data <- scan(what = list(person=0,treatment="",taste=0), multi.line=F)
1 KFC 6.8  1 Popeyes 7.2  1 Churchs 7.0 
2 KFC 7.7  2 Popeyes 7.4  2 Churchs 6.8 
3 KFC 1.8  3 Popeyes 1.6  3 Churchs 1.5 
4 KFC 6.3  4 Popeyes 7.4  4 Churchs 6.7 
5 KFC 8.1  5 Popeyes 7.9  5 Churchs 7.1 
6 KFC 7.8  6 Popeyes 8.4  6 Churchs 8.0 
7 KFC 8.8  7 Popeyes 9.5  7 Churchs 8.7 

data
biscuits <- data.frame(data)
rm(data)

biscuits$person <- as.factor(biscuits$person)
biscuits$treatment <- as.factor(biscuits$treatment)

boxplot(taste ~ treatment, data=biscuits, main = "Biscuit data")


 # RCB analysis
biscuit.rcb <- lm(taste ~ person +treatment, data=biscuits)

anova(biscuit.rcb)

 # check model assumptions

par(mfrow=c(2,2))  
plot(biscuit.rcb)
par(mfrow=c(1,1))  

 # create the variable to use in performing Tukey's additivity test

biscuits$pred <- biscuit.rcb$fitted.values
biscuits$res <- biscuit.rcb$residuals

biscuits$Tukey <- biscuits$pred^2/(2*mean(biscuits$taste))


 # refit the model with the new variable
 # its P value is for the test of additivity

biscuit.rcb2 <- lm(taste ~ person +treatment +Tukey, data=biscuits)

anova(biscuit.rcb2)

 # if we had rejected the null hypothesis of additivity, we could 
 # have used 1 - betahat  for the estimate of the Tukey variable slope
 # to suggest a power transformation to restore additivity

summary(biscuit.rcb2)