- APPENDIX D -
nonresponse ('bias') and
field administration files
These files are available only by special permission from
the NES Board of Overseers. A written request must be submitted
to the National Election Studies at the address provided on
p.iii.
1986 Nonresponse file.
This rectangular datafile contains 3833 cases from 2176
complete or partial interviews, 824 refusals, 215 no-
contacts, and 618 non-sample cases.
Vars included are the following:
1-61 cover sheet (control file) information
62-75 sampling information
76-86 interviewer information
87-109 non-interview information
110-170 Census segment profile information
----------------------------------------
1988 Nonresponse file.
This rectangular datafile contains 3503 cases from 2040
complete or partial interviews, 643 refusals, 210 no-
contacts, and 610 non-sample cass.
Variables include those from the Interviewer
Description booklet, a form used for the first time in
1988 (completed by interviewers) to describe attempts
at contact with each household--and, subsequently, with
each designated R-- within the 1988 NES sample.
Vars included are the following:
1-60 pre cover sheet (control file)
70-76 interviewer
77-90 sampling
201-263 Interviewer Description Booklet
5000+ post cover sheet (control file)
With special permission, a collection of 105 Census
variables describing demographics at the segment level
may also be obtained. Contact the NES staff for more
information.
----------------------------------------
1990 Nonresponse file.
This structured datafile is more detailed than previous
files. Of the 3420* lines of sample that were originally
issued for the 1990 election study, 1980* sample lines
became completed interviews, others were non-sample, and
still others were designated non-interviews. For each line
of sample, in records called 'sample unit' records, this
datafile provides sampling information, disposition of the
case, control record variables, and data about the
interviewer(s). Census tract information for each sample
address is available by contacting NES staff.
Almost every 'sample unit' record is also associated with
another (or more) records(s): a record for each personal
call to the sample address by a field interviewer, the
'call' record. There are 15,311* call records in this file,
each containing variables for the day of the week, time of
call, and call disposition. 6,310* of the 15,311* personal
calls made by interviewers resulted in contacts with someone
in the household, and, if contact was made during a call, an
additional record containing variables describing the nature
of the contact is present.
The structure of the datafile contains 2 levels and 3
groups. The first level (also identified as group 1) is the
'sample unit' level, with 3420 cases. The second level
(consisting of groups 2 and 3) is the 'call' level, where it
is possible to have more than one record per sample unit.
The 1990 Post ID is the link variable which links call
number from level 2 to its corresponding sample unit on
level 1. To link a contact record (level 3) with its
corresponding call (level 2), the Post ID in conjunction
with the call number is used.
To illustrate the file structure, a portion from this
datafile might be as follows:
GROUP CASE ID CALL # DATA
1 0001 'sample unit' data (sampling,
inteviewer, etc.)
2 0001 01 call data for personal
call number 1
2 0001 02 call data for personal
call number 2
2 0001 03 call data for personal
call number 3
2 0001 04 call data for personal
call number 4
3 0001 04 -contact data for
personal call number 4
2 0001 05 call data for personal
call number 5
3 0001 05 -contact data for
personal call number 5
1 0002 'sample unit' data (sampling,
interviewer, etc.)
2 0002 01 call data for personal
call number 1
3 0002 01 -contact data for
personal call number 1
1 0003 'sample unit' data (sampling,
interviewer, etc.)
2 0003 01 call data for personal
call number 1
2 0003 02 call data for personal
call number 2
2 0003 03 call data for personal
call number 3
3 0003 03 -contact data for
personal call number 3
* Note that these numbers reflect a correction to the
1990 Bias File resulting from correction of the 1990
PrePost data file; 20 cases were deleted from the 1990
Study data which were discovered to be either: 1)
fabricated by an interviewer [16 cases] or 2)
interviews of ineligible respondents [4 cases].
Removal of these 20 interview cases from the 1990 Bias
file resulted in a reduced number of call records and
contact records as well as fewer records at the sample
unit level.
----------------------------------------
1992 Nonresponse file.
This rectangular datafile contains 3690 cases from 2485
complete or partial ("short form") interviews, 497
refusals, 64 no-contacts, and 431 non-sample cass.
Variables include those from the Pre and Post
coversheets, sampling vars, administration and
interviewer vars.
Vars included are the following:
1-69 pre administration
70-86 sampling
87-99 pre interviewer
100-141 post administration
142-154 post interviewer
155-160 associated IW vars
----------------------------------------