- Joined
- Dec 18, 2015
- Messages
- 3,216
- Reaction score
- 4,930
I post the ASTRO jobsite data, and it's up to others to interpret. It could be a red flag, or not. Depends very much on context. Context is important. For example, in one analysis of SBRT vs surgery, it was reported...
... that after treatment, SBRT patients would've chosen the opposite treatment at a much higher rate than surgery patients. This is a good example of twisting p-values through a particular viewpoint. Because one could say:
After a lung cancer treatment, surgery or SBRT patients would choose surgery at a non-dissimilar rate (p=0.3).
OR
After SBRT, SBRT patients would choose an opposite treatment at a much different rate than surgery patients would (p<0.001).
That said, I have compared job "web availability" (compute the residents per year/instantaneous web job count ratio) for rad onc vs other specialties and it's always p<0.05 for rad onc job web availability vs others. I've never seen it not be.