comparing means with their base values

[Bhupinder]

Hi,

I have soil nutrients dataset from three years recieving different levels of treatments. My interest is to know whether there was any significant change occured over the time or not. So, basically I want to measure the change occured at the end of first year vs at the end of third year.

My question is do we need to set up my dependent variable as change measured or I can simply run it as a repeated measures to observed the change.

Thanks
Bhupinder

[Doc_Duke]

Those are two ways to attack it. They involve different assumptions, so there is no uniformly "best" way.

There is a third way, and that is to include all three time points in a GLM as dependent variables. Yet different assumptions. However, it allows you to explicitly test if the change over time is linear or not (using orthogonal polynomials).

[Bhupinder]

I have few random parameters so have to use MIXED model. The other things is that it is not possible to compare the data from second year due to some restrictions. So, basically, I have only two years data (2007 and 2009). Do I need to set it up as repeated measures or can use them both as dependent variables?

[Bhupinder]

I have few random parameters so have to use MIXED model. The other things is that it is not possible to compare the data from second year due to some restrictions. So, basically, I have only two years data (2007 and 2009). Do I need to set it up as repeated measures or can use them both as dependent variables?