Quadratic TOT IV Model Question

[Scott Holupka]

I'm currently estimating TOT models for an RCT study where I'm using randomization into the Tx group as my IV. I've been able to extend my analysis to examining dosage by replacing Tx receipt with a dosage measure, again using randomization into Tx as my IV. I would also like to examine a quadratic model using linear and squared dosage variables but I'm not sure how to do this if my only IV is the binary randomization term. Is it possible to even run this sort of quadratic IV?

Thanks,

Scott