Previous: Inferences on a fitted value

Original Model: Y_i = b₀ + b₁X_i + e_i,where e_i ~ N(0,s²) and cov(e_i,e_j)=0 for i¹ j.

Prediction Interval for Y_h,new (a future value corresponding to some given x_h)

A point predictor for Y_h,new is Y_h= b₀+b₁x_h, and Y_h,new is a Normal r.v.

However, its variance is 

Thus a 100(1-a )% prediction interval for Y_h,new is given by

Previous: Inferences on a fitted value