Bacteria: Bacteria virus in plant sap
In byandell/pda: Practial Data Analysis for Designed Experiments

Description Usage Format Source References See Also Examples

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. <P> 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.

1	data(Bacteria)

Bacteria data frame with 192 observations on 8 variables.

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

Michael D Samuel (mailto:michael_samuel@nbs.gov), National Wildlife Health Center, Madison, WI (http://www.emtc.nbs.gov/nwhchome.html)

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.

Bacteria

# full data set--code not ready yet!
# model fits for bill and leg
data( Bacteria )
Bacteria$id <- factor( Bacteria$id )
Bacteria$cage <- factor( Bacteria$cage )
Bacteria$day <- ordered( Bacteria$day )
contrasts( Bacteria$temp ) <- contr.sum
contrasts( Bacteria$bact ) <- contr.sum
contrasts( Bacteria$inoc ) <- contr.sum

Bacteria.bill <- aov( bill ~ temp * bact * inoc, Bacteria )
Bacteria.leg <- aov( leg ~ temp * bact * inoc, Bacteria )

Bacteria.rcbd <- aov( bill ~ inoc + bact * temp, Bacteria )

# D:10.2 Bacteria Effect Plot by Bird

effect.plot( Bacteria.bill, xaxt = "n" )
title( "D:10.2 Bacteria Effect Plot by Bird" )
axis( 1, 1:8, c("temp","bact","inoc","T:B","T:I","B:I","T:B:I","resid") )

# H:22.2 Bacteria Interaction Plot Adjusted for Runs

effect.plot( Bacteria.rcbd,
   xlab="(a) terms", ylab="MS adjusted effects",
   xlim = c(0.5,5.5), xaxt = "n",
   width = .2 )
title( "H:22.2a Effect Plot Adjusted by Run" )
axis( 1, 1:5, c("run","bact","temp","inter","resid") )

margin.plot( Bacteria.rcbd, factors = c("temp","bact"),
  xlab = "(b) temperature by bacteria",
  ylab = "bill length (mm)",
  main = "H:22.2a Margin Plot Adjusted by Run" )