using
>>>
>>> D = spm_eeg_load(file);
>>> a=conditions(D);
>>> intr=[0.24 0.25];
>>> d(con, 1)= mean(D.selectdata('Fz' , intr, a{con}));
>>>
>>>
>>> I then merged all individual files, averaged them, filtered,
>>> baseline
>>> corrected.
>>>
>>> The data I get from the extraction is not the same as the data I
>>> get from
>>> the grand average. Not only are the numbers different, the order of
>>> conditions is also different.
>>>
>>> Could you possibly shed any light on this, please?
>>>
>>> Cheers,
>>> Debbie
>>>
>>> ------------------------
>>> Deborah Talmi, Ph.D.
>>> School of Psychological Sciences, University of Manchester
>>> (0)161 275 1968
>
|