Panel Data

[Wayne Robinson]

Hi, I am working with panel data and want to create a dummy variable based on the median of each cross section for the entire sample and also at each time period. How do I do that?

[Fei Wang]

Below is an example. Not sure if what you were asking. For example, the value of wanted in 2011 is 2 which is the median of 2, 1, 2, 0, 4 (the 2011 values of id 1-5).

* Example generated by -dataex-. For more info, type help dataex
clear
input float(id year v)
1 2011 2
1 2012 1
1 2013 4
2 2011 1
2 2012 5
2 2013 1
3 2011 2
3 2012 2
3 2013 3
4 2011 0
4 2012 2
4 2013 0
5 2011 4
5 2012 4
5 2013 3
end

. bys year: egen wanted = median(v)

. sort id year

. list

     +------------------------+
     | id   year   v   wanted |
     |------------------------|
  1. |  1   2011   2        2 |
  2. |  1   2012   1        2 |
  3. |  1   2013   4        3 |
  4. |  2   2011   1        2 |
  5. |  2   2012   5        2 |
     |------------------------|
  6. |  2   2013   1        3 |
  7. |  3   2011   2        2 |
  8. |  3   2012   2        2 |
  9. |  3   2013   3        3 |
 10. |  4   2011   0        2 |
     |------------------------|
 11. |  4   2012   2        2 |
 12. |  4   2013   0        3 |
 13. |  5   2011   4        2 |
 14. |  5   2012   4        2 |
 15. |  5   2013   3        3 |
     +------------------------+

[Wayne Robinson]

Thanks. This is helpful