You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In the NBS implementation, when using the default 'both' for tails, the absolute value of the t test is used when thresholding prior to detecting cluster sizes. So an edge with a negative t score can appear to contribute to the 'size' of a cluster which also contains edges with positive t scores. Shouldn't the two conditions be treated separately, constructing one null distribution for positive t-score clusters, and another for negative ones? This would certainly make more sense if the cluster mass was used ('Fundamentals of Brain Network Analysis', Fornito et al., p407). Unfortunately, this does not appear to be discussed in the paper cited either.
The text was updated successfully, but these errors were encountered:
In the NBS implementation, when using the default 'both' for tails, the absolute value of the t test is used when thresholding prior to detecting cluster sizes. So an edge with a negative t score can appear to contribute to the 'size' of a cluster which also contains edges with positive t scores. Shouldn't the two conditions be treated separately, constructing one null distribution for positive t-score clusters, and another for negative ones? This would certainly make more sense if the cluster mass was used ('Fundamentals of Brain Network Analysis', Fornito et al., p407). Unfortunately, this does not appear to be discussed in the paper cited either.
The text was updated successfully, but these errors were encountered: