Quality Statement
Workplace address is the physical location of a workplace. Distinguishing details can include the building name; street number, name and type; suburb or rural locality; and city, town, or district. Workplace address relates to a person’s main job, that is, the job in which they worked the most hours. The address identifies where an individual carries out their work and could be their employer’s address, home address, or another location.
Moderate quality
Data quality processes section below has more detail on the rating.
Priority level 2
A priority level is assigned to all census concepts: priority 1, 2, or 3 (with 1 being highest and 3 being the lowest priority).
Workplace address is a priority 2 concept. Priority 2 concepts cover key subject populations that are important for policy development, evaluation, or monitoring. These concepts are given second priority in terms of quality, time, and resources across all phases of a census.
The census priority level for workplace address remains the same as 2018.
The 2023 Census: Final content report has more information on priority ratings for census concepts.
Employed census usually resident population count aged 15 years and over
‘Subject population’ means the people, families, households, or dwellings that the variable applies to.
The workplace address classification consists of a combination of classifications that are ordinarily stored independently of each other. There is a hierarchic relationship between the New Zealand geographic classifications of meshblock, statistical area 2, statistical area 3, territorial authority, and regional council. They are combined for the purpose of providing as much detail as possible for workplace address. Ideally workplace address would be defined at the meshblock level. Where this is not possible, the next most detailed geography is used.
Workplace address is a flat classification. The standard codes are:
Workplace or Educational Institution Address V2.0.0
- Meshblock codes (7 digits)
- Statistical Area 2 codes (6 digits) prefixed by '9'
- Statistical Area 3 codes (5 digits) prefixed by '99'
- Territorial Authority codes (3 digits) prefixed by '9999'
- Regional Council codes (2 digits) prefixed by '99999'
- 8888888 Overseas
- 9999977 Response unidentifiable
- 9999988 Response outside scope
- 9999996 No fixed address
- 9999998 New Zealand not further defined
- 9999999 Not stated
Address information (name of building; street number and street name or name of shopping centre; suburb or rural locality; and city, town, or district) is used to place a person into the classification.
The 2023 classification used for workplace address is similar to 2018 in structure; however area codes have been updated to reflect the 2023 geographic pattern, including the removal of urban rural areas and the addition of statistical area 3 categories.
Workplace address indicator is classified into the following categories:
Census workplace address indicator V1.0.0 - level 1 of 1
Code | Category |
---|---|
1 | Worked at home |
2 | Worked away from home |
9 | Not stated |
Workplace address indicator uses a 1 level flat classification presented in the table above. 'Not stated' is a residual code.
Standards and classifications has more information on what classifications are, how they are reviewed, where they are stored, and how to provide feedback on them.
Workplace address information is collected from the individual form (question 46 paper form).
Individuals were only required to answer this question if they indicated in earlier questions that they were 15 years or older, and they were employed. Respondents were instructed to answer this question regarding the job they work the most hours in.
The workplace address question is comprised of two sections. The first determines if respondents work at home or away from home. If they work away from home, full address information is collected.
There were differences in question layout, and the way a person could respond between the modes of collection (online and paper forms).
On the online form:
- routing ensured the workplace address question was only available to employed adults who mostly work away from home
- multi-response was not possible
- Online, respondents were able to use an as-you-type list of business names and an address list for businesses to find and code their workplace addresses.
On the paper form:
- anyone could see and respond to the workplace address question even if they:
- were not in the subject population
- were not asked to give an address because they mostly work at home,or
- had left the tick boxes blank
- were not in the subject population
- multiple answers to the tick boxes were possible, but this was resolved in processing by comparing workplace address to usual residence address
- a free text response field was available to write workplace address.
Data from online forms may therefore be of higher overall quality than data from paper forms. However, processing checks and edits were in place to improve the quality of the paper forms.
Stats NZ Store House has samples for both the individual and dwelling paper forms.
Data-use outside Stats NZ:
- in conjunction with other transport and location data from the census (for example, travel to education or travel to work) to plan and manage transport and other infrastructure (for example, by transport planners in large cities where there are congestion issues)
- to assess daytime population in specific areas for civil defence purposes
- to measure the number of people working from home
- for monitoring investment in certain travel modes, such as investments to support walking and cycling
- to focus targeted initiatives aimed at encouraging more people to use public transport.
Data-use by Stats NZ:
- in conjunction with main means of travel to work to measure commuting flows of the employed population.
Alternative data sources were used for missing census responses and responses that could not be classified or did not provide the type of information asked for. The table below shows the distribution of data sources for workplace address data.
