Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | David Kantor <kantor.d@att.net> |
To | statalist@hsphsun2.harvard.edu |
Subject | Re: st: basic question |
Date | Mon, 22 Aug 2011 22:32:35 -0400 |
At 09:24 PM 8/22/2011, you wrote:
[...] My variables meas: v9532: income from main job v9982: income from secondary job v1022: third or more of income jobs But most of the people have only one job, so they get missing to v9982 and v1022...
In that case, the missing values ought to be recoded to 0. Thus, if I don't have a second job, then the income from my second job is 0 -- not missing. Similarly for the third.
But you may need to be more cautious. Do these variables have either... 1, a special value that signifies "unknown", or2, an additional variable that tells you that the value for v9982 is unknown? (Similarly for v1022.) If so, then you need to code the income variables to reflect these conditions -- to distinguish "no second job" from "has a second job, but income is unknown". The former should correspond to 0; the latter to missing, in which case the expression v9532 + v9902 + v1022 will (correctly) yield missing.
Other respondents mentioned egen... rowtotal, which will treat missings as zero. This will work with the values as given, but it may be best to code your values as I have indicated -- 0 rather than missing, where appropriate -- and use +. Note that, given the proper coding of the variables, egen...rowtotal will give you something else: the sum of known values.
HTH --David P.S., as a minor point, I suspect that you mean v10022 where you wrote v1022. * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/