COR583a
February 23, 1990

Construction of Sample of 105 Pairs of 1975 WLS Respondents 
Who Were Married to One Another at the Time of the 1975 Survey

Many WLS respondents married other WLS respondents, and in about
105 cases/pairs, both spouses were interviewed in 1975.  Data for
these pairs of respondents are unique and important: Among other
things, they permit analyses of the reliability of respondents'
reports of their spouses' and their spouses' families'
characteristics.

The sample was selected from data in a file which contains the
latest locating information for 1975 and 1977 respondents, as of
1975 or 1977.  The data were collected in the course of tracing
and interviewing respondents and their siblings in 1975 and 1977. 

Initially, the following variables for all cases were extracted:
IDSWL, Phone Number, Address, and Zip Code.  

Next, the following variables were extracted from the master data
file and were matched to the locator file extract: Marital Status
and 1975 Response Status.  With this information, the locator
file extract was limited to cases in which the individual
responded to the 1975 survey and was currently married at the
time of that survey.

The sample, now limited to married 1975 respondents, was then
split in two: Male respondents were put in one file, and female
respondents were put in another file.  Then, records for male and
female respondents were matched by phone number.  That is,
married men were matched to married women if their phone numbers
were identical.  This technique yielded 111 pairs of respondents. 
Finally, men's and women's zip codes and addresses were examined. 

In six cases either the address or zip code were very different,
and these records were dropped from the file.  These may have
been cases in which one respondent did not have a telephone, and
listed a friend or relative's phone number instead.  Further
examination of the original 1975 questionnaires for several dozen
cases gave further support for the assertion that these pairs
were married: in the cases which were examined, each respondent
listed the same number of children and listed very similar names
and birthdates for those children (the names sometimes varied
slightly; for example, the wife may have listed "Robert" while
the husband listed "Bob.")