NSFH2 DATA FILES
Caseids are in columns 1-6 for NSFH2 data. Columns 1-5
contain the numeric caseid which is used to link all data files for one
case. Column 6 contains an alpha character which designates the type
of wave 2 interview:
R = primary respondent
S = spouse/partner from wave 1
N = new spouse partner that was not present at wave 1
P = parent interview
O = older focal child interview (age 18-23)
Y = younger focal child interview (age 10-17)
B = best measures data, which compares primary R and spouse/partner
income reports
The NSFH2 data files that are available include:
1. Main respondent - interview data
2. Current spouse/partner - interview data
3. Ex-spouse/partner - interview data
These files contain data collected by the interviewer in CAPI
(Computer-Assisted Personal Interview). The structure of these files
is identical; the variable locations in the main respondent codebook
apply to all three of these files. There are a few variables that
are applicable only to certain types of respondents - the codebook
and/or the instrument indicate when a variable is inapplicable to
certain respondent types. The inapplicable fields contain 9s.
The main respondent file includes one record for each of 10,007 main
respondents interviewed. The current spouse/partner file includes a
record for each current spouse/partner who was interviewed. There
are some cases where there was an interview with the spouse/partner,
but no interview with the main respondent.
Cohabitation is difficult to define precisely. For purposes of the
logic of the NSFH2 interview, a respondent was defined as cohabiting
if he or she had an opposite sex person in the household who was
identified as "lover/partner" on the relationship variable. Because
of differing definitions of the situation the partners, changes in
living arrangements between interviews of partners, and, for other
reasons, there are a small number of cases where one partner is
classified as cohabiting and the other is not.
There are a small number of cases where we are uncertain whether a
spouse interview was completed with a new spouse/partner or with
the NSFH1 spouse/partner. There are also a small number of cases
where one partner reports that they are married and the other
reports they are cohabiting. There are also a small number of cases
where one partner reports that they are cohabiting and the other
reports that they are roommates, or that the "partner" is not in the
household.
This will have an impact almost entirely on spouse records and not
on the main respondent record.
4. Main respondent - SE data
5. Current spouse/partner - SE data
6. Ex-spouse/partner - SE data
These files include data collected in the self-administered
questionnaires. The structure of each of these files is identical
for the three respondent types.
There are a few cases where there are interview data, but no SE data.
These are cases where the SE form was not completed by the respondent,
not returned by the interviewer, or lost before data entry.
When there are interview data but no SE data, there is an SE record
which is completely filled with 9s.
7. Constructed variables - main respondent
8. Constructed variables - current spouse/partner
9. Constructed variables - ex-spouse partner
10. Constructed best measures - comparison of main respondent
income reports and current spouse/partner income reports
These files contain variables constructed from the interview data, for
example, income measures aggregated over household members or
over income types, union history, education completed, etc. Case
weights are in this file.
The structure of the first 3 files is identical. The best measures
file was created by comparing the primary respondent and current
spouse/partner income reports.
11. Status file
This file shows which interview type is available for which cases, and
it shows the result code for cases that were not interviewed. There is
one case for each NSFH1 respondent. Among other uses, this file
permits the analysis of response rates.
12. Focal child telephone interview data - 10-17 year olds
13. Focal child telephone interview data - 18-23 year olds
Different interviews were administered to focal children depending on their
age. (Ages were classified by the age reported by the parent (usually R1)
at the time of the parent interview. There are separate data files for the two
age groups. While some questions are identical in the two interviews, they
have different item names and locations
14. Parent telephone interview data
An interview was conducted with a parent of the main respondent. A parent
was selected randomly for interview when both parents were living. Under
some circumstances the other parent could be substituted for the selected
parent.
15. Proxy data - telephone interviews with proxy respondent when
respondent was deceased or too ill to interview
Number of Records/size of Files:
Number of Record Bytes
Records Length (*1000)
1. Main R - interview data 10005 8231 82,350
2. Curr spouse/partner- int data 5624 8231 46,300
3. Ex-spouse/partner - int data 789 8231 6,500
4. Main respondent - SE data 10005 728 7,300
5. Current spouse/partner - SE data 5624 728 4,100
6. Ex-spouse/partner - SE data 789 728 575
7. Main R - constructed variables 10005 1999 20,000
8. Curr spouse/partner - const vars 5624 1999 11,240
9. Ex-spouse/partner - const vars 789 1999 1,580
10. Constructed best measures vars 10211 1553 15,870
11. Status file 13007 47 612
12. Focal child data - age 10-17 1415 1844 2,620
13. Focal child data - age 18-23 1090 1844 2,020
14. Parent interview data 3347 2333 7,840
15. Proxy interview data 802 157 130
All data files are fixed length records in ascii format.
