Generating data for one variable based on two or more variables that have data.

[Patrick Kafeero]

Hello,
I'm kindly requesting for guidance on this.
I am trying to create a variable age whose information is based on four variables.
In this case, if an observation/id has a value for one variable, it should have data for the other 3.
An example of the code is below.

* Example generated by -dataex-. For more info, type help dataex
clear
input float(ptid age_m age_f age_j age_p)
1 23  .  .  .
2  . 44  .  .
3  . 16  .  .
4 45  .  .  .
5  . 51  .  .
6 65  .  .  .
7  .  . 27  .
8  .  .  . 36
end

I'm not sure if the explanation is helpful but I want to create a variable age with just one value based on the variables age_m age_f age_j and age_p

I'll be grateful for any help.

Thanks.

[Carlo Lazzaro]

Patrick:
do you mean something along the following lines?

. egen wanted=rowtotal(age_*)

. list

     +-----------------------------------------------+
     | ptid   age_m   age_f   age_j   age_p   wanted |
     |-----------------------------------------------|
  1. |    1      23       .       .       .       23 |
  2. |    2       .      44       .       .       44 |
  3. |    3       .      16       .       .       16 |
  4. |    4      45       .       .       .       45 |
  5. |    5       .      51       .       .       51 |
     |-----------------------------------------------|
  6. |    6      65       .       .       .       65 |
  7. |    7       .       .      27       .       27 |
  8. |    8       .       .       .      36       36 |
     +-----------------------------------------------+

.

[Patrick Kafeero]

This is very helpful Carlo. Thanks a lot.

Would you consider a similar approach if you were dealing with dates or strings?

Thanks,
Paddy

[Carlo Lazzaro]

Patrick:
as far as I know, that -egen-function does not work with -string- variables.
As far as dates are concerned I've neither used -rowtotal()- with dates, nor I can figure out an instance in which that approach is useful.