IDs split in two columns

[Isaac Hance]

I am using a dataset that, for some reason, has a column called id_b, and another called id_c. Every row with a value in id_b has nothing in id_c, and vice versa. I simply want to combine these columns into a column named id.

I have attempted to use egen id = rowmax (id_b id_c), but whilst the id_b and id_c columns are both long data, this produces a column which is a float, and when I convert this to long, these are rounding issues as the IDs are 8 or 9 numbers long.

[Rich Goldstein]

-egen- allows you to specify the type you want the new variable to be so you could just insert "long" or "double" after -egen- in your command; or if you wanted to not make a new variable you could use -replace- to

replace id_b=id_c if missing(id_b)

and there are other options; but be sure to read

h egen

note that since you did not supply a data example using -dataex-, there might be errors or typos above

[Devra Golbe]

It's a good idea to hold ID variables as string rather than numeric. See

help tostring

(That brings up the help entry for destring.)

[Isaac Hance]

-egen- allows you to specify the type you want the new variable to be so you could just insert "long" or "double" after -egen- in your command; or if you wanted to not make a new variable you could use -replace- to

replace id_b=id_c if missing(id_b)

and there are other options; but be sure to read

h egen

note that since you did not supply a data example using -dataex-, there might be errors or typos above

Thank you Rich - simply adding long to the command did exactly what I needed!