To fill in some background:

I am trying to figure out what sort of model to use for a given set of data. I tried to fit the SLR model and it failed linearity, constant variance, and normality. After modifying it using WLS, all conditions except for normality are satisfied.

I now want to do some statistical inference on the WLS. Without the normality condition, I then turned to bootstrapping. However, how do I create a prediction interval in this case. Can I use the regular formula Ynew = (t)*SE in this case?

Thanks =)