Mixed-effects negative binomial regression for a 3-wave panel data. Baseline sample size is 1144. After running menbreg and controlling for waves, the observation # is 1978,, and groups # 1186, larger than the baseline sample size. How is this possible?

Thanks for help!