ATT after CEM using model "linear-RE" -- what does p=2.000000 mean?

[tkallen10]

Using the "linear-RE" model in ATT after CEM produces confidence intervals much narrower than other methods, in my dataset (around 7 million records), but the p.value is always over 1.0, usually exactly 2.0. What does this mean? How do I fix this? Thanks, Tim