Workplace address 2013 Census (information about this variable and its quality)

##Related variables

• Workplace address indicator – indicates whether a respondent worked at home or away from home

##Where the data comes from

Question 39 on the individual form.

Respondents such as milk vendors and sales representatives who have no fixed workplace are asked to state the address of the depot, headquarters, or reporting point from which they operate.

Respondents who have no fixed workplace address at all are asked to respond that they have ‘no fixed workplace address’.

##How this data is classified

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 geographic classifications of meshblock, area unit, territorial authority and regional council. For example, meshblocks aggregate to form area units, and area units aggregate to form both territorial authorities and regional councils. The categories of each of these ordinarily independent classifications sit alongside each other at the same level of the workplace address classification.

Workplace address is therefore a flat classification. The standard codes are:

Meshblock codes (7 digits)

Area unit codes (6 digits) prefixed by '9'

Territorial authority (3 digits) prefixed by '9999'

Regional council (2 digits) prefixed by '99999'

to

9999996 No fixed workplace address

9999998 New Zealand not further defined

For further information about this classification, refer to the:

• 2013 Census data dictionary
• Statistical standard for workplace address.

For background information on classifications and standards, refer to the Classifications and related statistical standards page.

##Subject population

The subject population for this variable is the employed census usually resident population aged 15 years and over.

The subject population is the people, families, households, or dwellings to whom the variable applies.

##Non-response and data that could not be classified

The variable Workplace address does not have a non-response category.

Respondents who were employed but did not state a workplace address were classified as 'New Zealand not further defined'. Respondents who stated a workplace address that could not be coded were also classified as 'New Zealand not further defined'.

• 8.5 percent of responses were coded to the 'New Zealand not further defined' category in 2013.

Respondents who state that they work across multiple areas (eg builders) are coded to 'No fixed address'.

• 0.4 percent of responses were coded to this category in 2013.

Note that there is a difference between the workplace address variable and the workplace address indicator. The workplace address indicator describes whether the respondent works at home or away from home, whereas the workplace address variable consists of the meshblock in which the physical address of the respondent's workplace is situated.

• The non-response rate for the workplace address indicator was 4.7 percent in the 2013 Census, 5.9 percent in the 2006 Census and 12.2 percent in the 2001 Census.

For more information on non-response, refer to the 2013 Census data user guide.

##How this data is used

Data from this variable is used to:

• code industry and sector of ownership variables from Statistics NZ's Business Frame
• measure traffic flows for those who are employed (together with main means of travel to work)
• assess daytime population in specific areas for town and traffic planning and civil defence purposes.

##Data quality processes

All census data was checked thoroughly during processing and evaluation, to ensure that it met quality standards and is suitable for use. These quality checks included edits.

All data must meet minimum quality standards to make it suitable for use.

Quality level

A quality level is assigned to all census variables: foremost, defining, or supplementary.

Workplace address is a supplementary variable. Supplementary variables do not fit in directly with the main purpose of a census, but are still important to certain groups. These variables are given third priority in terms of effort and resources.

Mode of collection impacts – online forms compared with paper forms

The online forms had built-in editing functionality that directed respondents to the appropriate questions and ensured that their responses were valid. As a result of this, data from online forms may be of higher overall quality than data from paper forms. The significance of this will depend on the particular type of analysis being done. There will always be a mode effect but this cannot be measured. Statistics NZ design and test to minimise the effects of mode for all questions.

There were differences between how the forms were completed online and on paper for this variable:

• The online form allowed only one response to be selected for the work at home / work away from home tick boxes for the workplace address question. If a further response was selected, the response given previously disappeared. Multiple responses to this question were possible when forms were completed on paper.
• On the online form, it was only possible to give text responses if 'Work away from home' was marked. When forms were completed on paper, it was possible to give a text response but not mark the 'Work away from home' tick box.
• On the online form, only people who gave a New Zealand address in question 5, were 15 years of age or older, and answered that they were in employment in question 32 were able to respond to the workplace address question (question 39). On the paper form it was possible for overseas visitors, respondents under 15 years of age, or those unemployed to respond to question 39.

##Quality assessment of data and data quality issues for this variable

Overall quality assessment

Moderate: fit for use – with some data quality issues to be aware of. 2013 Census variable quality rating scale gives more detail.

