[R-sig-eco] How to include initial variation when testing temporal differentiation among treatments?

[V. Coudrain]

Dear all,

I aim to assess the impact of different treatments on a response variable over time. Specifically at time 0, after all samples have been treated alike for some period, I have recorded initial values of my response variables and implemented three different treatments. Then, I have assessed the value of my response variables after 2 and 4 years to determine whether the samples evolved differently according to the different treatments. 
For modelling the system, I thought about computing a linear model with an interaction among time and treatments, with the idea that a significant interaction will indicate differentiation among treatments. However, I do not know how to handle the values at time 0, because, although at the start of the experiment the treatments were not implemented, the initial values and variation of the samples belonging to each "future" treatment should be taken into account. Can I just treat them as normal treatment points even if the treatments wer not yet implemented?

Thank you,
Valerie

___________________________________________________________
Mode, hifi, maison,… J'achète malin. Je compare les prix avec 
	[[alternative HTML version deleted]]