data <- scan(what = list(block=0,treatment="",y=0,x=0), multi.line=F)
  1 Check 23.1 13  1 Full 30.1 7  1 N0 26.4 10  1 P0 26.2 8 
  2 Check 20.9 12  2 Full 31.8 5  2 N0 27.2 9  2 P0 25.3 9 
  3 Check 28.3 7  3 Full 32.4 6  3 N0 28.6 6  3 P0 29.7 7 
  4 Check 25.0 9  4 Full 30.6 7  4 N0 28.5 6  4 P0 26.0 7 
  5 Check 25.1 8  5 Full 27.5 9  5 N0 30.8 5  5 P0 24.9 9 

data



table17.7 <- data.frame(data)
rm(data)

table17.7$block  <- as.factor(table17.7$block)
table17.7$treatment  <- as.factor(table17.7$treatment)


 # look at the data

plot(table17.7$x,table17.7$y,pch=as.numeric(table17.7$treatment),
      col=as.numeric(table17.7$treatment),ylab="y",xlab="x")
legend("topright",levels(table17.7$treatment),pch=1:4,col=1:4)

 # standard ANCOVA assuming equal slopes
 # first declare 'treatment' to be an ordered factor with 'Full' as the 
 #   first level, so that Dunnett's test will compare all levels to it

table17.7$treatment <- ordered(table17.7$treatment, levels = c("Full","Check","N0","P0"))

tab17.7.lm1 <- lm(y ~ x +block +treatment , data=table17.7)
anova(tab17.7.lm1)
summary(tab17.7.lm1)

 # check assumptions

windows()
par(mfrow=c(2,2))  
plot(tab17.7.lm1)
par(mfrow=c(1,1))  


 # calculate adjustedMeans and standard errors from the effects package

library(effects)
adjustedMeans <- effect("treatment",tab17.7.lm1) 
adjustedMeans
adjustedMeans$se


 # multiple comparisons using the multcomp package
 # using Dunnett's method against the first level (defined as "Full" above)

library(multcomp)
summary(glht(tab17.7.lm1,linfct=mcp(treatment="Dunnett")))