Relationship is a variable that collects the familial and non-familial relationships of each person:
- to every other person, in a defined group of people, in a private dwelling (relationship matrix) or
- in a defined group of people to one person in a private dwelling (the reference person).
It identifies and classifies the key relationships between persons in the same private dwelling and may include visitors in the private dwelling at the time of the survey.
The relationship matrix allows for the collection of all relationships between all household members (when respondents are asked for their relationship to every other household member) and can be used to derive the familial and non-familial relationships of each person within a family or household.
The reference person approach allows for the collection of familial and non-familial relationships of each person in a defined group of people to one person (the reference person). This information can be collected for all persons in the private dwelling at the time of the survey, including visitors.
In some surveys, where household and family type information is output only for people usually resident in the household, the usual residence indicator may be used to determine usual residents in a private dwelling. For other social surveys, where the usual residence indicator is not derived, people surveyed in the same private dwelling may be identified by using a set of defining questions, as determined by the scope and coverage rules for the survey. The living arrangements information may help to further identify relationship information for people who are in complex families and households where the relationship matrix is not used.
'Dwelling' and 'household' are supporting concepts and are defined in the Glossary. The following supporting concepts are also defined in the Glossary :
- familial relationship
- non-familial relationship
- reference person
- usual residence.
The previous versions of the Relationship standard recommended the selection of the one reference person in the household to whom the relationship of all other members can be reported. However, it is important to note that this approach does not and cannot identify all the relationships in a given household, especially where a household has multiple families. Therefore, a more elaborative method has been developed, namely the household relationship matrix approach. The household relationship matrix approach allows for the collection of all relationships between all household members.
Where feasible, the relationship matrix is the recommended approach for collecting information on the relationship variable. Otherwise, users are recommended to use the reference person approach combined with living arrangements. In some surveys, where household and family type information is output only for people usually resident in the household, the usual residence indicator may be used to determine usual residents in a private dwelling. For other social surveys, where the usual residence indicator is not derived, people surveyed in the same private dwelling may be identified by using a set of defining questions as determined by the scope and coverage rules for the survey.
The living arrangements information may help to further identify relationship information for people who are in complex families and households where the relationship matrix is not used. The combination of this information may be used to derive familial and non-familial relationships of those present in the private dwelling at the time of the survey, this also includes absentees.
Guidelines and requirements for using the 'relationship to reference person' or the 'household relationship matrix' approach are provided in the Questionnaire module page.
There are also certain constraints on the definition of relationship. Firstly, respondents sometimes intentionally mis-report their actual relationships. This could be because they find the question an invasion of their privacy, or there might be a particular relationship they do not wish to disclose. For example, a beneficiary may not want to report that they are living with a partner.
The second constraint is that relationships between people are not always clear-cut, since people relate to each other in several ways. For example, a person could be a brother, a landlord, a flatmate and an employer of the other person. Which relationship would we expect to be reported? In this scenario, because the familial relationships are presented first, we would expect that the respondent would report the familial relationship – brother. But it is worth noting that we must rely on the discretion of the respondent to report the most significant aspect of multifaceted relationships.
The third constraint is that relationships can be mis-reported unintentionally. Respondents can get confused answering the questionnaire module and may reverse their relationships to other people. For example, an aunt who is filling out the questionnaire module for relationship may accidentally write 'aunt' instead of niece when reporting the relationship of the niece to her. Reversed relationships can sometimes be corrected by checking other information provided by the respondent. Surveys should respect respondents' intentions and allow changes where it is clear which piece of information is correct and is considered worthwhile for output purposes. Since relationship information is not used directly in output but rather input to family coding it is important that the results in family coding are correct as this is key information used as a base to derive the family and household output variables.
Finally, notions of what constitutes a family can vary considerably. Some respondents define their family as only relatives with whom they live. Other respondents' definition of family includes relatives who live in other dwellings. However, there is also the notion that a family may include people who are unrelated, whether or not they usually live together. Statistics New Zealand definitions for what constitutes a familial versus a non-familial relationship and related versus unrelated people are covered briefly in the explanatory notes.