Data sources for workplace address data, as a percentage of the employed Census usually resident population count aged 15 years and over, 2023 Census | ||
---|---|---|
Source of workplace address data | Percent | |
2023 Census response | 78.8 | |
Historical census | 0.0 | |
Admin data | 7.6 | |
Deterministic derivation | 13.6 | |
Deterministic derivation from census response | 8.4 | |
Deterministic derivation from admin data | 5.1 | |
Deterministic derivation from CANCEIS(1) | 0.2 | |
Statistical imputation | 0.0 | |
No information | 0.0 | |
Total | 100.0 | |
1. CANCEIS = imputation based on CANadian Census Edit and Imputation System Note: Due to rounding, individual figures may not always sum to the stated total(s) or score contributions. |
While the data sources themselves have not changed since the 2018 Census, how they are reported has changed. In the 2023 Census, workplace address data is primarily sourced from 2023 Census response.
Where this is not possible, responses have been deterministically derived using another census variable (usual residence). In 2018, this was recorded to the data source for the usual residence address (census response, response from a partial form or admin data). If an address could not be sourced from the Statistical Business Register (SBR), admin data was used where their employer information was available. The data source was admin data from Inland Revenue.
Deterministic derivation includes the following situations:
- Where a respondent stated they worked from home, their usual residence address was used to derive workplace address. The data source would be deterministic derivation from census response.
- Where a workplace address could not be sourced from a census response, individuals were linked to the Statistical Business Register (SBR) by the name of their employer given on the census form. The data source would be deterministic derivation from census response.
- Where an individual was imputed to working at home, usual residence address was used to derive workplace address. Usual residence address data source determined the type of deterministic derivation being from either 2023 Census response (either from an individual form or household listing), admin data, or from CANCEIS imputation.
- Remaining missing responses, and individuals that work at ‘No Fixed Address’ were coded to the territorial authority of the individual’s usual residence address.
Editing, data sources, and imputation in the 2023 Census describes how data quality is improved by editing, and how missing and residual responses are filled with alternative data sources (admin data and historical census responses) or statistical imputation. The paper also describes the use of CANCEIS (the CANadian Census Editing and Imputation System), which is used to perform imputation. This webpage also contains a spreadsheet that provides additional detail on the admin data sources.
Missing and residual responses represent data gaps where individuals either did not provide answers (missing responses) or provided answers that were not valid (residual responses).
Missing or residual responses are replaced by alternative data sourcing or the individual’s usual residence address territorial authority. As such, there are no residual values in the output data for the subject population.
Percentage of ‘Not stated’ for the workplace address indicator in the employed census usually resident population count aged 15 years and over:
- 2023: 0.0 percent
- 2018: 0.0 percent
- 2013: 4.7 percent
Where a territorial authority spans multiple regional council boundaries and doesn’t map directly to a single regional council, they are coded as response unidentifiable for the regional council classification.
Percentage of regional council ‘Response unidentifiable’ for the employed census usually resident population count aged 15 years and over:
- 2023: 0.6 percent
- 2018: 0.5 percent
- 2013: 0.2 percent
Overall data quality: Moderate
Data has been evaluated to assess whether it meets quality standards and is suitable for use.
Three quality metrics contribute to the overall quality rating:
- data sources and coverage
- consistency and coherence
- accuracy of responses.
The lowest rated metric determines the overall quality rating.
Data quality assurance in the 2023 Census provides more information on the quality rating scale.
Data sources and coverage: Moderate quality
The quality of all the data sources that contribute to the output for the variable were assessed. To calculate the 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:
- 0.98–1.00 = very high
- 0.95–<0.98 = high
- 0.90–<0.95 = moderate
- 0.75–<0.90 = poor
- <0.75 = very poor.
While the majority of the data has come from census responses, the available alternative data sources of admin data and deterministic derivation have lower ratings, resulting in a score of 0.90 and a metric rating of moderate.
Data sources and coverage rating calculation for workplace address data, as a percentage of the employed census usually resident population count aged 15 years and over, 2023 Census | |||
---|---|---|---|
Source for workplace address data | Rating | Percent | Score contribution |
2023 Census response | 1.00 | 78.75 | 0.79 |
Admin data | 0.50 | 7.62 | 0.04 |
Deterministic derivation from census response | 0.56 | 8.36 | 0.05 |
Deterministic derivation from admin data | 0.54 | 5.10 | 0.03 |
Deterministic derivation from CANCEIS(1) | 0.52 | 0.17 | <0.01 |
No information | 0.00 | 0.00 | 0.00 |
Total | 100.00 | 0.90 | |
1. CANCEIS = imputation based on CANadian Census Edit and Imputation System. Note: Due to rounding, individual figures may not always sum to the stated total(s) or score contributions. |
The introduction of deterministic derivation as a data source provides greater transparency of how the data is sourced and better reflects the accuracy of the data, as reflected in the ratings.
