Families and households: household composition (information about this variable and its quality)

Description

Household composition classifies households according to the relationships between usually resident people. The classification is based on how many and what type(s) of family nuclei were present in a household, and whether or not there were related or unrelated people present.

Statistics

Representation

Variable Details

Other Variable Information

Following a detailed investigation into potential quality concerns the quality rating of this variable has been changed from very poor quality to moderate quality.

Families and households in the 2018 Census: Data sources, family coding, and data quality has detailed information on the quality assessment for these variables and some changes in our methodology.

Priority level

Priority level 2

We assign a priority level to all census variables: Priority 1, 2, or 3 (with 1 being highest and 3 being the lowest priority).

Household composition is a priority 2 variable. Priority 2 variables cover key subject populations that are important for policy development, evaluation, or monitoring. These variables are given second priority in terms of quality, time, and resources across all phases of a census.

The census priority level for household composition is the same as 2013.

Quality Management Strategy and the Information by variable for household composition (2013) have more information on the priority rating.

Overall quality rating for 2018 Census

Moderate quality

Data quality processes section below has more detail on the rating for this variable.

Subject population

Households in occupied private dwellings (visitor-only private dwellings are excluded)

‘Subject population’ means the people, families, households, or dwellings to whom the variable applies.

How this data is classified

Household Composition Standard Classification 2008 V1.0.0

Household composition is a hierarchical classification with three levels. Level one of the classification has 6 categories, level two has 19 categories while level three contains 31 categories.

The level one categories are:

1 One-family household (with or without other people)

2 Two-family household (with or without other people)

3 Three or more family household (with or without other people)

4 Other multi-person household

5 One-person household

6 Household composition unidentifiable

At level two of the classification, one-family households are classified according to family type, and whether there are other people present. For two-family households in which both families contain children, the criterion is the number of parents in each family. Two-family households that contain at least one 'couple-only family' are not classified to the same level of detail as two-family households in which both families contain children.

At level three of the classification, the criterion for classifying one-family households is whether the other people present are related or unrelated to the family nucleus. In two-family households, the criterion for classification at level three is whether the families are related or unrelated.

The Standards and Classifications page provides background information on classifications and standards.

Question format

Household composition is derived from information about all the people who usually live in the household: residents who are present on census night and residents who are absent on census night (absentees). The following variables are used to determine relationships between the usual residents of the household:

  • Relationship to reference person
  • Absentee relationship to reference person
  • Living arrangements

For online forms the household set-up form asks for residents present on census night, any absentees, and the relationship of each resident to the reference person. Living arrangements is asked on the individual form.

On paper forms the dwelling form asks for residents present on census night and their relationship to reference person (question 17). Any absentees are listed on the dwelling form with absentee relationship to reference person (question 20). Living arrangements is asked on the individual form (question 17).

Both forms collect further information about absentees (usual residents who are absent on census night) regarding whether they are in New Zealand or overseas, and whether they are away for less than 12 months or 12 months or more. Absentees away from New Zealand for 12 months or more are not included as usual residents of the household.

Stats NZ Store House has samples for both the individual and dwelling paper forms.

How this data is used

Outside Stats NZ

  • By central government agencies, local authorities, private organisations, and researchers in the formulation of social policy, for planning and monitoring programmes, and for research purposes.

Within Stats NZ

  • To derive household and family projections.

2018 data sources

We used alternative data sources for missing census responses and responses that could not be classified or did not provide the type of information asked for. Where possible, we used administrative data from the Integrated Data Infrastructure (IDI).

The table below shows the breakdown of the data sources used to derive households.

2018 household composition –
households in occupied private dwellings
Source Percent
Response from 2018 Census 93.4 percent
Response from 2018 Census and administrative data* 0.8 percent
Administrative data 2.2 percent
No information 3.6 percent
Total 100 percent
Due to rounding, individual figures may not always sum to the stated total(s)
*Households where some members were enumerated from 2018 Census forms, and
others from administrative data.
 

‘No information’ in the table above is the percentage of non-responding households. A household is classified as non-responding when we have evidence that a private dwelling was occupied on census night but we did not receive a response and we were unable to identify usual residents of the household in administrative data. The usual residents of these non-responding dwellings may have been present in administrative data, but we did not have enough evidence to place them into a specific household. In these cases, individuals were placed into a meshblock instead. Non-responding households were coded to ‘household composition unidentifiable’.

Administrative data sources

We enumerated individuals from the range of sources available in the Integrated Data Infrastructure (IDI).

The potential for linked administrative data to provide household and family information provides information on sources of administrative data used.

Overview of statistical methods for adding admin records to the 2018 Census dataset provides information on the linking of census responses to the IDI and has more information on the timeliness of administrative data.

Please note that when examining household composition data for specific population groups within the subject population, the percentage that is from administrative data may differ from that for the overall subject population.

