Drink: Water Temperature Effect on Cow Drinking

In byandell/pda: Practial Data Analysis for Designed Experiments

Description

A dairy scientist is interested in the effect on milk yield of feeding cows hot (lukewarm, actually) instead of cold water. This may have economic importance if the temperature of water can alter milk yield by even a pound per week. Animals were put on hot (or cold) water for three weeks, with measurements taken in the final week (as 7-day milk yield) of the period. Each cow was given both hot and cold water over a six week (two periods), with cows randomized as to whether they received hot or cold water first in each pair. Cows might be treated over several pairs of periods during the course of the study. Milk yield should gradually decrease over time, regardless of treatment. This decline is confounded with the hot/cold treatment for any given cow, but can be sorted out by comparing cows given hot or cold first. There is a covariate dim (days in milk) that indicates how long the cow has been producing milk; milk yield tends to rise initially and then gradually fall, with a total lactation (milk producing) time of roughly 305 days. In addition the month of entry into the study is included to help assess seasonal effects if any.

Usage

data(Drink)

Format

Drink data frame with 261 observations on 6 variables.

[,1]	cow	factor	cow identifier
[,2]	hc	factor	Heifer or Cow
[,3]	trt	factor	treatment group
[,4]	per	factor	period (Early/Middle/Late)
[,5]	dmi	numeric	dry matter intake (DMI)
[,6]	coce	factor	plot code

Source

Dave Combs

References

Wattiaux MA, Combs DK and Shaver RD (1994) 'Lactational responses to ruminally undegradable protein by dairy cows fed diets based on alfalfa silage', J Dairy Science 77, 1604-1617.

Examples

# Average measurement per cow
data( Drink )

Drink2 <- Drink[Drink$period < 3,]
Drink4 <- Drink[!is.na(match(Drink$cow,Drink$cow[Drink$period == 4])),]

Drink$cow <- factor( Drink$cow )
Drink$month <- ordered( Drink$month )
Drink$period <- ordered( Drink$period )
Drink.fit <- aov( milk ~ cow + period * water, Drink )

Drink2$cow <- factor( Drink2$cow )
Drink2$month <- ordered( Drink2$month )
Drink2$period <- ordered( Drink2$period )
Drink2.fit <- aov( milk ~ cow + period * water, Drink2 )

Drink4$cow <- factor( Drink4$cow )
Drink4$month <- ordered( Drink4$month )
Drink4$period <- ordered( Drink4$period )
Drink4.fit <- aov( milk ~ cow + period * water, Drink4 )

# I:27.1 Drink cross-over interaction plots
lsd.plot( Drink2.fit, factors = c("period","water"),
  xpos = 1.25, xlab = "(a) cows in first two periods",
  ylab = "7-day milk yield (lb)",
  main = "Figure 27.1",
  more = TRUE, split = c(1,1,2,1) )
lsd.plot( Drink4.fit, factors = c("period","water"),
  xpos = 1.75, xlab = "(b) cows in all four periods", ylab = "",
  main = "Drink",
  split = c(2,1,2,1) )

# better approach: mixed model fit using lme()
library(lme4)
Drink2.lme <- lmer(milk ~ period * water + (1|cow),
  data = Drink2 )
summary(Drink2.lme)
anova(Drink2.lme)
VarCorr( Drink2.lme)

int.plot( Drink2.lme, factors = c("period","water"),
   xpos = 1.25, xlab = "(a) cows in first two periods",
   ylab = "7-day milk yield (lb)", bar = "ellipse" )

Drink4.lme <- lmer(milk ~ period * water + (1|cow),
  data = Drink4 )
Drink4.lme
anova(Drink4.lme)
VarCorr( Drink4.lme)

int.plot( Drink4.lme, factors = c("period","water"),
   xpos = 1.75, xlab = "(b) cows in all four periods",
   ylab = "", bar = "ellipse" )
int.plot( Drink4.lme, factors = c("water","period"),
   xpos = 1.75, xlab = "(c) cold vs hot",
   ylab = "", bar = "ellipse" )

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