Hi statalisters ... I need some help to interpret the coefficients of some mixed-effects models in my master thesis. I'll explain shortly. I performed three experimental treatments (G, C, CG). My dataset is a balanced panel in which 30 subjects per treatment are observed for 7 periods. I implemented a mixed-effects model like the following:

y = C + CG + period + C # period + CG # period + L1.M

where C and G are the dummy variables that identify my treatments. (I omitted G because it is my baseline treatment), period is my trend, C # period and CG # period are interaction terms to evaluate differences in trends across treatments, L1.M is the lag of another variable of interests. The most important treatment comparison for me is between CG and G so I would be interested in the CG coefficient. The code entered in stata is as follows:

xtmixed ce ib4.treatment treatment ## c.period L1.mpcr || group:

Now, my first model is

y = C + CG + L1.M

the code being used is this >> xtmixed ce ib4.treatment L1.mpcr || group:

here the CG cofficient is positive and highly significant. To evaluate differences in trends across treatments I add period and the two interaction terms thus running the complete model above mentioned. Here CG coefficient is no more significant but the two interaction terms are significant. What does it means? Maybe treatment CG is not statistically different from G? But interaction terms seems to reveal differences between trends in CG and G... I appreciate any useful clarification. I attached a word file in which I paste my stata output.

Thank you in advance!!! Array