On 6/7/2019 2:29 AM, Avi Gharehgazlou wrote:
External Email - Use Caution
Hi experts,
I want to examine between-group differences in LGI controlling for age, sex and surface area (all except sex demeaned). To do so, I ran the DODS model on QDEC selecting:
- “diagnosis” and “sex” as discrete (fixed factors)
- “age” and “surface area” as nuisance factors
- And selected this question --> “Does the average pial_lgi, accounting for sex, differ between TD and NDD? Nuissance factors: Age Surface area” 1) I wanted to ask you what the significant results from the above analysis means? I was under the impression that the significant clusters represent clusters where there are significant group differences in LGI while controlling for age, surface area and sex. Correct. But I am confused because I know from previous threads that you recommend to run DODS to check for interactions first, and if there is no interaction to then run DOSS to see differences between groups. So: 2) Do my significant results mean that I have an interaction between diagnosis and sex OR diagnosis and age OR diagnosis and surface area? No, not from that contrast. QDEC might put out another contrasts that test for that. Otherwise, you will have to use the command line stream. 3) Considering that I have an interaction, how do I now examine between-group differences in LGI? Is what I have done (through the DODS model) correct? If there is an interaction, then interpreting the results becomes more complicated as the size and direction then become dependent on the age and surface area. Test for the interaction and hope that there is none:).
I'd really appreciate your help,
Thanks so much,
Avi
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edumailto:Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer