Multinomial Generalized Logit Modeling with PROC BGLIMM

The script creates a 'school' dataset containing learning preferences. It then uses the BGLIMM (Bayesian Generalized Linear Mixed Models) procedure to fit statistical models on a nominal response variable ('Style') with a multinomial distribution and a generalized logit (glogit) link function. Two models are tested: one with interaction and one without, with specific contrast estimations.

Data Analysis

Type : CREATION_INTERNE

Data is directly defined within the script via a DATA Step using DATALINES ('school' dataset).

1 Code Block

DATA STEP Data

Explanation :
Creation of the 'school' dataset. The INPUT statement uses the '@@' symbol (double trailing @) to read multiple observations per data line.

Copied!

1	DATA school;
2	LENGTH Program $ 9;
3	INPUT School Program $ Style $ Count @code_sas_json/8_SAS_Intro_ReadFile_MultiCol_@@.json;
4	DATALINES;
5	1 regular self 10 1 regular team 17 1 regular class 26
6	1 afternoon self 5 1 afternoon team 12 1 afternoon class 50
7	2 regular self 21 2 regular team 17 2 regular class 26
8	2 afternoon self 16 2 afternoon team 12 2 afternoon class 36
9	3 regular self 15 3 regular team 15 3 regular class 16
10	3 afternoon self 12 3 afternoon team 12 3 afternoon class 20
11	;

2 Code Block

PROC BGLIMM

Explanation :
Execution of the BGLIMM procedure to fit a model including the main effects 'School', 'Program', and their interaction 'School*Program'. The 'Count' variable is used as a frequency variable.

Copied!

1	PROC BGLIMM DATA=school seed=123 dic;
2	freq Count;
3	class School Program;
4	model Style(ref="class")=School Program School*Program / dist=multinomial
5	link=glogit;
6	RUN;

3 Code Block

PROC BGLIMM

Explanation :
Second execution of PROC BGLIMM on a model without interaction. ESTIMATE statements are added to calculate and test specific contrasts (comparisons between schools and programs) with the 'exp' option to exponentiate results (odds ratios) and 'bycat' for estimations by response category.

Copied!

1	PROC BGLIMM DATA=school seed=123 dic;
2	freq Count;
3	class School Program;
4	model Style(ref="class")= School Program / dist=multinomial
5	link=glogit;
6	estimate 'School 1 vs 3' School 1 0 -1 / exp bycat;
7	estimate 'School 2 vs 3' School 0 1 -1 / exp bycat;
8	estimate 'Afternoon vs Regular' Program 1 -1 / exp bycat;
9	RUN;

This material is provided "as is" by We Are Cas. There are no warranties, expressed or implied, as to merchantability or fitness for a particular purpose regarding the materials or code contained herein. We Are Cas is not responsible for errors in this material as it now exists or will exist, nor does We Are Cas provide technical support for it.

Retour à la liste