File Naming Conventions:
First letter: type of file
I = Instrument
C = Codebook
D = Data
L = Layout file: lists variable names and column locations
2nd letter/number: type of respondent
M = Main Interview Respondents - Primary respondents and spouses/
partners; caseids end in R, S and N.
S = Spouses/Partners; caseids end in S and N
R = Primary Respondents only; caseids end in R.
C = Current spouses; caseids end in S and N.
X = Ex-spouses; caseids end in S.
P = Parent; caseids end in P
K = Kid; caseids end in Y or O
Z = Proxy; caseids end in R
SURVEY INSTRUMENTS
In the traditional "paper and pencil" interview there is a printed
interview that the interviewer follows, which (hopefully) contains
the skip logic of the interview in clear and unambiguous language.
In addition to being a roadmap for the interviewer, it is also a
very useful document for researchers using the data. In a
computer-assisted interview (CATI or CAPI) there is not a similar
document that is comprehensible to persons not familiar with the
programming language.
We have produced an interview "instrument" which includes the skip logic,
interviewer instructions, and other information that should allow
researchers to understand precisely what "stimulus" respondents were given
in each question.
CODEBOOKS
Main Interview Materials:
There are three separate data files for three different respondent
types:
1. Main respondents (the original NSFH1 primary respondent,
also referred to as R1)
2. Current spouse/partner of R1 (some of whom were the
spouse/partner at NSFH1 (R2) and others are newly acquired
spouse/partners (New Spouse/Partners)
3. Ex-spouse/partners (NSFH1 Spouse/Partners (R2's) who are
not Current Spouse/Partners.
The layout (variable locations) for the records in each of
these files is identical. The variable names for items on each of
these files are the same except for the prefix letter (M for Main
Respondents, C (current) or X (ex) for spouse/partners).
The variable locations, coding categories, and (with the exception
of the first character) variable names are identical. There are two main
interview codebooks: one contains frequencies for primary respondents and
the other has frequencies for spouses/partners (one column for current, one
column for ex-sp/p).
SE and Constructed Variables:
Again, variable locations for the SE information are the same for all three
respondent types. There are two codebooks: one containing primary respondent
frequencies, the other for spouse/partners.
The variable locations for the constructed variables are also the same for all
three respondent types, but the codebook does not contain frequencies for any
respondents. There is only one version of the codebook which describes the
column location and defines each variable. The best measures are a comparison
of the primary respondent and spouse/partner's income reports. The data and
layout are separate files from the other constructed variables, but they
are in the same codebook as the other constructed variables.
Other Codebook Information:
The codebook includes global skips, but not all question-and response-
specific skips. For example, if left-handed respondents skipped a section
of the interview, this will be noted in the codebook. But if persons
answering yes to a question skip the following questions, this is not shown
in the codebook. Or if a question about marital status is not asked
regarding household members age 15 and younger, this will not appear in the
codebook. All of this skip logic is in the instruments and skip maps.
The codebook also does not include interviewer instructions, and may not
include all question wording variation (fills).
Zero frequency variables in main interview codebooks:
The codebook for the main interview is designed to be used with
current and ex-spouse records, as well as with main respondent
records. The frequencies are for main respondents only. A few variables
shown in the codebook are applicable only to current or ex-spouses.
These are shown in the codebook with "0" frequencies. (An example is
variable ML92, which is applicable only to spouses. In this instance
the question wording had to be different for spouses.)
Codebook Examples:
This is the codebook entry for two variables in the care-giving and
receiving section of the interview.
CKPT:ME3
. .
a. <l1a> If R has any conditions on card 5, GO TO ME3 <C3a>
b. <l2a> If R does not have any long-term physical or mental
condition or disability, GO TO CKPT:ME5 <l3>
. .
1270-1272 ME3 (C3a)
What is the most serious condition that you have?