Issues to note

• Most of the problems with workplace address stemmed from coding issues. Sometimes these were caused by respondents supplying insufficient detail to workplace addresses to be coded. Where possible workplace addresses were coded using Statistics NZ's Business Frame.
• At times problems exist when the legal name of a business in the Business Frame does not match with that given in the census. This is a common difficulty our coders all over the country face, particularly when respondents give common or trading names which differ from a business legal name. While the census implements a number of strategies to evaluate and edit data to ensure its quality, the self-completion nature of the census imposes limitations on getting exact data at very low geographic levels. Where we cannot get a direct match to the Business Frame, industry and sector of employment is coded via a separate industry codefile – and the workplace meshblock might be coded to a higher geographic level, or not coded at all (ie end up as 'New Zealand not further defined'), depending on the level of detail the respondent gave.
• Some care needs to be taken with small numbers when usual residence by workplace address cross-tabulations are used, due to some issues when workplace coding at a detailed level. Problems included: operators sometimes miscoding the workplace address written by a respondent; coding a workplace address to the head office instead of the regional office, and insufficient detail for a workplace address to be coded at a small area level (such as meshblock or area unit). The issue with regional offices being coded to head offices was resolved for the majority of cases (as these occurred in Auckland, Wellington, and Christchurch) but there may be some cases when an apparent long-distance commute between other regional centres may be due to miscoding. This situation would have occurred in previous censuses.
• Note that inconsistencies can occur when comparing 'Worked at home' counts from travel to work variable with 'Work at home' for workplace address indicator variable due to differences in the reference period. The travel to work question is based on the one day, Tuesday, 5 March 2013, while the work questions are based on the seven days ended Sunday, 3 March 2013.

##Comparing this data with previous census data

This data is broadly comparable with data from the 2006 and 2001 Censuses. Changes in the data over this time period may be partly due to changes in the collection, definition, or classification of the data rather than to real change.

The Business Frame match rate for workplace address has increased from 60.2 percent in 2006 to 64.6 percent in 2013. This will improve the quality of both workplace address and industry/sector coding relative to 2006.

An issue occurred in 2006 where usable responses were coded to 'New Zealand not further defined' instead of the correct meshblock, leading to a higher ‘New Zealand not further defined’ count. This has affected the comparability of the data over time. In 2001, 12.5 percent of the subject population was coded to 'New Zealand not further defined', compared with 14.5 percent in 2006. Due to better coding in 2013 only 8.5 percent of responses were coded to 'New Zealand not further defined ' in 2013.

Changes in meshblock boundaries and other geographic classifications will also affect the comparability of the data for this variable over time.

There was a slight change in the structure of the question from that used in the 2001 Census. In 2013 and 2006, there were four boxes for the address details and, in 2001, there was only one. There is some indication that this has lowered the non-response rate for the 'Workplace address' indicator variable, but no real evidence to suggest it has affected the results for 'Workplace address' itself.

##Comparing this data with data from other sources

Census is the only information source that provides comprehensive information for small areas and small populations. However alternative sources of information about this subject are available:

• Household Labour Force Survey (HLFS)
• Linked Employer-Employee Database (LEED).

Data from these alternative sources may show differences from census data for several reasons. These could be due to differences in scope, coverage, non-response rates, data being collected at different periods of time, alternative sources being sample surveys and as such subject to sampling errors, or differences in question wordings and method of delivery (self-administered versus interviewer-administered). Data users are advised to familiarise themselves with the strengths and limitations of individual data sources before comparing with census data.

##Further information about this data

All percentages for workplace address indicator in census publications have been calculated using 'Total stated' as the denominator.

When using this data, be aware of the following:

• In the 2013 Census approximately 35 percent of workplace addresses were manually coded from the workplace address written on census forms. The other 65 percent were coded by locating the respondents' business on Statistics NZ’s Business Frame, which is a comprehensive database of New Zealand businesses and their structure.
• The Business Frame is derived and maintained from tax records with the primary purpose of acting as a statistical register for drawing and maintaining business surveys operated by Statistics NZ. Using a single centralised frame for all Statistics NZ business and economic surveys allows better management of respondent burden and ensures that classifications and statistical unit definitions are applied consistently. The quality of workplace address information in the census can be affected by a number of factors, including:
  • Timing – for the 2013 Census, the Business Frame codefile was updated twice, firstly on 2 April 2013 before census data processing commenced. The second time, 10 July 2013, was during processing after the Business Frame completed their annual update. There can also be time lags in terms of businesses starting /ending / moving and this information being updated on the frame.
  • The Business Frame maintenance strategy – this is stratified so that larger businesses receive more attention. Small businesses receive less frequent attention in order to reduce respondent burden.
  • The level of detail that respondents provide – where incomplete information is given, the response may be coded to a higher geographic level, or may not be able to be coded at all.
  • Business Frame matching – this is done on the legal name of the business, not the trading name, which can result in a non-match. (Wherever possible, non-matches are coded through a separate codefile).

Contact our Information Centre for further information about using this variable.