Hi @yichao,
In the most basic scenario, where you want to test the hypothesis that the rate of change of e.g. FDC with respect to age is greater than zero, and don’t want to bother with parameter normalisation, you would do:
Name | CI | Age
----------+----+----
Control01 | 1 | 38
Control02 | 1 | 40
Control03 | 1 | 45
Control04 | 1 | 50
t-test: 0 | 1
If there is no relationship between age and FDC, i.e. the null hypothesis is true, then the second GLM beta coefficient will be zero. If you want to test the hypothesis that FDC decreases as a function of age, i.e. the second beta coefficient is negative, then the contrast vector [ 0 -1] would produce positive t-statistics in that instance.
Cheers
Rob