"If you want to predict (=estimate) the value of an individual using
regression, then there will be a degree of uncertainty, arising from the
variability of the observations about this line. Hence the expression."
This is different again. SPSS's SEE is the standard deviation of the
residuals _in the present sample_. The estimated value of Y for an
individual has two sources of uncertainty: the one just referred to, and
the uncertainty in the location of the line, or of the vector of "slopes".
By the way, I was unnecessarily rude about the programmers: the founding
fathers were not so much unprofessional as amateur.
|