kempton.slatehall {agridat} | R Documentation |
Yields for a Slate Hall Farm 1976 spring wheat trial.
A data frame with 150 observations on the following 5 variables.
rep
rep, 6 levels
row
row
col
column
gen
genotype, 25 levels
yield
yield (grams/plot)
The trial was a balanced lattice with 25 varieties in 6 replicates, 10 ranges of 15 columns. The plot size was 1.5 meters by 4 meters. Each row within a rep is an (incomplete) block.
Field width: 15 columns * 1.5m = 22.5m
Field length: 10 ranges * 4m = 40m
R A Kempton and P N Fox. (1997). Statistical Methods for Plant Variety Evaluation, Chapman and Hall. Page 84.
Julian Besag and David Higdon. 1993. Bayesian Inference for Agricultural Field Experiments. Bull. Int. Statist. Table 4.1.
Gilmour, Arthur R and Robin Thompson and Brian R Cullis. (1994). Average Information REML: An Efficient Algorithm for Variance Parameter Estimation in Linear Mixed Models, Biometrics, 51, 1440-1450.
data(kempton.slatehall) dat <- kempton.slatehall # Besag 1993 figure 4.1 (left panel) if(require(desplot)){ grays <- colorRampPalette(c("#d9d9d9","#252525")) desplot(yield ~ col * row, dat, aspect=40/22.5, # true aspect num=gen, out1=rep, col.regions=grays, # unknown aspect main="kempton.slatehall - spring wheat yields") } # ---------------------------------------------------------------------------- ## Not run: # Incomplete block model of Gilmour et al 1995 require(lme4) require(lucid) dat <- transform(dat, xf=factor(col), yf=factor(row)) m1 <- lmer(yield ~ gen + (1|rep) + (1|rep:yf) + (1|rep:xf), data=dat) vc(m1) ## groups name variance stddev ## rep:xf (Intercept) 14810 121.7 ## rep:yf (Intercept) 15600 124.9 ## rep (Intercept) 4262 65.29 ## Residual 8062 89.79 ## End(Not run) # ---------------------------------------------------------------------------- ## Not run: # Incomplete block model of Gilmour et al 1995 # asreml3 require(asreml) m2 <- asreml(yield ~ gen, random = ~ rep/(xf+yf), data=dat) vc(m2) ## effect component std.error z.ratio constr ## rep!rep.var 4262 6890 0.62 pos ## rep:xf!rep.var 14810 4865 3 pos ## rep:yf!rep.var 15600 5091 3.1 pos ## R!variance 8062 1340 6 pos # Table 4 predict(m2, data=dat, classify="gen")$predictions$pvals ## End(Not run) # ---------------------------------------------------------------------------- ## Not run: # Incomplete block model of Gilmour et al 1995 ## require(asreml4) ## require(lucid) ## m2 <- asreml(yield ~ gen, random = ~ rep/(xf+yf), data=dat) ## vc(m2) ## ## effect component std.error z.ratio bound ## ## rep 4262 6890 0.62 P 0 ## ## rep:yf 15600 5091 3.1 P 0 ## ## rep:xf 14810 4865 3 P 0 ## ## units(R) 8062 1340 6 P 0 ## # Table 4 ## predict(m2, data=dat, classify="gen")$pvals ## ## gen predicted.value std.error status ## ## 1 G01 1280 60.2 Estimable ## ## 2 G02 1550 60.2 Estimable ## ## 3 G03 1420 60.2 Estimable ## ## 4 G04 1450 60.2 Estimable ## ## 5 G05 1530 60.2 Estimable ## End(Not run) # ----------------------------------------------------------------------------