Five treatments with two controls may be analysed in strip-plot arrangement n effectiveness of treatment may be observed.

Hi Achal, thanks for your suggestion. What actually i wanna do is something very "simple". I know from y data point that the curves are different from the controls but i need to show with a statistic test they actually are different.

It seemsthat Dunnett test can compare controls against treatments. But not sure if it consider the whole curve (data point) or only each time point. Also my curves are not linear....

Hello Danilo,

Here concerning the commands :

You must install first "multcomp" package with the command (but I think that some other packages do it well too) :

> install.packages("multcomp")

And then load the library :

> library(multcomp)

> Group Value data model summary(glht(model, linfct=mcp(Group="Dunnett")))

Simultaneous Tests for General Linear Hypotheses

Multiple Comparisons of Means: Dunnett Contrasts

Fit: aov(formula = Value ~ Group, data = data)

Linear Hypotheses:

Estimate Std. Error t value Pr(>|t|)

B - A == 0 2.1510 0.7512 2.864 0.0513 .

C - A == 0 1.6262 0.7512 2.165 0.1441

D - A == 0 4.9818 0.7512 6.632 your_table model summary(glht(model, linfct=mcp(Group="Dunnett")))

Please note that the control is the first group.

Concerning other points :

• Dunnett test is a procedure for multiple-to-one comparisons of mean. So it "reveals" significant difference based on mean and standard deviation. By the way, because of this, you should check if your data are following normal distribution...
• Yes, you must enter the "raw data" in R, the value of each point. Mean calculation is part of data analysis.
• If I understand well, you have two control. So, you'll have to run this test twice, for each of them.
• You said you have time course experiment... So what is your response ? Is it dependent to time ? If it is you have to include this variable (the time) in your model.

Hoping this would help.

Regards

Thanks Cyril, it gave me some good point to start with.

Just to answer your questions:

my data are in % therefore i expect they are normal distributed. However, because are obtained detracting the blank, i have some data that are negative and other that are over 100%. I think i need to make =100 all the data >100% and =0 all the data < 0.

What you call "Value" in the script is the response from my time course experiment. all responses have been collected at the same time point (total time point 15*3).

I will try as you suggested.

Thanks again

Danilo