See Appendix A for codes
@ Unwt Wgted
Freq Pct
001-989 Condition given 4085 40.67
990 No specific condition given 88 0.89
997 Refused 2 0.01
998 Don't know 19 0.15
999 Inap/No Answer 5814 58.29
Total 10008 100.00
In this example:
CKPT:ME3
a. <l1a> If R has any conditions on card 5, GO TO ME3 <C3a>
b. <l2a> If R does not have any long-term physical or mental
condition or disability, GO TO CKPT:ME5 <l3>
This is a checkpoint. It is a skip condition (that can be skipped to from a
point earlier in the interview). Checkpoints are identified by the CKPT
designation followed by item name of the variable immediately following (ME3)
1270-1272 ME3 (C3a)
"1270-1272" is the column location of the item.
"ME3" is the item name. The convention for naming items is discussed below.
"(C3a)" is the original item name in the CAPI program. This is included for
NSFH staff use; it will not be useful to users of the data.
"What is the most serious condition that you have?"
This is the text of the question. Interviewer instructions and other
clarifying information are not included in the codebook. They are included in
the instrument.
"See Appendix A for codes"
This is clarifying information that will be useful to the user of the data.
In this instance it tells us that the codes are not here in the codebook but
are in Appendix A. When there are numerous coding categories for a variable
- e.g., occupation or religious preference, the coding categories are
contained in appendices to the codebook.
@ Unwt Wgted
Freq Pct
001-989 Condition given 4085 40.67
990 No specific condition given 88 0.89
997 Refused 2 0.01
998 Don't know 19 0.15
999 Inap/No Answer 5814 58.29
Total 10008 100.00
This tells us the codes for the variable and the frequency distribution. For
main respondent data, both unweighted frequency distributions and a weighted
percentage distributions are shown. For some data only the unweighted
frequency distributions are shown.
In this example, since coding categories are not shown (they are in an
appendix), the valid substantive codes are all combined (001-989). In such
cases frequencies are always shown for non-substantive codes (Refused, DK).
The 5814 cases coded 999 are inapplicable/no answer. In this example the
series of questions is asked of all respondents, but the particular
question is asked only of persons who reported on either of the two
previous questions that they had a disability.
We do not show response-specific skips in the codebook - i.e., in the
example that follows, only respondents answering "Yes" (1) are asked
the immediately following question about the care-giver; Others ["No" (2),
"Refused" (7) or "Don't Know" (8)] skip to a later question.
1362 ME52 (C63)
At any time in the last 12 months, did you RECEIVE help
with personal care from anyone WHO WAS LIVING WITH YOU,
because of a health condition, illness, or disability?
@ Unwt Wgted
Freq Pct
1 Yes 667 6.75
2 No 9337 93.23
7 Refused 0 0.00
8 Don't know 1 0.00
9 Inap/No Answer 3 0.02
Total 10008 100.00
Aggregated variables
Near the beginning of the interview there is a series of questions
about each member of the respondent's household. Included in this
series is a question on marital status asked with reference to all
household members (excluding the respondent) age 17 and older.
(Other examples where this applies include questions relating to
spells of employment and questions relating to the number of children
moving in and out of the parental household.)
The codebook entry for marital status of household members is:
0000 MB6P01-MB6P16
What is (his/her) marital status? Is (he/she):
@ Unwt Wgted
Freq Pct
1 married, 6370 4.58
2 separated because of marital problems, 190 0.11
3 divorced, 614 0.36
4 widowed, or 342 0.24
5 never married? 2963 2.01
7 Refused 10 0.01
8 Don't know 16 0.01
9 Inap/No Answer 149623 92.67
Total 160128 100.00
The item names for this variable are MB6P01-MB6P16. Space is
available for up to 16 household members. Hence there are 160128
possible responses to this question (10008 respondents * 16
potential persons per household). Of these, 149623 are inapplicable
- the person does not exist (household has less than 16 members) or
the person is less than age 17. The marital status distribution of
the 10,505 household members who are age 17 and older is shown in
the remaining rows of the table.
No column location is shown in this codebook entry. The "0000" is
simply a flag showing that this is an "aggregated variable". Column
locations for aggregate variables are normally shown in tables following
the item or series of items. In this example:
Household roster other than respondent--
Column locations
Person sex age marital cohab relation-
Number status status ship to R
________________________________________________________________________
1. 0024-0025 0026 0027-0028 0029 0030 0031-0032
2. 0048-0049 0050 0051-0052 0053 0054 0055-0056
3. 0072-0073 0074 0075-0076 0077 0078 0079-0080
4. 0096-0097 0098 0099-0100 0101 0102 0103-0104
.
.
.
