The following example is used to illustrate the process; however, the process
is most useful when we have a large number of independent variables.
Example 8.10 Refer to the data on liver transplants of Example 8.7 consisting
of three independent variables: hepatic systolic pressure at transplant time
(called pressure 1 ), age (at the second measurement time), and gender of the
child. The results for individual terms were shown in Example 8.7; these in-
dicate that the pressure at transplant time (pressure 1 ) is the most significant
variable.
Step 1:Variable PRESSURE1 is entered. The model with only pressure at
transplant time yields
SSR¼ 1460 :47 with 1 dfAnalysis of variables not in the model: With the addition of age, we have
SSR¼ 1796 :61 with 2 df
MSE¼ 47 : 69 ðdf¼ 18 Þleading to anFstatistic of 7.05ðp¼ 0 : 0161 Þ. Variable AGE is entered next; the
remaining variable (gender) does not meet the criterion of 0.1 (or 10%) level to
enter the model.
Step 2:Variable AGE is entered. The final model consists of two indepen-
dent variables with the results shown in Table 8.13.
Note: The SAS program of Example 8.7 should be changed to
PROC REG;
MODEL POST = PRE SEX AGE/SELECTION = STEPWISE;
to specify the stepwise process.
8.3 NOTES ON COMPUTATIONS
Samples of SAS program instructions were provided for all procedures at the
end of Examples 8.1, 8.7, and 8.10. Regression analyses can also be imple-
mented easily using Microsoft’s Excel; however, you needData Analysis,an
TABLE 8.13
Factor Coe‰cient Standard Error FStatistic pValue
Pressure 1 0.3833 0.0806 22.60 0.0002
Age 0.9918 0.3735 7.05 0.0161
NOTES ON COMPUTATIONS 305