BactRoom: Bacteria Virus in Plant Sap by Room

In byandell/pda: Practial Data Analysis for Designed Experiments

Description

An experiment performed at the National Wildlife Health Research Center under the supervision of Michael Samuel (Samuel, Goldberg, Thomas and Sharp 1995) examined the effect of a certain bacteria strain (mycoplasma) on the development of birds. Unfortunately, due to the danger of aerial infection, they had to isolate treatment groups in different rooms. A very conservative approach would take each chamber as an experimental unit, regardless of the number of birds per room. There were four treatments: cold, cold+myco, warm and warm+myco. In addition, the experiment was run in two runs since there were only 4 rooms.

Measurements were taken on almost 200 chicks in these rooms. It is possible to think of the experiment as having chicks (or eggs) randomly assigned to the rooms, and to consider the sample in each room as a random sample from a population of chicks exposed to that environment (combination of temperature and presence or absence of mycoplasma bacteria). This experiment on bird development was conducted in two runs separated by several weeks. Several things could have changed in that time, including the mycoplasma culture, seasonal changes of chick growth and food or water conditions. The scientist inoculated eggs in the first run, but decided to switch to inoculating young chicks in the later run. Earlier analysis in the text has assumed that inoculation could be just considered as another factor. Here it is viewed as a blocking factor with no replication. That is, strictly speaking it is not possible to assess the main effect of inoculation method since there is no replication of runs. However, it would be possible to assume that interactions with run were interactions with inoculation.

Usage

data(BactRoom)

Format

BactRoom data frame with 8 observations on 7 variables.

[,1]	bact	factor	bacteria or control
[,2]	temp	factor	room temperature
[,3]	inoc	factor	when innoculated
[,4]	bill	numeric	bill length
[,5]	leg	numeric	leg lenght
[,6]	code	factor	plot code
[,7]	trt	trt	bact*trt combination

Source

Michael D Samuel (mailto:[e-mail]), National Wildlife Health Center, Madison, WI (http://www.emtc.nbs.gov/nwhchome.html)

References

Samuel MD, Goldberg DR, Thomas CB and Sharp P (1995) “Effects of Mycoplasma anatis and cold stress on hatching success and growth of mallard ducklings,” J of Wildlife Diseases 31, 172-178.

See Also

Bacteria

Examples

data( BactRoom )
BactRoom$trt <- factor( BactRoom$trt )
BactRoom$code <- factor( BactRoom$code )
contrasts( BactRoom$temp ) <- contr.sum
contrasts( BactRoom$bact ) <- contr.sum
contrasts( BactRoom$inoc ) <- contr.sum

BactRoom.badd <- aov( bill ~ temp + bact + inoc, BactRoom )
summary( BactRoom.badd )

BactRoom.full <- aov( bill ~ temp * bact, BactRoom )
lsmean( BactRoom.full )
summary( BactRoom.full )

BactRoom.bill <- aov( bill ~ temp * bact * inoc, BactRoom )
lsmean( BactRoom.bill )

# C:8.1 Bacteria Interaction Plots

BactRoom.trt <- aov( bill ~ trt * inoc, BactRoom )
int.plot( BactRoom.trt, split = c(1,1,2,1), more = TRUE,
  bar.plot="none",
  xlab="(a) treatment by inoc", ylab="bill length (mm)",
  main = "Figure C:8.1a Wrong Way" )

int.plot( BactRoom.bill, bar.plot="none", split = c(2,1,2,1),
  xlab="(b) temp by bact", ylab="", lty = c(1,1,2,2),
  main = "Figure C:8.1b Right Way", white = FALSE )

# C:8.2 Bacteria Two Factor Interaction Plots

lsd.plot( BactRoom.full,
  xlab="(a) interaction plot", ylab="bill length (mm)",
  ypos=40, lty = 1,
  main = "Figure C:8.2a Interaction with LSD",
  split = c(1,1,2,1), more = TRUE )

margin.plot( BactRoom.full, ylim=range( BactRoom$bill ),
  effects=TRUE, xlab="(b) margin plot", ylab="",
  ypos=40, lty = 1, split = c(2,1,2,1),
  main = "Figure C:8.2b Margin Plot with LSD" )

# C:8.3 Bacteria Three Factor Interaction Plots

Bacteria.ylab <- c("","bill length (mm)")
Bacteria.xlab <- c("(a)","(b)")
Bacteria.main = c( "Figure C:8.3", "Three-Factor Interactions" )
Bacteria.inoc = levels( BactRoom$inoc )
for ( i in 1:2) {
  tmpdata = BactRoom[ BactRoom$inoc == Bacteria.inoc[i], ]
  print( xyplot( bill ~ temp, tmpdata, groups = bact, type="b",
    xlab = paste( Bacteria.xlab[i], Bacteria.inoc[i], "inoc" ),
    ylab = Bacteria.ylab[i],
    main = Bacteria.main[i] ),
    split = c(i,1,2,1), more = ( i == 1 ))
}
rm( Bacteria.xlab, Bacteria.ylab, Bacteria.main, Bacteria.inoc, tmpdata )

# C: Bacteria Half-Normal and Effect Plots
# order -- temp:bact:inoc temp:bact bact:inoc temp:inoc inoc bact temp

BactRoom.coef = sort( abs( coef( BactRoom.bill )[-1] ))
BactRoom.coef = BactRoom.coef * sqrt( length( BactRoom.coef ))

qhalfnorm = function(p)qnorm((1+p)/2)
qhalfquant <- qhalfnorm(ppoints(length(BactRoom.coef)))
print( qqmath( ~ BactRoom.coef, distribution = qhalfnorm,
  prepanel = prepanel.qqmathline,
  xlim = range(qhalfquant) + c(-.2,.2), ## make room for text
  panel = function(x, ...) {
    panel.qqmathline(x, distribution = qhalfnorm, lty = 2 )
    panel.text(qhalfquant, x, names(BactRoom.coef))
  },
  xlab = "(a) half-normal quantiles", ylab = "effects",
  main = "Figure C:8.4a Half-Normal Plot" ))

effect.plot( BactRoom.bill, xlim = c(.5,7.5), xaxt = "n",
   xlab = "(b) terms", ylab = "MS adjusted effects" )
title( "Figure C:8.4b Effects Plot" )
axis( 1, seq(1,7,by=2), c("T","I","IT","BIT") )
axis( 1, seq(2,6,by=2), c("B","BT","BI") )

byandell/pda documentation built on May 13, 2019, 9:27 a.m.