agridat (version 1.16)

gomez.fractionalfactorial: Rice fractional factorial experiment 1/2 2^6.

Description

Rice fractional factorial experiment 1/2 2^6. Two reps with 2 blocks in each rep.

Format

A data frame with 64 observations on the following 6 variables.

yield

grain yield in tons/ha

rep

replicate, 2 levels

block

block within rep, 2 levels

trt

treatment, levels (1) to abcdef

col

column position in the field

row

row position in the field

Details

Grain yield from a 2^6 fractional factorial experiment in blocks of 16 plots each, with two replications.

Gomez has some inconsistencies. One example:

Page 171: treatment (1) in rep 1, block 2 and rep 2, block 1.

Page 172: treatment (1) in Rep 1, block 1 and rep 2, block 1.

This data uses the layout shown on page 171.

Examples

# NOT RUN {
data(gomez.fractionalfactorial)
dat <- gomez.fractionalfactorial

# trt abcdef has the highest yield
# Gomez, Figure 4.8
if(require(desplot)){
  desplot(yield~col*row, dat,
          # aspect unknown
          text=trt, shorten="none", show.key=FALSE, cex=1,
          main="gomez.fractionalfactorial - treatment & yield")
}

# Split treatment into individual factors
dat <- transform(dat,
                 a = -1 + 2 * grepl('a',trt),
                  b = -1 + 2 * grepl('b',trt),
                 c = -1 + 2 * grepl('c',trt),
                 d = -1 + 2 * grepl('d',trt),
                 e = -1 + 2 * grepl('e',trt),
                 f = -1 + 2 * grepl('f',trt))

# Gomez table 4.24, trt SS totalled together.
# Why didn't Gomez nest block within rep?
m0 <- lm(yield ~ rep * block + trt, dat)
anova(m0)

# Gomez table 4.24, trt SS split apart
m1 <- lm(yield ~ rep * block + (a+b+c+d+e+f)^3, dat)
anova(m1)

# }
# NOT RUN {
  require(FrF2)
  aliases(m1)
  MEPlot(m1, select=3:8,
         main="gomez.fractionalfactorial - main effects plot")
# }
# NOT RUN {
# }