Is it reasonable to chose p<0.002 as the clusterize threshold?

Hi AFNI experts,
I ran 3dClustSim with my data and found the cluster could not survive at p=0.001. But it could survive at p=0.002.
I remember that Cox mentioned the threshold could be loosed to p=0.002, but I can’t remember more details. (Bootcamp at ShenZhen last December)
Could you please tell me where can I find the reference for choosing p=0.002 so I can use it to support my study?
Thank you very much!

Thresholding is quite arbitrary to start with. A couple of options:

  1. Use -ETAC with 3dttest++;
  2. Use p=0.002 or even 0.005 as your threshold if it’s an event-related design;
  3. ROI-based analysis:,157054,157054#msg-157054