Sas code and output

Multi-centre analysis cf = post-treatment cold feet (1-5), cfl = pre-treatment cold feet.

Table 4.5 Treatment effect estimates.

Model

Treatment effect (SE)

A-B

A-C

B-C

1 Uncorrelated

0.72 (0.20)

-0.54(0.25)

-1.26 (0.23)

2 Compound symmetry

0.65 (0.29)

-0.61 (0.35)

-1.26 (0.33)

3 Toeplitz

0.65 (0.29)

-0.57(0.35)

-1.21 (0.33)

Table 4.6 Odds ratios calculated from Model 2.

Effect

Odds ratio

A/B

0.52 (0.30, 0.92)

A/C

1.84 (0.93, 3.65)

B/C

3.52 (1.84, 6.73)

Model 1

PROC GENMOD DATA=b; CLASS centre treat cf1; MODEL cf=cf1 treat/ TYPE3 WALD DIST=MULT COVB; ESTIMATE 'A-B' treat 1 -1 0/ ALPHA=0.05 EXP; ESTIMATE 'A-C' treat 1 0 -1/ ALPHA=0.05 EXP; ESTIMATE 'B-C' treat 0 1 -1/ ALPHA=0.05 EXP;

The GENMOD Procedure Model Information

Data Set Distribution Link Function Dependent Variable

WORK.B Multinomial Cumulative Logit cf

Number of Observations Read 2 83 Number of Observations Used 2 83

Class Level Information Class Levels Values centre 29 1 2 3 4 5 6 7 8 9 11 12 13 14 15 18 23 24 25

Response Profile

Ordered Value

Total Frequency

226 16 17 8 16

PROC GENMOD is modeling the probabilities of levels of cf having LOWER Ordered Values in the response profile table. One way to change this to model the probabilities of HIGHER Ordered Values is to specify the DESCENDING option in the PROC statement.

Parameter Information

Parameter

Effect

treat

cfl

Prml

cfl

l

Prm2

cfl

2

Prm3

cfl

3

Prm4

cfl

4

Prm5

cfl

5

Prm6

treat

A

Prm7

treat

B

Prm8

treat

C

Criteria For Assessing Goodness Of Fit Criterion DF Value Value/DF

Log Likelihood -183.0400

Algorithm converged.

Analysis Of Parameter Estimates

Standard Wald 95% Chi-

Parameter DF Estimate Error Confidence Limits Square Pr > ChiSq

Intercept1

l

-l.

3254

0.

8l52

-2.

923l

0.

2724

2.

64

0.

l040

Intercept2

l

-0.

. 7739

0.

8090

-2.

3594

0.

8ll7

0.

92

0.

3388

Intercept3

l

0.

. 0703

0.

8003

-l.

4982

l.

6388

0.

0l

0.

9300

Intercept4

l

0.

.6673

0.

7995

-0.

8998

2.

2343

0.

70

0.

.4040

cf1

l

l

4.

. 0832

0.

7794

2.

5556

5.

6l09

27.

44

<.

000l

cf1

2

l

3.

l632

l.

0050

l.

l935

5.

l329

9.

9l

0.

00l6

cf1

3

l

l.

2l27

0.

8428

-0.

439l

2.

8646

2.

07

0.

l502

cf1

4

l

2.

967l

l.

4723

0.

08l4

5.

8529

4.

06

0.

0439

cf1

5

0

0.

0000

0.

0000

0.

0000

0.

0000

treat

A

l

-0.

8745

0.

4709

-l.

7974

0.

0484

3.

45

0.

0633

treat

B

l

-l.

. 5292

0.

4536

-2.

4l82

-0.

6402

ll.

37

0.

0007

treat

C

0

0.

0000

0.

0000

0.

0000

0.

0000

Scale

0

l.

0000

0.

0000

l.

0000

l.

0000

NOTE: The scale parameter was held fixed.

Wald Statistics For Type 3 Analysis Chi-

Source DF Square Pr > ChiSq cf1 4 53.20 <.0001

treat 2 11.79 0.0028

Contrast Estimate Results

Standard

Label Estimate Error Alpha Confidence

Chi-

Limits Square Pr > ChiSq

0.6547 0.3640 0.05

1.9246 0.7005 0.05

0.4171 0.1964 0.05

0.2167 0.0983 0.05

0.9430 3.9280

0.1657 1.0496

0.0891 0.5272

0.0721

0.0633

0.0007

Was this article helpful?

0 0

Post a comment