Family Data » Creating Parent Files with One Record Per Respondent

  1. Current Configurations
  2. Need for Creating Standard Parent Files
  3. Creating Files with One Record per Respondent
    1. HRS 2000, 1998, 1996: Distribution Format Is One Record per Household
    2. HRS 2002 and Ahead 1995 and 1993: Distribution Format Is One Record per Respondent
    3. HRS 1994: Distribution Format Includes Many Types of Records
    4. HRS 1992: Distribution Format Is One Record per Parent
  4. Building Parent Files: Detailed Information and Sample Solutions in SAS, SPSS or Stata

1. Current Configurations

Over the course of HRS and AHEAD data collection and distribution, variables about parents have been asked and distributed in a variety of different ways. The table below provides a brief overview.

Data collection If coupled, who was asked family information How distributed
2002 HRS Each respondent One record per respondent with information about mother and father
2000 HRS Family respondent One record per household with information about family respondent's mother, father, mother-in-law, and father-in-law
1998 HRS Family respondent One record per household with information about family respondent's mother, father,  mother-in-law, and father-in-law
1996 HRS Family respondent One record per household with information about family respondent's mother, father,  mother-in-law, and father-in-law
1995 AHEAD Each respondent One record per respondent with information about mother and father
1994 HRS Family respondent Various record types
1993 AHEAD Each respondent One record per respondent with information about mother and father
1992 HRS Female respondent (See 1992 V901 for details) Two or four records per household with information about mother, father, mother-in-law or father-in-law

2. Need for Creating Standard Parent Files

To facilitate longitudinal analysis, the analyst may wish to create files using a standard format for parent information for all waves of data. Two options and possible analytic uses are described below.

  • For example, for an analysis of the flow of resources between individual respondents and their parents, it may be useful to create a file with one record for each respondent for whom parent information is available. Each record in such a file would contain information about both the respondent's mother and father. This document details the construction of such a file for each wave of the data. These files would also facilitate the construction of a file tracking the status of the respondent's parents over time.

  • Alternatively, for an analysis of the flow of resources between respondents' households and their parents, it may be useful to create a file with one record for each household for which any parent information is available; each record in this file would contain information about the family respondent's mother and father AND about the mother and father of the spouse or partner (if any) of the family respondent. While the general information in this document will be helpful, the details of constructing such files are left to the reader.

3. Creating Files with One Record per Respondent

The tasks required to create files with a standardized format of one record per respondent including both mother and father information from the distribution files are described, in brief, in this document. Detailed descriptions are provided in companion documents.

The record should contain information about both the respondent's mother and father:

HHID PN xPN_SP xSUBHH Mom Still Alive How Old Mom Age Mom Died Mom Need Help Dad Still Alive How Old Dad Age Dad Died Dad Need Help
1234 10 20 0 yes 92 -- no yes 95 -- yes
1234 20 10 0 yes 84 -- yes no -- 43 --

A record will only be created if both respondent and parent information is available (the intersection of the Venn Diagram, below). A record will not be created if either 1) parent information is not available because available because an interview was not obtained from the spouse (or partner) designated as the family respondent , or 2) respondent information is not available because an interview was not obtained from the spouse (or partner) of the family respondent.

[Venn Diagram: Union of respondent and parent information]

3a. HRS 2000, 1998, 1996: Distribution Format Is One Record per Household

Parent information was asked of the designated family respondent. She (or he) provided information about her (or his) mother, father and, if married or partnered, mother-in-law and father-in-law. These variables are distributed in a household-level file, which contains the PN of the family respondent and of the responding non-family respondent, if any.

HHID GSUBHH GPN _FAM GPN _NFAM Mom Alive Age Mom Dad Alive Age Dad Mom IL Alive Age Mom IL Dad IL Alive Age Dad IL
1234 0 20 10 yes 84 no -- yes 92 yes 95

In order to create a file with one record per respondent, up to two records will need to be generated from this record with the parent variables and parent-in-law variables in a standard output position. The appropriate value will need to be assigned to the PN and xPN_SP variables from xPN_FAM or xPN_NFAM. (Where x represents the year’s prefix: G=2000, F=1998, E=1996).

3b. HRS 2002 and Ahead 1995 and 1993: Distribution Format Is One Record per Respondent

The 2002, 1995, and 1993 parent variables are included in the respondent-level files; no transformation is necessary to create a file with one record per respondent.

HHID PN Mom Still Alive How Old Mom Age Mom Died Mom Need Help Dad Still Alive How Old Dad Age Dad Died Dad Need Help
1234 10 yes 92 -- no yes 95 -- yes
1234 20 yes 84 -- yes no -- 43 --

3c. HRS 1994: Distribution Format Includes Many Types of Records

Parent information was asked of the designated family respondent. She (or he) provided information about her (or his) mother, father and, if married (or partnered), mother-in-law and father-in-law. This information was combined and reformatted for distribution. The distribution file contains records for mothers, fathers, parents, mothers and stepfathers, fathers and stepmothers, stepfathers, and stepmothers of family respondents and analogous records for in-laws. The records contain PN, the person number of respondent about whose parents the information was provided, CPARCODE, the relationship, mother, father, or parents, and CSUBHH.  In addition to parent information obtained from core interviews, this file also contains information about parents, potentially the same parents, from the exit interviews.

HHID PN CPARCODE CSUBHH Still Alive How Old Age Died Need Help Married To Whomever
1234 10 1 0 yes 92 -- no --
1234 10 2 0 yes 95 -- yes --
1234 10 3 0 -- -- -- -- yes
1234 20 1 0 yes 84 -- yes no
1234 20 2 0 no -- 43 -- --

In order to create file with one record per respondent, the various records with information about the mother, father, parents, mother and stepfather, and father and stepmother will need to be combined into one record with information about the PN’s parents.

3d. HRS 1992: Distribution Format Is One Record per Parent

The distribution file contains two or four records per household with information about a mother or father, and, for households with a married (or partnered) couple, a mother-in-law or father-in-law. APARTYPE designates whether the record is about the mother or father of the Section E respondent or the husband/partner of the Section E respondent.

HHID PN ATYPEPAR Still Alive How Old Age Died Need Help
1234 10 1 yes 84 -- yes
1234 10 2 no -- 43 --
1234 20 1 yes 92 -- no
1234 20 2 yes 95 -- yes

In order to create file with one record per respondent, two records, a mother record and a father record, will need to be combined. The mother variables and father variables will need distinct names, or locations, in the output file. 

4. Building Parent Files: Detailed Information and Sample Solutions in SAS, SPSS or Stata

More detailed information about the reformatting needed to create a respondent-level file with information about both parents is provided. Also provided are parent and parent-in-law cross-reference tables for 2000, 1998, and 1996.

Year Detailed Information and Cross-Reference Tables
2002 No reformatting needed
All
2000
1998
1996
Creating 2000, 1998, or 1996 Parent Files with One Record Per Respondent
     2000 Parent to Parent-in-Law Variable Cross-Reference Table
     1998 Parent to Parent-in-Law Variable Cross-Reference Table
     1996 Parent to Parent-in-Law Variable Cross-Reference Table
1995 No reformatting needed
1994 Creating a 1994 Parent File with One Record Per Respondent
1993 No reformatting needed
1992 Creating a 1992 Parent File with One Record Per Respondent

Sample code in SAS, SPSS, or Stata is available.


Top