topic Re: PROC GENMOD Repeated Measures Anova coding help? in Statistical Procedures

PROC GENMOD Repeated Measures Anova coding help?

JOEY21 — Mon, 08 Jun 2020 21:44:57 GMT

Hi all,

I'm trying to analyze if there are differences in Cover Averages of native plants between 4 different treatments over time. My data is organized as follows:

MONTH	BLOCK/TREATMENT	TREATMENT	COVER AVERAGE
1	1/1	1	0.234
2	1/1	1	0.2105
3	1/1	1	0.1675
4	1/1	1	0.0775
5	1/1	1	0
6	1/1	1	0.0255
7	1/1	1	0.1505
8	1/1	1	0.141
9	1/1	1	0.141
1	2/1	1	0.053
2	2/1	1	0.0325
3	2/1	1	0.015
4	2/1	1	0
5	2/1	1	0
6	2/1	1	0.001
7	2/1	1	0.0075
8	2/1	1	0.015
9	2/1	1	0.0155

My code is here:

PROC GENMOD DATA=WORK.IMPORT DESCENDING;
CLASS BLOCK_TREATMENT TREATMENT COVER_AVERAGE MONTH;
MODEL COVER_AVERAGE = BLOCK_TREATMENT TREATMENT MONTH;
REPEATED SUBJECT = BLOCK_TREATMENT / type=IND;
RUN;

Right now my output doesn't make much sense and every online article is over my head. I'm stuck with what I'm doing wrong.

Re: PROC GENMOD Repeated Measures Anova coding help?

PaigeMiller — Mon, 08 Jun 2020 22:13:23 GMT

Show us the output. Explain what parts don't make sense.

Re: PROC GENMOD Repeated Measures Anova coding help?

JOEY21 — Mon, 08 Jun 2020 23:02:13 GMT

My apologies, the output is attached.

Re: PROC GENMOD Repeated Measures Anova coding help?

PaigeMiller — Tue, 09 Jun 2020 11:06:38 GMT

I asked you to provide two pieces of information. You only provided one piece of information... I need the other piece of information.

Re: PROC GENMOD Repeated Measures Anova coding help?

SteveDenham — Tue, 09 Jun 2020 12:41:42 GMT

First off, the response values should not be modeled with a multinomial distribution. That implies levels, and nothing will ever converge with only one observation per level. So, remove cover_average from the CLASS statement. Then you may want to consider what distribution to use to model this variable. I would suggest a beta distribution, but it only has support on the open interval (0, 1), and I see you have some zeroes in your responses. Consider adding a very small bit to the zeroes (say 1e-6) and try fitting this code:

PROC GENMOD DATA=WORK.IMPORTMODIFIED ;
CLASS BLOCK_TREATMENT TREATMENT COVER_AVERAGE MONTH;
MODEL COVER_AVERAGE = BLOCK_TREATMENT TREATMENT MONTH/ dist=beta;
REPEATED SUBJECT = BLOCK_TREATMENT / type=IND;
RUN;

Perhaps a better approach would be to use PROC GLIMMIX, so that a covariance structure could be applied to the repeated measure MONTH.

PROC glimmix DATA=WORK.IMPORTMODIFIED ;
CLASS BLOCK_TREATMENT TREATMENT COVER_AVERAGE MONTH;
MODEL COVER_AVERAGE = BLOCK_TREATMENT TREATMENT MONTH treatment*month/ dist=beta;
random block_treatment;
 random month/ SUBJECT = BLOCK_TREATMENT type=AR(1);
RUN;

My changes are all in lower case.

SteveDenham