[R-sig-ME] Treatment:Block interactions in lme (Schabenberger & Pierce 2002)

[Christoph Scherber]

Dear all,

I have tried to reproduce the SAS example from Schabenberger & Pierce (2002, p. 474 f.), but
couldn´t figure out so far how to fit a farm:treatment (tx) random interaction term.

The data are structured as follows:
- yield is the response variable
- tx (treatment) is the only fixed effect
- farm, block and tx are random effects

Schabenberger & Pierce fit the following model:

fixed effects: tx
random effects: block/farm plus *random interaction term for farm and treatment*

I´ve started with lme as follows:

farmdata=data.frame(farm,tx,block,yield)
m1=lme(yield~treatment,random=~1|farm/block/treatment,farmdata)

But this is obviously wrong as I´d be interested in the treatment:farm interaction.

I guess I´d need to set this up using the pdMat() classes (pdIdent, pdBlocked)?

Any help would be greatly appreciated!

Best wishes
Christoph

(using R 2.15.1 on Windows 7 Professional 32-bit).

####

farm	tx	block	yield
1	A	1	30.8601
1	A	2	30.3216
1	A	3	32.3135
1	B	1	33.3089
1	B	2	30.943
1	B	3	35.2381
2	A	1	31.3924
2	A	2	30.6227
2	A	3	29.9349
2	B	1	27.8756
2	B	2	25.2517
2	B	3	21.7799
3	A	1	39.216
3	A	2	38.9588
3	A	3	35.386
3	B	1	41.9458
3	B	2	43.3782
3	B	3	41.0985
4	A	1	37.1987
4	A	2	36.1024
4	A	3	35.8494
4	B	1	30.9673
4	B	2	32.5553
4	B	3	33.0424
5	A	1	24.9758
5	A	2	22.0432
5	A	3	22.9343
5	B	1	23.3903
5	B	2	24.5053
5	B	3	23.2413
6	A	1	28.059
6	A	2	27.9827
6	A	3	25.1334
6	B	1	28.6984
6	B	2	25.6783
6	B	3	25.8853
7	A	1	27.8232
7	A	2	25.3218
7	A	3	26.5197
7	B	1	37.2286
7	B	2	34.4472
7	B	3	32.4972
8	A	1	29.4101
8	A	2	26.6327
8	A	3	29.6004
8	B	1	30.9808
8	B	2	30.7152
8	B	3	30.6305




-- 
PD Dr Christoph Scherber
Georg-August University Goettingen
Department of Crop Science
Agroecology
Grisebachstrasse 6
D-37077 Goettingen
Germany
phone 0049 (0)551 39 8807
fax 0049 (0)551 39 8806
http://www.gwdg.de/~cscherb1