Dear Statalists,

I would like to fill missing observations in a panel. I found the mipolate command, that seems adequate. The issue now is that I would like to compute the average for missing values between two observed ones, and carry values backwards/forwards if one of those end/start points is missing. The mipolate command seems to do this, but only for the mid point in the interval it takes the mean. It replaces all other values by cascading forwards/backwards.
So in the extract bellow, flow_ex in row 16. 17. and 18. should be =50.976. Is there a way to adjust the command?

Thanks for your feedback,
Yannik


Code:
by census: mipolate flow_combined1 year, generate(flow_extra2) nearest
list year census flow_combined1 flow_extra2 if census==11 

                    
        +--------------------------------------+
        | year   census   flow_c~1   flow_ex~2 |
        |--------------------------------------|
    11. | 1979       11          0           0 |
    12. | 1978       11         10          10 |
    13. | 1977       11          0           0 |
    14. | 1976       11         10          10 |
    15. | 1975       11          0           0 |
        |--------------------------------------|
    16. | 1974       11          .           0 |
    17. | 1973       11          .   50.976059 |
    18. | 1972       11          .   101.95212 |
    19. | 1971       11   101.9521   101.95212 |
    20. | 1970       11     70.672   70.671997 |
        +--------------------------------------+