16. 0384-0385 0386 0387-0388 0389 0390 0391-0392
This shows that marital statuses for the other members of the household
are found in columns 29, 53, 77, 101, . . . and 389.
Note also in this example, the codebook shows that the computer varied the
question wording depending on the sex of the person being talked about
(his/her) and (he/she).
SKIP MAP
We have created a "skip map" which shows, in graphic form, the skip
logic of the entire interview. The skip map is extremely useful in
understanding which respondents get asked which questions. Our
experience with Nsfhhelp has shown that most of the questions asking
why N's don't seem to match can be quickly answered from the skip map.
Users are encouraged to use the skip map along with the codebook and
interview schedule to understand the logic of the interview. These
skip maps are the *.pdf files in the NSFH public access directory.
APPENDICES
A Medical Condition Codes
B Religion Codes
C Reasons Stopped Living with Parents Codes
D Reasons Started Living with Parents Codes
E States Included in Each Census Region
F Occupation Codes
G Occupational Socioeconomic Status Codes:
Total-based and Male-based
H Conversion of Dates to Century Months
I Industry Codes
J Categorical & Constructed Income Variables
K Cause of Death Codes
L Relationship Codes for Proxy Interview
M Showcards used with Main Interview
N Field Report (includes interviewing procedures)
O Weight Report
P SE Measures
Q Person Numbers
VARIABLE NAMING CONVENTIONS
All item names are 8 characters or less; all begin with an alpha
character. Thus they are legal SPSS variable names.
Main Interview
(1) First character (letter)
M Primary (Main) Respondent interview item
C Current spouses/partners
X Ex spouses/partners
(2) Second character (letter) - refers to the module in which the item occurs
Interview Modules (in order they appear throughout the interview):
A Birthdate, is time 1 spouse/partner still with R?
Sex, age and householder status
B Current marital status and household composition
C R's children living elsewhere and spouse/partner's children
living elsewhere
R's children mentioned at time 1 but not time 2
D Comings and goings of children between time 1 and time 2
E Health and giving/receiving care
F R's parents
R living with parents
G Help given to and received from adult children 19+
H Siblings, grandchildren, background of new spouses
I Marriage/cohabitation history since time 1
J Fertility history since time 1
K Problem inventory - children age 5-17
L Original Focal Child, now age 5-17
K Problem inventory - children age 18-23
L Original Focal Child, now age 18-23
K Problem inventory - children age 4 and under
L New Focal Child, age 4 and under
M Child under 18 with other parent absent from the household
R's child under 18 who is absent from the household
N Residence history since time 1
Religion
Education history since time 1
O Work history since time 1
P Current employment
Q Income, assets, debts and economic transfers
R Interviewer observations, info. about Focal Child and Parent
selected for telephone interviews
S Additional variables - these variables do not appear in the
interview as questions
(3) Third character (number) - sequential within the module
For sequences in table format, this will be followed by a
letter "P" (for person) or "T" (for time) and then a number
which is sequential within the table
Self-enumerated
(1) First character (letter)
M Primary Respondent interview item
C Current spouses/partners
X Ex spouses/partners
(2) Second character is always the letter T
(3) Third character and fourth character for SE10-18 (number)
- corresponds to the SE (values=1-18)
- an exception to this is the STATUS variable at the beginning
of each SE
(4) Fourth character (or fifth character for SE10-18) - usually the item
number in the SE questionnaire
Parent Interview
(1) First character is always the letter P
(2) Second character (letter) - refers to the module in which the item occurs.
Interview Modules (in order by how they appear throughout
the interview):
A Result code of parent interview, which parent
C Household composition and children living elsewhere
D Marriage/cohabitation; Health and well-being
E Health and giving/receiving care
F Parents
G Help given to and received from adult children 19+
H Siblings, grandchildren
P Current employment
Q Income, assets, debts and economic transfers
(3) Third character (number) - sequential within the module
For sequences in table format, this will be followed by a
letter "P" (for person) or "T" (for time) and then a number
which is sequential within the table
Child Interview
(1) First character is always the letter F
(2) Second character (letter)
Y = Younger focal child interview; child age 10-17
O = Older focal child interview; child age 18-23
(3) Third character (number) - sequential throughout the interview;
numbers start again at 1 for beginning of older child interview.
