Dear SPM users,
I've ran a full factorial design in VBM where the factor 'diagnosis' (2 levels, OCD-HC) interacts with the covariate 'semantic fluency score' (a continuous variable). If I understood correctly, VBM tests if the beta for the correlation between the two factors is significantly different from 0. Then, in the interaction contrasts, it tests whether there is a significant difference between the slopes of association in the two study population. Did I get it right that such difference is tested comparing the beta? If it is so I could conclude that in certain areas, the relative strength of the parameter of interest (e.g. GM volume) in predicting the continuous variable value, is greater in one sample (compared to the other).
Any help is more than appreciated,
La Sapienza Universita' di Roma
Institute of Medical Psychology and Behavioral Neurobiology