person-period dataset

[Man Yang]

Hi all, I am trying to convert a dataset to person-period format, which means each subject has one row of data per time period. First of all, I am not sure whether the current data format satisfy the requirements of being called a person-period. Below is part of what the data looks like:

. list UCIn age CENSOR EVENT PERIOD in 1/40

     +--------------------------------------+
     | UCIn   age   CENSOR   EVENT   PERIOD |
     |--------------------------------------|
  1. |   10    17        1       0        2 |
  2. |   10    18        1       0        3 |
  3. |   10    20        1       0        5 |
  4. |   10    22        1       0        7 |
  5. |   10    23        1       0        8 |
     |--------------------------------------|
  6. |   15    16        1       .        1 |
  7. |   15    17        1       0        2 |
  8. |   15    18        1       0        3 |
  9. |   15    19        1       0        4 |
 10. |   15    20        1       0        5 |
     |--------------------------------------|
 11. |   15    21        1       0        6 |
 12. |   15    22        1       .        7 |
 13. |   15    23        1       .        8 |
 14. |   21    17        1       .        2 |
 15. |   21    18        1       0        3 |
     |--------------------------------------|
 16. |   21    19        1       0        4 |
 17. |   21    20        1       0        5 |
 18. |   21    21        1       0        6 |
 19. |   21    22        1       0        7 |
 20. |   21    23        1       0        8 |
     |--------------------------------------|
 21. |   21    25        1       0       10 |
 22. |   34    16        1       .        1 |
 23. |   34    19        1       0        4 |
 24. |   34    20        1       0        5 |
 25. |   34    21        1       0        6 |
     |--------------------------------------|
 26. |   34    23        1       0        8 |
 27. |   34    24        1       0        9 |
 28. |   36    16        0       .        1 |
 29. |   36    17        0       .        2 |
 30. |   36    19        0       1        4 |
     |--------------------------------------|
 31. |   36    20        0       1        5 |
 32. |   36    21        0       1        6 |
 33. |   36    22        0       1        7 |
 34. |   36    23        0       1        8 |
 35. |   39    17        1       .        2 |
     |--------------------------------------|
 36. |   39    18        1       0        3 |
 37. |   39    19        1       0        4 |
 38. |   39    22        1       .        7 |
 39. |   39    23        1       .        8 |
 40. |   41    17        1       .        2 |
     +--------------------------------------+

I think the part that I am not so sure about is whether the remaining data should still be kept in the dataset once a subject has the event occur. For example, at age 22 the event for the first time changed from 0 to 1 for a participant and does the information for this participant after age 22 is still needed? In my current dataset, as you can see, participant's information was kept until a certain age regardless of whether they've had the event occurred. Is this an issue for person-period dataset? Also, should I drop the missing values under the EVENT variable and only keep the cases of 0 or 1? Thanks!

[Phil Bromiley]

You are more likely to obtain a helpful answer if you follow the FAQ on asking questions - provide Stata code in code delimiters, Stata output, and sample data using dateex. Also, try to simplify your presentation to the core issues.

You should xtset your data first - xtset UCIn age if you want to use age as the date variable. This looks like a panel (time series cross sectional) data set.

You'll have to tell us a lot more about what you're doing. What question are you trying to answer? What is your dv? What kind of model do you plan to run? How are you using censor and event? I'm not sure what the missing data mean as well.