Depending on the complexity of your experimental design structure , this could be analysed as a paired or matched t test, where you have a before and after within a subject as a pair.

As there is likely to a lack of independence of the two measure you can look at the difference between the before and after as the response variable.

Dear Miranda, thanks for your reply. Do you mean that repeated measures ANOVA is not the right approach?

Your before and after measures are likely to have some sort of relationship. What are you looking at - is it a complex design - like a factorial ? If it is simple I would do a paired t test .

I have two groups, wt and ko.

1- We test the subjects at baseline

2- Then we apply a treatment

3- We test again after the treatment

Ultimately we want to know if

a- they behave differently at baseline because of their different genetic makeup and

b-i f they have had a different response to the treatment because of their different genetic makeup.

I hope that all makes sense?

This example is something similar using glm.

It may help, especially since it uses Preweight as an offset

## Example with offsets from Venables & Ripley (2002, p.189)

library(MASS)

utils::data(anorexia, package = "MASS")

# Factor of three levels: "Cont" , "CBT" and "FT"

anorex.1