Unbalanced Variance Analysis with EFFECTPLOT (ICOMEP2)

Difficulty Level

Beginner

Published on : 14/01/2024

This script illustrates how to analyze a two-factor experimental design (drug and disease) with interactions, on unbalanced data. It uses the GENMOD procedure to fit the model and the EFFECTPLOT statement to generate various plots: effect plots, sliced interactions, boxplots, and mosaic plots.

Data Analysis

Type : CREATION_INTERNE

Data is dynamically generated in the DATA step 'a' via the DATALINES statement. It simulates a clinical study with 'drug', 'disease' variables, and the 'y' response.

1 Code Block

DATA STEP Data

Explanation :
Creation of data table 'a'. The DATA step reads drug and disease identifiers, then loops to read up to 6 'y' measurements per input line.

Copied!

1	DATA a;
2	INPUT drug disease @;
3	DO i=1 to 6;
4	INPUT y @;
5	OUTPUT;
6	END;
7	DATALINES;
8	1 1 42 44 36 13 19 22
9	1 2 33 . 26 . 33 21
10	1 3 31 -3 . 25 25 24
11	2 1 28 . 23 34 42 13
12	2 2 . 34 33 31 . 36
13	2 3 3 26 28 32 4 16
14	3 1 . . 1 29 . 19
15	3 2 . 11 9 7 1 -6
16	3 3 21 1 . 9 3 .
17	4 1 24 . 9 22 -2 15
18	4 2 27 12 12 -5 16 15
19	4 3 22 7 25 5 12 .
20	;

2 Code Block

PROC GENMOD

Explanation :
ODS graphics activation. Execution of PROC GENMOD to model the response 'y' (default normal distribution). 'effectplot / obs' displays the effect plot with observations. 'effectplot interaction(sliceby=disease)' displays interactions by slicing by the disease variable, with confidence limits (clm).

Copied!

1	ods graphics on;
2	PROC GENMOD DATA=a;
3	class drug disease;
4	model y=disease drug disease*drug / d=n;
5	effectplot / obs;
6	effectplot interaction(sliceby=disease) / clm;
7	RUN;

3 Code Block

PROC GENMOD

Explanation :
Second analysis with complementary visualizations: box plots, cross-interactions with observations, mosaic plots, and interactions faceted by disease (plotby).

Copied!

1	PROC GENMOD DATA=a;
2	class drug disease;
3	model y=drug disease drug*disease / d=n;
4	effectplot box;
5	effectplot interaction(x=drug*disease) / obs;
6	effectplot mosaic;
7	effectplot interaction(plotby=disease);
8	RUN;
9	ods graphics off;

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.

Difficulty Level

Published on : 14/01/2024

Data Analysis

1 Code Block

2 Code Block

3 Code Block

Related Documentation