Consistency and coherence: Moderate quality
Workplace address data is mostly consistent with expectations across consistency checks. There is an overall difference in the data compared with expectations and benchmarks, which can be explained through a combination of real-world change, incorporation of other sources of data, or a change in how the variable has been collected.
The coherence between expectations and the data is high, however consistency with 2018 Census (especially at lower geography levels) is lower due to significant changes to the coding logic, alongside improvements to data processing, coding specificity, and data capture.
These changes include:
- coding logic improvements, alongside improvements to the processing of the data, specificity of coding, and the proportion of responses that have been coded, that has led to higher quality data compared with 2018.
- the change to alternative data sourcing to allow workplace address outside of the usual residence territorial authority in scenarios where it is plausible (for example, Lower Hutt city to Wellington city)
- coding all remaining residual values to the territorial authority of the individual's usual residence address has had a notable impact on the data. In 2018, non-response was coded to the regional council of the usual residence address.
These changes appear to have improved the quality of the data overall with more census responses coded, and a much greater specificity of coding (for example, more coded to the meshblock level). However, such substantial change from 2018 makes the data somewhat inconsistent with time series, particularly for alternatively sourced values.
Accuracy of responses: Moderate quality
Workplace address data has various data quality issues involving several aspects of the data, or entire level of a hierarchical classification. The data quality issues could include problems with the classification or coding of data, such as vague responses resulting in coding issues, or responses that cannot be coded to a specific (non-residual) category, thereby reducing the amount of useful, meaningful data available for analysis.
The data quality of workplace address data has improved through coding to lower levels of geography, higher quality scanning of paper forms, and more scanning repair, but unavoidable quality issues still occur, due to respondents misunderstanding or misinterpreting the question and providing vague answers. This was an issue particularly for paper responses and online responses that did not use the as-you-type (AYT) list. Coding was higher quality for online responses that used the AYT list.
Users should note the following improvements since 2018:
- scanning repair improvements
- expanded use of manual intervention
- the pre-emptive inclusion of place look-ups to code responses that are not addresses
- extensive revision of the coding rules both prior to the data being received and in response to identified improvements after analysing the data.
Overall, there were improvements in coding to lower levels of geography compared with 2018.
Workplace address data can be used and compared with data from the 2013 and 2018 Censuses.
When using this data, be aware that:
- This data represents where people travel to for work based on their usual travel patterns. This is not necessarily reflective of the address of their employer.
- Substantial changes since 2018 that have improved the accuracy of coding may produce unexpected results (especially at lower geography levels). Care is advised when making time series comparisons at low levels of geography.
- Coding rules when matching to the business register were relaxed to include travel between neighbouring territorial authorities (such as Wellington and Porirua cities or Christchurch city and Selwyn district. This is likely to have resulted in travel between neighbouring territorial authorities increasing so caution should be used when interpreting change.
- The workplace address for people who work at home is the same as their usual residence address – if someone is working remotely, their workplace when they are not working at home is not captured.
- People who work at home should be excluded from the data if the data is being used to capture travel patterns and interpret change between census years.
- The majority of records coded to ‘Territorial authority not further defined’ have been coded to the territorial authority of their usual residence address as there was no other information available for their workplace address. Although most people live and work in the same territorial authority, this is not necessarily accurate for everyone.
Furthermore, there are a lot of complexities for users to be aware of at different geographic levels:
- Workplace addresses were coded to the most specific geographic area possible from the available information – this varies depending on the specificity of a response, and the available alternative data. In some data the workplace address matches the meshblock (and often the address) of their usual residence address, despite the respondent ticking 'work away from home'. Because the respondent said that they mostly work away from home, these records have remained in the data as working away from home.
Comparisons to other data sources
Although surveys and sources other than the census collect workplace address data, data users are advised to familiarise themselves with the strengths and limitations of the sources before use.
Key considerations when comparing workplace address information from the 2023 Census with other sources include:
- Census aims to be a national count of all individuals in a population while other surveys (such as the Household and Labour Force Survey and the General Social Survey) measuring this variable are only based on a sample of the population.
To assess how this concept aligns with the variables from the previous census, use the links:
Contact our Information centre for further information about using this concept.