Model checking

Residual plots of scaled residuals are used to detect any outliers or a general lack of normality. The graphs of the scaled residuals in Figure 6.1(a) and Figure 6.1(b) indicate no appreciable departure from normality, nor any definite outlying observations.

Table 6.2 Comparing treatment effects between Models 1 and 6.

Model 1 Model 6 (separate

(compound Toeplitz pattern symmetry) for each treatment)

Model 1 Model 6 (separate

(compound Toeplitz pattern symmetry) for each treatment)

Overall treatment effects

A -

B

1.22

1.03)

1.25

0.99)

A -

C

3.01

1.02)

3.04

1.08)

B -

C

1.79

1.03)

1.79

1.00)

Visit 3 treatment effects

A-

B

1.35

1.26)

1.36

1.26)

A-

C

3.40

1.25)

3.42

1.29)

B-

C

2.05

1.28)

2.06

1.24)

Visit 4 treatment effects

A-

B

0.56

1.28)

0.56

1.28)

A-

C

1.86

1.27)

1.89

1.30)

B-

C

1.30

1.28)

1.34

1.25)

Visit 5 treatment effects

A-

B

2.91

1.30)

3.00

1.31)

A-

C

4.67

1.29)

4.77

1.32)

B-

C

1.76

1.29)

1.77

1.26)

Visit 6 treatment effects

A-

B

0.05

1.32)

0.09

1.33)

A-

C

2.09

1.30)

2.10

1.34)

B-

C

2.04

1.31)

2.01

1.27)

Treatment-visit p-value

0.22

0.11

Overall treatment effects in

1.23

1.02)

1.23

1.02)

model omitting treatment-visit effects

3.03

1.03)

3.01

1.02)

1.80

1.03)

1.78

1.03)

Figure 6.1(a) Residual plots ( using outp and proc gplot).
Figure 6.1(b) Residual plots and statistics (via residual option, sas Version 9).

Was this article helpful?

0 0

Post a comment