help
help > RE: How to use "age" as covariates?
Apr 17, 2015 03:04 AM | Alfonso Nieto-Castanon - Boston University
RE: How to use "age" as covariates?
Hi Hengshuang,
Not exactly, that is an "groups by performance interaction" (corrected by age and gender effects), so basically you would interpret a significant negative-contrast result as indicating that the association between performance and A-B functional connectivity is higher in AD compared to HC. In other words, you are looking at a regression line between performance and functional connectivity, separately for the HC and AD groups, and you are finding that the slope of this regression line for the AD group is higher than the slope for the HC group.
Hope this helps
Alfonso
ps. as post hoc analyses you should actually look at these regression separately within each group, e.g. contrast [0 0 1 0 0 0] or [0 0 0 1 0 0] to help you better interpret these results; e.g. are these associations between A-B FC and performance 'positive' associations -i.e. higher performance associated with higher FC- or 'negative' associations -i.e. higher performance associated with lower FC-? Similarly you could also look at the average A-B functional connectivity, e.g. contrast [1 0 0 0 0 0] or [0 1 0 0 0 0] to again help you properly interpret these results -e.g. are these A-B FC values reflecting 'positive' connectivity or anti-correlations?
Originally posted by Hengshuang LIU:
Not exactly, that is an "groups by performance interaction" (corrected by age and gender effects), so basically you would interpret a significant negative-contrast result as indicating that the association between performance and A-B functional connectivity is higher in AD compared to HC. In other words, you are looking at a regression line between performance and functional connectivity, separately for the HC and AD groups, and you are finding that the slope of this regression line for the AD group is higher than the slope for the HC group.
Hope this helps
Alfonso
ps. as post hoc analyses you should actually look at these regression separately within each group, e.g. contrast [0 0 1 0 0 0] or [0 0 0 1 0 0] to help you better interpret these results; e.g. are these associations between A-B FC and performance 'positive' associations -i.e. higher performance associated with higher FC- or 'negative' associations -i.e. higher performance associated with lower FC-? Similarly you could also look at the average A-B functional connectivity, e.g. contrast [1 0 0 0 0 0] or [0 1 0 0 0 0] to again help you properly interpret these results -e.g. are these A-B FC values reflecting 'positive' connectivity or anti-correlations?
Originally posted by Hengshuang LIU:
Hi,
Recently I encountered a similar question (sorry to borrow the following example):
When using "regression" to compare between two groups (HC and AD) the association between FC and performance:
Select: HC, AD, performance_HC, performance_AD, age, gender; Enter contrast: [0 0 1 -1 0 0]
and then I got negative (in blue) connectivity between area A and area B.
Then how to interpret this? the performance difference between groups (HC>AD) could predict the A-B FC difference between groups (AD>HC)?
Thanks!
Recently I encountered a similar question (sorry to borrow the following example):
When using "regression" to compare between two groups (HC and AD) the association between FC and performance:
Select: HC, AD, performance_HC, performance_AD, age, gender; Enter contrast: [0 0 1 -1 0 0]
and then I got negative (in blue) connectivity between area A and area B.
Then how to interpret this? the performance difference between groups (HC>AD) could predict the A-B FC difference between groups (AD>HC)?
Thanks!