12-4 PREDICTION OF NEW OBSERVATIONS 439The estimated mean response at this point is found from Equation 12-35 asThe variance of is estimated byTherefore, a 95% CI on the mean pull strength at this point is found from Equation 12-38 aswhich reduces toSome computer software packages will provide estimates of the mean for a point of interest x 0
and the associated CI. Table 12-4 shows the Minitab output for Example 12-7. Both the esti-
mate of the mean and the 95% CI are provided.12-4 PREDICTION OF NEW OBSERVATIONSA regression model can be used to predict new or future observationson the response
variable Ycorresponding to particular values of the independent variables, say, x 01 , x 02 , p, x 0 k.
If , a point estimate of the future observation Y 0 at the point x 01 ,
x 02 , p, x 0 kis(12-39)A 100(1)% prediction intervalfor this future observationisyˆ 0 x¿ 0 ̨ˆx¿ 0 3 1, x 01 , x 02 ,p, x 0 k 426.66
Y|x 0
28.6627.662.074 1 0.23266
Y 0 x 0
27.662.074 1 0.232665.2352 1 0.04444 2 0.23266£.214653 .007491 .000340
.007491 .001671 .000019
.000340 .000019 .0000015§ £1
8
275ˆ^2 x 0 ¿ 1 ¿ 2 ^1 x 0 5.2352 3 1 8 275 4 §ˆY 0 x 0ˆY|x 0 x 0 ¿ˆ 31 8 2754 £
2.26379
2.74427
0.01253§27.66This prediction interval is a generalization of the prediction interval given in Equation 11-33
for a future observation in simple linear regression. If you compare the prediction interval
Equation 12-40 with the expression for the confidence interval on the mean, Equation 12-38,
you will observe that the prediction interval is always wider than the confidence interval. The
confidence interval expresses the error in estimating the mean of a distribution, while the pre-
diction interval expresses the error in predicting a future observation from the distribution atY 0
yˆ 0 t2,np 2 ˆ^2 ̨ 11 x¿ 01 ¿ 2 ^1 x 02 (12-40)yˆ 0 t2,np 2 ˆ^2 ̨ 11 x 0 ¿ 1 ¿ 2 ^1 x 02c 12 .qxd 5/20/02 9:33 M Page 439 RK UL 6 RK UL 6:Desktop Folder:TEMP WORK:MONTGOMERY:REVISES UPLO D CH114 FIN L:Quark Files: