遇到了同样问题,
看到了这个帖子
https://www.stata.com/statalist/archive/2013-04/msg00833.html
“The "o." prefix is used by Stata for a variable that's been omitted because of collinearity or whatever. If you reestimate but omit the dropped variable by hand, xtoverid should then work.”
去掉omitted dummy,果然work了