Dear Nick,
Thanks for suggesting the -stable- option. It crossed my mind after reading the -sort- thread from last month. However, the -stable- option with -sort- doesn't keep the data in serial/pernum order at the -bysort- command. Did I mess something up? Test commands and sample output below.
Thanks,
Deb
. use "C:\dlgarvey\Census data\IPUMS data\temp.dta", clear
. sort serial pernum, stable
. gen test=momloc
. bysort serial (test): replace test=8 if momloc==1
(30 real changes made)
. l serial pernum momloc test
serial pernum momloc test
----------------------------------
1. 2408215 1 0 0
2. 2408215 2 0 0
3. 2408216 2 0 0
4. 2408216 3 0 0
5. 2408216 1 0 0
----------------------------------
6. 2408217 2 0 0
7. 2408217 1 0 0
8. 2408217 5 2 2
9. 2408217 4 2 2
10. 2408217 3 2 2
----------------------------------
11. 2408218 1 0 0
12. 2408219 1 0 0
13. 2408220 1 0 0
14. 2408221 3 0 0
15. 2408221 2 0 0
----------------------------------
16. 2408221 1 0 0
17. 2408222 1 0 0
18. 2408222 2 1 8
19. 2408222 3 1 8
20. 2408233 2 0 0
----------------------------------
21. 2408233 1 0 0
22. 2408233 4 2 2
23. 2408233 3 2 2
24. 2408323 2 0 0
25. 2408323 5 0 0
----------------------------------
26. 2408323 1 0 0
27. 2408323 3 2 2
28. 2408323 4 2 2
29. 2408423 1 0 0
30. 2408423 3 0 0
----------------------------------
31. 2408423 2 1 8
32. 2408423 5 3 3
33. 2408423 4 3 3
34. 2408434 1 0 0
35. 2408434 2 0 0
--Break--
----
Date: Tue, 19 Jul 2005 00:13:34 +0100
From: "Nick Cox" <[email protected]>
Subject: st: RE: Bysort on non-selection variables
Two suggestions:
1. Do you also need a unique id?
gen uniqueid = _n
2. Check out the -stable- option.
Nick
[email protected]
Deborah Garvey, Ph.D.
> I have data with a household identifier (serial) and a person
> locator within each household (pernum).
>
> Is there any way to maintain a sort order of PERNUM within
> SERIAL, i.e., the results of a command like -sort serial
> pernum- for a -bysort- that is sorted on a within-household
> indicator variable, i.e. -bysort serial (indicatorvar):
> several expressions to condition the value of a PERNUM-level variable-
>
> My question may seem odd, but the person's location in the
> household is an important criterion for determining the
> indicator variable's value. Unfortunately, the value of the
> indicator variable may be the same for more than one person
> in the household, so I can't use -bysort serial pernum
> (indicatorvar): stuff-
******************************
Deborah Garvey, Ph.D.
Department of Economics
Kenna Hall
Santa Clara University
Santa Clara, CA 95053
408/554-5580
408/554-2331 (FAX)
[email protected]
http://lsb.scu.edu/~dgarvey
**********************************
This message scanned for viruses and SPAM at SCU (MGW2)
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