Missing and residual responses

Household composition does not have a non-response or ‘not stated’ category. ‘No information’ in the data sources table is the percentage of households in occupied private dwellings which we have no census or admin information about usual residents for and were therefore coded to ‘household composition unidentifiable’. An additional 0.3% of households had usual residents identified through census or admin data but were also coded to ‘household composition unidentifiable’ as we were unable to determine relationships between the residents.

Percentage of ‘household composition unidentifiable’ for households in occupied private dwellings:

  • 2018: 3.9 percent
  • 2013: 2.6 percent
  • 2006: 1.9 percent.

Data quality processes

Overall quality rating: Moderate quality

Data was evaluated to assess whether it meets quality standards and is suitable for use.

Three quality metrics contributed to the overall quality rating:

  • data sources and coverage
  • consistency and coherence
  • data quality.

The lowest rated metric determines the overall quality rating.

Data quality assurance for 2018 Census provides more information on the quality rating scale.

Data sources and coverage: High quality

We have assessed the quality of all the data sources that contribute to the output for the variable. To calculate a data sources and coverage quality score for a variable, each data source is rated and multiplied by the proportion it contributes to the total output.

The rating for a valid census response is defined as 1.00. Ratings for other sources are the best estimates available of their quality relative to a census response. Each source that contributes to the output for that variable is then multiplied by the proportion it contributes to the total output. The total score then determines the metric rating according to the following range:

  • 98–100 = very high
  • 95–<98 = high
  • 90–<95 = moderate
  • 75–<90 = poor
  • <75 = very poor.

For household composition, the data source ratings represent the likelihood that the correct people are placed within the correct household. The proportions of data from the 2018 Census and administrative data contributed to the score of 0.96, determining the high quality rating.

Quality rating calculation table for the sources of household composition data – 2018 households in occupied private dwellings
Source Rating Percent of total Score contribution
2018 Census form 1.00 93.39 0.93
2018 Census and admin data* 1.00 0.82 0.01
Admin data 0.64 2.16 0.01
No Information 0.00 3.63 0.00
Total 100.00 0.96
Due to rounding, individual figures may not always sum to the stated total(s) or score contributions.
*Households where some members were enumerated from 2018 Census forms, and others from administrative data.
     

Consistency and coherence: High quality

Household composition data is consistent with expectations across nearly all consistency checks, with some minor variation from expectations or benchmarks that makes sense due to real-world change, incorporation of other sources of data, or a change in how the variable has been collected.

Data quality: Moderate quality

Data quality was determined by assessing how many records in the subject population had been flagged as having a potential error. An error flag indicates a household may have been miscoded or may have a missing person, and therefore may have an incorrect household composition.

Family and household variables with error flag rates below 6 percent were given a data quality rating of moderate, those with between 6 percent and 9.99 percent were given a rating of poor, while variables with error flag rates of 10 percent or greater were given a rating of very poor.

The households in occupied private dwellings subject population had an error flag rate of less than 6 percent, determining the moderate quality rating for the household composition variable.

Families and households in the 2018 Census: Data sources, family coding, and data quality has more information on error flags.

Quality issues to note

  • Households with 6 or more usual residents, and households with more than one family have a higher proportion of error flags than smaller households.

Quality ratings of related variables

The table below summarises the quality ratings of variables related to household composition.

Quality ratings of related variables
Variable Data sources and coverage Consistency and coherence Data quality Overall quality rating
Count of households High Very high High High
Count of people in households Moderate High Moderate Moderate
Household composition by child dependency status High High Moderate Moderate
Number of usual residents in household High High Moderate Moderate
Number of usual residents aged 15 and over in household High High Moderate Moderate
Number of usual residents aged under 15 in household High High Moderate Moderate
Number of dependent children in household High High Moderate Moderate
Age of youngest child in household High High Moderate Moderate
Age of youngest dependent child in household High High Moderate Moderate

Quality issues to note for related variables:

Recommendations for use and further information

The overall quality of household composition data is moderate and comparable with 2006 and 2013 data.

When using this data you should be aware that:

  • data has been assessed to be consistent at the regional council level of geography. Some variation is possible at geographies below this level.
  • new methodology has been used to assess the data sources and coverage, and data quality metrics. Families and households in the 2018 Census: Data sources, family coding, and data quality has further information on the quality assessment methodology for this variable.
  • edits for consistency with derived family data were carried out for those under 15 with conflicting individual and dwelling form age information. But a small number of errors remain in the data due to errors in either age or relationship to reference person.

Contact our Information Centre for further information about using this variable

Revision Information

Currently viewing revision 8 by on 21/05/2020 9:48:46 p.m.

Revision 8 *
24/05/2020 10:56:09 p.m.
Revision 7
30/04/2020 2:17:29 a.m.
Revision 5
11/03/2020 11:39:04 p.m.
Revision 4
3/10/2019 2:16:37 a.m.
Revision 3
22/09/2019 9:53:26 p.m.

Show / Hide more...

Identifiers

DDI Agency
nz.govt.stats
DDI Id
cc378460-2d1e-453d-9ef0-66480ed0eae1
DDI Version
8

Download

DDI 3 Download

Select the languages to display