Can you relate in multiple ways? Multiple linear regression and stepwise regression - Improving the User Experience through Practical Data Analytics - page 250

FIGURE 10.7
SPSS output; illustrative example.
There are some complex ways to ind the exact 95% conidence interval for the
prediction, by hand using the Excel output, or by simply using SPSS. However,
assuming a sample size used in the regression that is not very small (a rule of thumb
might be at least n = 25), and for predicting values of the X or X's not too far from
where your data values are, there is a simple formula that provides an approximate
95% conidence interval for the prediction—close enough for virtually any real-
world decision that is based on the conidence interval.
The formula is
Yc±2*(StandardErroroftheEstimate)
where the value, “2,” corresponds to 95% conidence. If you wanted 99% conidence,
you would replace the “2” with “2.6” and for 90% conidence with “1.65.” The really
good news is that the “Standard Error of the Estimate” is obtained from the regres-
sion output, in both Excel and in SPSS. In Figure 10.7 , you can see it in the irst
block of SPSS output (see arrow in Figure 10.7 ); the value is 7.205. If you go back
to Figure 10.2 , the Excel output, you can see the same 7.205 value, called “Standard
Error” by Excel, since its output format does not allow the entire expression to be
written out (see the two dashed arrows in Figure 10.2 ).