[FieldTrip] ft_regressconfound for between sample covariates (e.g., age)

[Paul Dhami]

Greetings,

I have a sample in which for each person, I have calculated their
time-frequency response during a task. Accordingly, I have a
time-frequency-channel representation for all individuals across their
trials (I did not keep individual trials).

Before performing cluster-based statistics, I'm interested in removing the
variance from some demographic and clinical variables (e.g., age). Can this
be done with the ft_regressconfound function? The only mention I could find
of this being possible with between sample data (and not at the individual
trial level) is here:
https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fmailman.science.ru.nl%2Fpipermail%2Ffieldtrip%2F2019-April%2F038828.html&data=05%7C02%7Cfieldtrip%40science.ru.nl%7Cdd23bc0c75b14002b93208ddc3063246%7C084578d9400d4a5aa7c7e76ca47af400%7C1%7C0%7C638881153559609872%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=atlGx3kbfWrRio26TLHEcNfL3%2B9u37wjrmsHe21Nxcg%3D&reserved=0.

To create my between-sample design matrix, I create a matrix which has my
nuisance variables together (with categorical variables converted using
dummyvar and the first column removed). But when calling
ft_regressconfound, I get the issue of too many input variables.

Any insight into this would be helpful.

Paul
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20250714/51afeacb/attachment.htm>