topic Re: Model option of Solution in Proc GLM/Mixed in Statistical Procedures

Model option of Solution in Proc GLM/Mixed

Nancych — Wed, 14 Feb 2018 16:10:00 GMT

Hi,

does anyone know how to interprate the fixed effects in Solutions when the intercept is canceled, such as Drug A, D,G and Gender F as below?

The GLM Procedure

Dependent Variable: Y

Parameter Estimate StandardError t Value Pr > |t|Drug ADrug DDrug GGender FGender M

22.16666667	B	2.58800653	8.57	<.0001
39.06666667	B	2.58800653	15.10	<.0001
59.96666667	B	2.58800653	23.17	<.0001
4.26666667	B	2.58800653	1.65	0.1113
0.00000000	B	.	.	.

Please see my code below:

proc glm data=DrugTest;

class Drug gender;

model Y = Drug GENDER/ noint solution;

run;

Thank you so much!

Nancy

Re: Model option of Solution in Proc GLM/Mixed

Nancych — Wed, 14 Feb 2018 16:12:51 GMT

The results are re-posted as below

Parameter	Estimate	Standard Error	t Value	Pr > \|t\|
Drug A	22.16666667	2.58800653	8.57	<.0001
Drug D	39.06666667	2.58800653	15.10	<.0001
Drug G	59.96666667	2.58800653	23.17	<.0001
Gender F	4.26666667	2.58800653	1.65	0.1113
Gender M	0.00000000	.	.	.

Re: Model option of Solution in Proc GLM/Mixed

Rick_SAS — Wed, 14 Feb 2018 20:49:17 GMT

Do you know how to interpret the coefficients for the intercept case? It's the same idea for the no-intercept model, except the intercept term is assigned to the last level of DRUG.

For males, the mean response is 22, 38, and 60 for Drugs A, D, and G, respectively.

For females, the mean response is 4.3 units higher at each level of DRUG.

If you turn on ODS graphics and put

PLOTS=IntPlot;

on the PROC GLM statement, you can see plots of the regression.

Re: Model option of Solution in Proc GLM/Mixed

Nancych — Thu, 15 Feb 2018 03:02:11 GMT

Thank you very much, it is helpful!