help > not parallel: T-statistic and beta value
Showing 1-1 of 1 posts
Jan 15, 2014 08:01 AM | Takashi Hayaku
not parallel: T-statistic and beta value
In ROI-ROI analysis, larger beta values do not always have
corresponding T-statistics. Say, for a group of subjects, I have
two conditions and each subject has 40 ROIs. For condition 1, I get
500 connections; for condition 2, 450 connections. Either when
comparing connections in a given condition or comparing connections
linking the same pair of nodes between the two conditions, values
of T-statistic and beta are not always parallel. For example, a
smaller beta may have a bigger T value.
Generally, t value should be the square root of R*R/ [(1-R*R)/ (n-2)], where n is the sample size. I think, under abovementioned condition, n should be a fixed number. Thus the value of beta and t should always be parallel?
Hope you can clarify this point for me, thank you very much!
Generally, t value should be the square root of R*R/ [(1-R*R)/ (n-2)], where n is the sample size. I think, under abovementioned condition, n should be a fixed number. Thus the value of beta and t should always be parallel?
Hope you can clarify this point for me, thank you very much!