• This section describes variables related to household and family composition, household identification, linkages between members of multiple respondent households, and household residence.

Household members

The term "household" refers to all individuals sharing the respondent's primary residence at the time of the interview. For respondents living in temporary quarters (except temporary military quarters), the usual residence is defined as that person's permanent residence. For those living in their own dwelling unit or in military family housing, the usual residence is the person's dwelling unit. For example, if a male college student is living in a temporary residence, such as a fraternity, those who share his permanent residence, such as his parents' address, would be considered his household members. However, if that same college student were living in his own apartment, all those living in his apartment would be considered his household members.

Persons analyzing military households should note that household screener information was not collected for persons in the military sample. Thus, while military units are included in the total 8,770 unique households, military units cannot be multiple respondent households. Household specification for those respondents enlisted in the military is as follows: (1) for those in the military who are married but living in military quarters other than military family housing, the household is the household of the respondent's spouse, and (2) for those in the military who are unmarried, no household information is recorded.

During PAPI surveys, information about a respondent's household was gathered during a separately administered household interview. Three different Household Interview Forms "Household Interview Forms" were used prior to 1987: 

• Version A was completed by a parent of those respondents living in a parental household
• Version B was administered to youth not living at a permanent address
• and Version C was answered by those respondents living in their own dwelling unit or independent living quarters

Table 1 details, by survey year, the relevant universes and residence types specific to each version; notes on variations in administration of the forms are included. A series of variables entitled 'Version of Household Record Used' is available for the 1979-80 and 1982-86 survey years. To determine the version of the household interview used in 1981, it is necessary to match information from the variable, 'Type of Residence R is Living In,' to residence information that was included on the three different forms. Beginning in 1987, only one version of the household interview was used, as all respondents were 22 or older and living predominantly on their own. Since the introduction of CAPI interviews in 1993, household information has been collected in the first section of the main questionnaire rather than in a separate instrument.

All members of the respondent's household are enumerated each survey year on the household record; in 1978, household members were listed on the household screener. The relationship generally listed for each household member on the household record is relative to the youth respondent, such as 'Household Record - Relationship to Youth Member # 1.' For variables from the screener and for one series of 1979 household record variables, the relationship of household members (only family members in the screener) is relative to the householder. Anyone who lives in the residence but is temporarily away is listed; anyone who is there only temporarily is removed from the listing. For the screener and for interviews in which the respondent lives in a new household, that is, living with new people rather than living at a new address, the householder generally is listed first, followed by a spouse; any children; any other relatives; and any roomers, boarders, hired help, or other usual unrelated residents.

Family members

Within the listing of household members, family units are identified through family unit numbers and relationship codes. A family unit includes all those related by blood, marriage, or adoption. For each member of the household in every survey year, including the 1978 screener, the family unit number is listed on the "Household Enumeration" or the screener, such as 'Household Record - Family Unit # 1 Member # 1.' All family members in an interrelated group will share a family unit number, with number 1 assigned to the respondent's family. Each additional interrelated group or individual adult sharing the household but not related to another group or individual in the household constitutes an additional family unit. For example, if Mr. and Mrs. Brown are boarders in the same house with Mr. and Mrs. Smith, the Smiths are the first family unit and the Browns are a second family unit. The reliability of 1979-92 family unit numbers beyond those assigned to the respondent's family is questionable. Beginning in 1993, family units were assigned electronically; the definition of a family unit remains the same. Codes were added for partner's family/relations. All others are assigned a code of "9."

An enumeration of a respondent's children is also available. Several variables have been created as part of the Supplemental Fertility File (area of interest "Fertility and Relationship History/Created"), including variables such as '# of Own Children in Household,' 'Age of Youngest Child in Household,' and a variety of variables for each biological child listed, with some exceptions, in order of age. Unedited variables from the Children's Record Form (areas of interest "Child Record Form/Biological" and "Child Record Form/Nonbiological") are also available for both biological and nonbiological children. If there is a discrepancy between household rosters versus marriage event histories use the marriage event histories rather than the household roster data. See the "Fertility" section of this guide for more information about the collection of information on the respondent's children.

Additionally, information about household members includes, sex of the member, their relationship to the respondent, age, highest grade completed, and whether the member receives pay for work (age restrictions apply.)

Finally, information on whether the mother and father of each child (in 1991, new children only) live in the household is available for the 1987-2014 survey years. In all other years, information on whether the father of the child is present is available for children of female respondents.

Family size

Beginning with the 1990 release, a family size variable, comparable to the family size variable created for the computation of the 'Total Net Family Income' and 'Poverty Status' variables, was created for each year. The variable is constructed by simply cycling through the household record "relationship codes" and increasing the family size by one each time a qualifying relationship relative to the respondent is encountered. Qualifying relationships include all relations by blood, marriage, and adoption. Foster relationships, partners, boarders, guardians, and other individuals are not considered family members in the creation of this variable.

Program derivation

The SPSS program statements for a sample survey year (1979-92) FAMILY SIZE variable are as follows:

COUNT FAMSZXX=RELR1 TO RELR15 (0 THRU 32,37 THRU 44,47 THRU 49)

IF (WEIGHTXX EQ 0) FAMSZXX=-5

After 1993, the roster was expanded to accommodate up to 20 individuals. The SPSS program is the same but the number of relationships to check is five larger. Additionally, the respondent is not on the household roster after 1992, so FAMSZXX is initialized to "1."

