R help - Jan 2003 - Good advices or books on estimating the SPE?

Dear R'ers

I am trying to use some distribution to test the hypothesis
on my prediction error.
Given a large numbe of sample data with unknown distribution, I first find
some funtion to make regression.
Now for each new data point, I can calculate the prediction error xi. So how
to represent the SPE = sum_{xi^2}
in a general and reliable way with some statistics, like chi-square, t?
Please give me some advice or reference books on this topic.

Thanks for your time and attention.

Fred