If by a test you mean something that produces a p-value, you shouldn't do it. Your research apparently is applied, whicgh means you must be able to make quantitative comparisons. I would suggest a 2x3 factorial linear model (SPSS calls this Generalized Linear Models). Make sure to check that you want a coefficient table with confidence limits in the output. These are the numbers that tell you how strong the effects are, and how uncertain.
As Time is within-subject, you should run a multi-level model, with Time on participant level. That is very important for intervention studies, as the effect of any intervention is zero, when there was little to intervene. Thinbk of how much effect an aspirin has when there is no pain.
If you want to stick with a fixed effects model, you can simply reduce the time factor by doing T2 - T1 and interpret the result as net effect of the intervention.
You appear to have no idea of what you are doing. I recommend you discuss this with a consultant if this is a real piece of research. If it isn't what is it? David Booth