Hi Stata listers,

I am running an inverse hyperbolic sine regression with the survey command and have a question about back-transforming the inverse hyperbolic sine variable to create predictions. Here is my code:

Code:
gen ihs_inc=asinh(income)

svyset [pw=fex]

svy: reg ihs_inc i.male i.edu age

predict yhat

replace yhat=sinh(yhat)
That's a good start I think, but the next thing I want to do is:

Code:
replace yhat=exp(e(mse)/2)
Which, I believe, is the proper way to complete the reversal of the inverse hyperbolic sine transformation (please correct me if I am wrong).

However (and herein lies the problem), because I am using svy to regress, Stata does not produce e(mse). (I think this is linked to the fact that svy uses robust standard errors?)

Does anyone have any thoughts for how to get around this problem so that I can create the prediction?

Thanks very much in advance,
Jimmy