agridat (version 1.5)

graybill.heteroskedastic: Wheat varieties with heteroskedastic yields

Description

Wheat varieties with heteroskedastic yields

Usage

data(graybill.heteroskedastic)

source

F. A. Graybill, 1954. Variance heterogeneity in a randomized block design, Biometrics, 10, 516-520.

Details

Yield of 4 varieties of wheat at 13 locations in Oklahoma, USA. The data was used to explore variability between varieties.

References

Hans-Pieter Piepho, 1994. Missing observations in the analysis of stability, Heredity, 72, 141--145.

Examples

dat <- graybill.heteroskedastic

# Genotypes are obviously not homoscedastic
boxplot(yield ~ gen, dat)

# Shukla stability variance of each genotype, same as Grubbs' estimate
# Matches Piepho 1994 page 143.
# Do not do this!  Nowadays, use mixed models instead.
require("reshape2")
datm <- acast(dat, gen~env)
w <- datm
w <- sweep(w, 1, rowMeans(datm))
w <- sweep(w, 2, colMeans(datm))
w <- w + mean(datm)
w <- rowSums(w^2)
k=4; n=13
sig2 <- k*w/((k-2)*(n-1)) - sum(w)/((k-1)*(k-2)*(n-1))