I'm having a good deal of trouble navigating the data dictionary for variables in the U.S. Census Survey of Income and Program Participation (SIPP). Here's a specific puzzle I've been unable to solve:
I'm looking for data on an individual's receipt of unemployment benefits. However, there are at least two different variables with the same label in the core wave files:
euectyp5 byte %15.0g euectypm GI: Receipt of State unemployment comp.
er05 byte %15.0g er05l GI: Receipt of State Unemployment Comp.
And the variables do not have equivalent values:
. ta euectyp5 er05, mi
GI: Receipt of |
State | GI: Receipt of State
unemployment | Unemployment Comp. (ISS
comp. (ISS | Not in Un Yes No | Total
----------------+---------------------------------+----------
Not in Universe | 416,059 0 0 | 416,059
Yes | 0 3,746 1,790 | 5,536
No | 316 0 0 | 316
----------------+---------------------------------+----------
Total | 416,375 3,746 1,790 | 421,911
(This is all from 2008 Panel, Wave 1, but appears to be true in other waves as well. Have not checked other panels.)
The only thing I can think to do is to cross-reference with the questionnaire to determine what each question was actually asking, but the questionnaire itself (as downloaded here) does not actually have these variable names in the file. I started trying to read the 400-page questionnaire to deduce where each were coming from, but this was unsuccessful, not to mention unbearably tedious because the questions are full of stuff like [fill OTHSRCE_ARR(<33>):l](33) and other such macros that it is hard to track down the meaning of.
What am I missing? What is the difference between these variables: and, more generally, how do I go about figuring out other such puzzles when the variable names do not actually match those in the questionnaire and the questionnaire is so hard to read?
UPDATE:
I also wrote the census, as Albert recommended. they wrote back immediately and very helpfully. Here is the complete response:
The structure of the EUECTYP5 and ER05 variables can be confusing.
EUECTYP5 covers the entire reference period. So, respondents who have a 1 in this field have received state unemployment compensation at some point during the four-month reference period.
You can see this in the question for EUECTYP5 on page 7-189, from the data dictionary, SIPP 2008 Panel Waves 01-10 - Core Data Dictionary, available here: http://www.census.gov/programs-surveys/sipp/tech-documentation/data-dictionaries/data-dictionaries-2008.html Did ... receive any state unemployment compensation during the reference period?
ER05, however, is a monthly variable. So, it records receipt or non-receipt in each month. It appears on page 7-229 of that same data dictionary. Did ... receive income from State unemployment compensation in this month.
Usually, you could also see this in the universe statements, but unfortunately, there is an error in the EUECTYP5 universe statement posted in the data dictionary. It should read All persons 15+ at the end of the reference period.
The ER05 universe, however, is correct as posted: All persons 15+ at the end of the reference period indicating receipt of State unemployment compensation sometime during the reference period. (EUECTYP5 = 1)
So, the ER05 recipients are a subset of EUECTYP5 recipients. Respondents who received unemployment compensation in all 4 months of the reference period will have a "1" for both ER05 and EUECTYP5 in all 4 months of a wave. But, respondents who only received UC in one month will have ER05 and EUECTYP5 = 1 in that month, and will have EUECTYP5=1 and ER05=2 in the other 3 months of the wave.
As far as trying to figure out variable names from question fields, it really depends on the variable you are looking at and the section in which it appears. For some fields there is a pretty easy one-to-one correspondence between the variable and the relevant question. For others, however, it isn't as clear what the connection is.
I suggest looking at the user's guide, data dictionary, and user's notes, as you have been doing, and then writing in with questions. We're happy to help in whatever way we can.