For sequences in table format, this will be followed by a
letter "P" (for person) or "T" (for time) and then a number
which is sequential within the table
VI.D PROXY (Deceased or ill primary respondents)
(1) First two characters are always the letters ZZ
(2) Third character (number) - sequential throughout the interview
PERSON NUMBERS
NSFH1 Person Numbers Maintained in NSFH2
In NSFH1 person numbers (also referred to as household member numbers)
were assigned to all household members and to all sons and daughters and
step-sons and daughters mentioned in the interview. These numbers
had values of 1 (main respondent), 2-15 (persons in the household)
and 21-36 (sons and daughters and step-sons and -daughters living
elsewhere). When a person who had an NSFH1 person number was mentioned
in NSFH2, the same NSFH1 person number was assigned.
Person Numbers Assigned to Persons Not Mentioned in NSFH1
Persons mentioned in NSFH2 who were not mentioned in NSFH1 were assigned new
person numbers (values ranging from 41-56). These include new spouses and
partners; children born since NSFH1; sons and daughters who should have been
mentioned in NSFH1, but who were omitted; new roommates or relatives living
in the household; and new spouse/partners children living (living with R or
elsewhere). Also included are the ex-spouse's new household members.
The initial assignment of these new person numbers was automated - i.e., the
computer assigned them sequentially starting with "41" within the interview.
Potentially there were five separate interviews per case - main respondent,
current spouse/partner, ex-spouse/partner, focal child, and parent. Each
interview is self-contained and in the interview process there was no way to
cross-reference information obtained in the interview with another respondent
(e.g., the main respondent in the case of an interview with a spouse). The
person numbers had to be reconciled among the three interview types. For
example, a respondent may have separated, divorced, remarried and acquired
a step-child, and had a baby with the new spouse between NSFH1 and NSFH2.
The main respondent may have mentioned the new baby first (assigned a 41),
the new spouse second (42), and the step-child third (43). In the spouse
interview, the respondent (i.e., the new spouse) would be assigned a person
number of 41. If the spouse mentioned her child (R1's step-child) first
and the new baby second, the person numbers would be reversed when compared
to the main respondent's numbers. If the ex-spouse was living with a
new partner, the new partner would have been assigned a person number
of 41.
We have reassigned person numbers so that each person is assigned a unique
person number across all interviews for a particular case. This proved to be
a very complex and tedious process. The unambiguous cases are those where
the names and ages are the same, and the relationships are consistent (R1's
biological child; spouse's step-child, and the residence is the same (in- or
out- of Rs household)). There were a fairly large number of ambiguous
cases, and there are undoubtedly situations where a person was mentioned at
NSFH1 and at NSFH2 was assigned a different person number.
Children and step-children appear and disappear between the two interviews.
In the case of biological children (and step-children living with R at NSFH1)
we try to account for missed children in the lost child sequence. We made
no attempt in the interview to determine where children listed in NSFH2 but
not in NSFH1 have come from or why they were not reported in NSFH1.
Some Numbers are Not Assigned
Note that there may be sequential numbers that are not assigned -
for example there may be a 41 and a 43, but no 42. This may occur
for several reasons, but primarily in situations where the person
appeared to be a new person, but later examination revealed that it
was a person mentioned in NSFH1. It can also occur when the interviewer
enters a person on a roster and then later deletes the person.
Not Everyone Assigned a Person Number at NSFH1 was Asked About in NSFH2
Those not asked about include:
non-relatives in household at NSFH1
relatives other than spouse/partner and R's children and step-
children in household at NSFH1
r's step-children at time 1 who are the children of an ex-spouse or
partner at NSFH2
Person Numbers in the Parent Interview
In the parent interviews person numbers were assigned by the computer to all
household members and all sons and daughters and step-sons and -daughters.
These numbers have values of 61-75. No attempt has been made to identify
persons mentioned in the parent interview who were also mentioned in the
main respondent and/or spouse interviews. Thus in a few cases, when the
parent and the respondent lived in the same household, or when the respondent
and a sibling lived in the same household, the same person will have a
different person number in the main respondent and parent interviews. The
main respondent also will have a different number (other than "1") in the
parent interview. There is a variable identifying the main respondent on
the parent interview file. For reasons which cannot be identified a large
number of parents did not list the primary respondent when listing their
children. This is discussed in section X.H.
THE FOCAL CHILD
In the NSFH1 interview, respondents who had any biological, adopted,
step (including partner's), or foster child under the age of 18 living in
their household were ask a series of questions about a child, randomly
selected from among eligible children - the eligible child whose name
came first alphabetically. This is the focal1 child at NSFH1.