Household identification and linkages

The NLSY79 screening procedure allowed more than one member of a household to be selected for interviewing. The original 12,686 respondents were members of 8,770 households; 6,742 respondents or 53 percent of the sample were members of households from which more than one respondent originated, while 5,944 respondents or 47 percent were members of single respondent households (Table 2). To establish linkage of respondents originating from the same household, variables identify other interviewed household members and their relationships as of the 1979 interview. The 1979 variable providing the unique household identification number of each household is R00001.49, 'Household Identification Number' (HHID). The same HHID is assigned to all respondents who originated from the same household in 1979. In multiple-respondent households, the HHID corresponds to the lowest respondent 'Identification Code' of all respondents interviewed in that household; in single-respondent households, the HHID corresponds to the respondent 'Identification Code.' The HHID variable was constructed using other created variables from the NLSY79 main data set and exists only for 1979.  Multiple respondent households can also be identified through variables that identify other respondents in the same household and their relationship to the first respondent. Reference numbers include R00001.50-R00001.61, for example, 'Identification Code of 1^st Other Interviewed Youth in R's Household.'

Although these matches represent unique samples for a number of research topics, be aware that matches may be not be demographically representative due to the age restrictions applied to all members chosen from a household. The primary types of relationships that existed among respondents within multiple respondent households at the time the surveys began included brothers, sisters, husbands, and wives (Table 3). Other relationships included cousins, brothers- and sisters-in-law, step-brothers or -sisters, and other types of household members.

Household residence

Household residence refers to the type of dwelling or living situation of the respondent. Household residence information is available for the respondent at each survey point, for the respondent during his or her childhood, and for the respondent's children during recent surveys. The variable 'Type of Residence R is Living In' classifies the respondent's actual place of residence at the time of each survey. From 1979-86, it was created based upon responses to several questions asking about different types of dwelling units. In these years, several versions of the Household Interview Forms (the instrument completed before the main questionnaire and used to construct the household enumeration) existed. The universes for these different versions were dependent upon the type of dwelling unit in which the respondent lived (parental home, own dwelling unit, individual or group quarters), the sample type of the respondent (military or civilian), and who answered the household interview section questions (respondent or parent). The responses to questions designating type of residence from each of these versions were combined into one variable reflecting type of residence for the entire sample.

Beginning in 1987, the several versions of the Household Interview Forms were combined and all types of residences were coded in one question. Therefore, after 1986, this question is no longer considered a "created" variable. The 'Type of Residence R is Living In' variables include categories such as dorm, fraternity or sorority, hospital, jail or juvenile detention center, orphanage, religious institution, own dwelling unit, parents' household, and specific types of military quarters. The codes assigned to response categories for type of residence in 1979 differ significantly from those in other survey years. Also, in earlier years, respondents living in parental homes were treated as valid skips; in later years, these respondents were assigned a separate code that differs by year.

Retrospective information describing the respondent's childhood living arrangements was collected during 1988 in a three-part series of questions on the Childhood Residence Calendar. In Part 1, the respondent's identification of any type of parent with whom he or she lived for four or more months was recorded. Coding categories included biological, adoptive, or stepmother or father for each age from birth through 18 years, for example, 'Lived with Biological Mother at Birth,' 'Lived with Adoptive Father at Age 16.' Ages at which the respondent stopped living with a parent, the reason for ending shared living arrangements, and the frequency of visitation with the absent parent during the first year after coresidence ended were also collected. For those ages when the respondent reported not living with a parent, information was collected in part 2 of the Childhood Residence Calendar on:

• coresidence with grandparent(s), other relative(s), foster parent(s), or friend(s)
• residence in a children's home or orphanage, a group care home, a detention center/jail/prison, or another institution
• use of another type of arrangement
• for those ages ten and over, whether the respondent was left to be on his or her own

Variable titles for this series include 'Lived with Foster Parent(s) (Not Living With a Parent) at Age-7,' 'Lived in Children's Home/Orphanage (Not Living With a Parent) At Age-2,' 'Left to be on Own (Not Living With a Parent) at Age-15.' The number of foster or group care arrangements experienced by the individual is also recorded. Finally, in part 3 of this supplement for each age during which the respondent experienced more than one living arrangement when not living with a parent, the place at which s/he stayed the longest is identified. Data quality issues are discussed by Haurin (1991).

Information on the residence of respondents' children is available, for the most part, since the 1982 survey year. Note that edited variables based on the 1989 and 1991 raw data are not available until the subsequent year's release. These edited variables, cleaned and checked for consistency, include residences of each biological child in birth order (with some anomalies), such as 'Usual Residence of 7^th Child,' and combine information collected for residence of children of male respondents with that of children of female respondents. Coding categories include in the respondent's household, with other parent, with other relatives, in foster care, with adoptive parents, in a long-term care institution, away at school, deceased, lives part-time with both parents, lives part-time with the respondent and another person, and other. The unedited variables upon which the edited variables are based can be found in the "Fertility" section of the main youth questionnaire and include residence of all biological children; residence of all children born by the time of the previous interview, collected annually since 1984; and residence of all children born since the last interview, collected since 1983. Unedited residence information for nonbiological children is available for 1985, 1986, 1988, 1990, 1992, and from 1994 forward. Coding categories for all unedited variables are the same as for edited variables.

References

Haurin, R. Jean. "Childhood Residence Patterns: Evidence from the National Longitudinal Surveys of Work Experience of Youth." Columbus, OH: CHRR, The Ohio State University, 1991.

Morgan, William R. "Sibling Influences on the Career Plans of Male and Female Youth." Columbus, OH: CHRR, The Ohio State University, 1983.