enter image description here

I sas this example in the book R in action, the codes in R are as follows:

aggregate(response, by=list(trt), FUN=mean)
aggregate(response, by=list(trt), FUN=sd)
fit<-aov(response ~ trt)

tuk<-glht(fit, linfct=mcp(trt="Tukey"))
plot(cld(tuk, level=.05), col="lightgrey")

It is written in the book that:

From these results, you can see that taking the cholesterol-lowering drug in 5 mg doses four times a day was better than taking a 20 mg dose once per day. The competitor drugD wasn’t superior to this four-times-per-day regimen. But competitor drugE was superior to both drugD and all three dosage strategies for our focus drug

What I understund is:

four times a day is better than one time a day, because the mean is higher. (mean is the right value to watch, isn't?)

DurgE is the best among all, because of the mean.

What I could not understand is:

  1. Why is four times a day better than drugD? The mean of drugD is higher.

  2. How could four times a day be compared with drugD? They are not signifikantlly different. (they both have the alphabet c in this graphic)


1 Answer 1


@yue86231 I think you've simply misread the R manual. The book (as you've quoted) states that DrugD WASN'T (significantly) better than the 4x/day regimen. As you indicated, the post-hoc letters above the plot indicate a shared letter (c).

  • $\begingroup$ So "The competitor drugD wasn’t superior to this four-times-per-day regimen" can also be written as: "drugD wasn't significantly better than four time a day, even though the mean of drugD is higher than four times a day". Now I understand it, thanks! $\endgroup$
    – yue86231
    Commented Dec 26, 2014 at 14:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.