Translating run-together dates

[Abdan Syakura]

Dear all,

Sorry for basic question. I would like to translate the following date into Stata date format.

* Example generated by -dataex-. To install: ssc install dataex
clear
input str6 dscd str8 date
"ID:AAL" "10101990"
"ID:AAL" "10111990"
"ID:AAL" "1011990" 
"ID:AAL" "10121990"
"ID:AAL" "10151990"
"ID:AAL" "10161990"
"ID:AAL" "10171990"
"ID:AAL" "10181990"
"ID:AAL" "10191990"
"ID:AAL" "1021990" 
"ID:AAL" "10221990"
"ID:AAL" "10231990"
"ID:AAL" "10241990"
"ID:AAL" "1031990" 
"ID:AAL" "1041990" 
"ID:AAL" "1051990" 
"ID:AAL" "1081990" 
"ID:AAL" "1091990" 
"ID:AAL" "1101990" 
"ID:AAL" "11101989"
"ID:AAL" "1111990" 
"ID:AAL" "11131989"
"ID:AAL" "11141989"
"ID:AAL" "11151989"
"ID:AAL" "11161989"
"ID:AAL" "11171989"
"ID:AAL" "111990"  
"ID:AAL" "11201989"
"ID:AAL" "11211989"
"ID:AAL" "1121990" 
end

I tried both:

 gen date2 = date(date, "MDY")
format date2 %td

and

numdate daily date3 = date, pattern(MDY)

But both treat 1041990 (October 4 1990) and some others as missing. Could you please help? Thank you.



Best regards,

[Øyvind Snilsberg]

How do you know whether, e.g., 1111990 is 11 january 1990 or 1 november 1990?

[Nick Cox]

You have some rogue dates in there. This may help, but inspect the protocol and results for incomplete dates very carefully.

* Example generated by -dataex-. To install: ssc install dataex
clear
input str6 dscd str8 date
"ID:AAL" "10101990"
"ID:AAL" "10111990"
"ID:AAL" "1011990"
"ID:AAL" "10121990"
"ID:AAL" "10151990"
"ID:AAL" "10161990"
"ID:AAL" "10171990"
"ID:AAL" "10181990"
"ID:AAL" "10191990"
"ID:AAL" "1021990"
"ID:AAL" "10221990"
"ID:AAL" "10231990"
"ID:AAL" "10241990"
"ID:AAL" "1031990"
"ID:AAL" "1041990"
"ID:AAL" "1051990"
"ID:AAL" "1081990"
"ID:AAL" "1091990"
"ID:AAL" "1101990"
"ID:AAL" "11101989"
"ID:AAL" "1111990"
"ID:AAL" "11131989"
"ID:AAL" "11141989"
"ID:AAL" "11151989"
"ID:AAL" "11161989"
"ID:AAL" "11171989"
"ID:AAL" "111990"  
"ID:AAL" "11201989"
"ID:AAL" "11211989"
"ID:AAL" "1121990"
end

gen length = length(date)
gen wanted = cond(length == 8, daily(date, "MDY"), cond(length == 7, daily("0" + date, "MDY"), cond(length == 6, daily("01" + date, "DMY"), .)))

sort length wanted

list date wanted length, sepby(length)

    +----------------------------+
     |     date   wanted   length |
     |----------------------------|
  1. |   111990    11262        6 |
     |----------------------------|
  2. |  1011990    10958        7 |
  3. |  1021990    10959        7 |
  4. |  1031990    10960        7 |
  5. |  1041990    10961        7 |
  6. |  1051990    10962        7 |
  7. |  1081990    10965        7 |
  8. |  1091990    10966        7 |
  9. |  1101990    10967        7 |
 10. |  1111990    10968        7 |
 11. |  1121990    10969        7 |
     |----------------------------|
 12. | 11101989    10906        8 |
 13. | 11131989    10909        8 |
 14. | 11141989    10910        8 |
 15. | 11151989    10911        8 |
 16. | 11161989    10912        8 |
 17. | 11171989    10913        8 |
 18. | 11201989    10916        8 |
 19. | 11211989    10917        8 |
 20. | 10101990    11240        8 |
 21. | 10111990    11241        8 |
 22. | 10121990    11242        8 |
 23. | 10151990    11245        8 |
 24. | 10161990    11246        8 |
 25. | 10171990    11247        8 |
 26. | 10181990    11248        8 |
 27. | 10191990    11249        8 |
 28. | 10221990    11252        8 |
 29. | 10231990    11253        8 |
 30. | 10241990    11254        8 |
     +----------------------------+

Alternatively see https://www.stata-journal.com/articl...article=dm0062 for how to impute an unstated day of the month.