The relationship variable is used to collect the familial and non-familial relationships between persons in the same private dwelling and may include visitors to the dwelling on the night of the survey. However, for practical purposes, surveys may only wish to output family and household information for those persons usually resident in the same private dwelling. As a result, the household and family structures used by some Statistics NZ outputs may not adequately reflect the social and family relationships relevant in New Zealand. It is the subject area's responsibility to inform data users who is included or excluded from the collection of the relationship variable.
Changes since the 1999 review of the standard
Flexible approach to the use of the relationship classification
Classifications and Standards are proposing a new approach that will balance the needs of users who want simple classifications while at the same time ensuring greater consistency and comparability across collections.
To achieve this we have developed a core classification framework for some social topics including relationship. This core framework contains a comprehensive list of categories that meets the needs of all current collections. A survey will not need to use every level of a classification, nor will they need to use every category within a level.
Aggregations of the core classification framework will be versions of the core framework and will be developed by the classifications team, in consultation with survey areas, and then stored in CARS. This will consequently enable more comparison between surveys as the definitions and conceptual basis will be consistent.
Inclusions in the standard
The new Relationship standard will include:
- Relationship master classification (core classification framework)
- Alternate version for Census of Population and Dwelling
- Alternate version for the GSS and HLFS
- Alternate version for theFamily Survey.
For more information, see the downloadable Excel file on the Classification page.
The 2008 versions of the classification maintain a similar hierarchical structure to the previous version with three levels. This will continue to enable survey areas to utilise the classification in various amounts of detail.
The key changes made to the current Relationship classification are the inclusion of civil unions, de facto relationships and step relationships. Step relationships are now included as a response option to identify step families. Key familial relationships are maintained at level 1 and 2 to identify key family types and structures. Level 3 of the classifications provides detailed relationship information that will further assist in coding and deriving family and household variables.
The statistical unit for which relationship is an attribute is always a person. However, the variable can be collected for any subpopulation. Most collections are based on the ideology that the coverage of the relationship variable uses the rules
- of usual residence in dwellings, or
- people being present (or absent) at the time of the survey, or
- some variant of these as determined by the scope of the survey.
Which of these rules is used needs to be identified clearly to data users so they are able to determine who is included or excluded from the collection of the relationship variable.
Living arrangements versus relationship
The 'living arrangements' concept differs from 'relationship' in that living arrangements collects information about the type of relationships people have with all the people with whom they usually reside or live, but not who the relationship is with. For example, in a household with an adult and two children, a response to a simple living arrangements question 'Who do you usually live with?' can be 'I live with my child and one other unrelated person'. This does not identify which of the two children is the child of the respondent.
Relationship variable, on the other hand, collects information on the specific relationship(s) of each person in a defined group to one person (reference person) or to all persons in that group in the private dwelling. For the above example, with relationship, the survey can identify that person 2 is the child of the respondent person 1, or person 3 is unrelated to person 1.
Relationship identifies and classifies the key relationships between persons in the same private dwelling and may include visitors in the private dwelling at the time of the survey. The living arrangements information may help to further identify relationship information for people who are in complex families and households where the relationship matrix is not used.
Related versus unrelated people
The definitions of related and unrelated people, as well as familial and non-familial relationships are given in the Glossary. In summary, related people are people who have familial relationships (related by birth/biologically, or by registered marriage, civil union, de facto, fostering or adoption), and unrelated people have a non-familial relationship to reference person.
Criteria for reference person
The criteria for selecting the reference person are survey-specific. However, there are two general guidelines that should be observed by all surveys. Firstly, the reference person is preferably an adult. The questionnaire module and/or interviewer should specify that an adult should be the reference person. Only when an adult is unavailable is a child acceptable. Secondly, relationship should not be used as an output variable, regardless of the criteria used to select the reference person. In particular, relationship should not be used to summarise characteristics of households. This is because no commonalities can be assumed to exist across the total population of 'reference person' (even with a strict selection criteria) and the characteristics of the reference person should not be generalised to all the other people in the household.