Stats NZ has a new website.

For new releases go to

As we transition to our new site, you'll still find some Stats NZ information here on this archive site.

  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+
Results from the Census Test (March 2016)

This report provides an overview of the key results from the Census Test held in March 2016.

The Census Test (March 2016) allowed us to assess new ways of collecting information in preparation for the 2018 Census.

Read this report online or download the PDF from 'Available files' on the right-hand side of this page. If you have problems viewing the files, see opening files and PDFs.

About the Census Test (March 2016)

We asked 22,000 households in Auckland, Waikato, Wellington, and Canterbury to participate in the Census Test (March 2016).

For the test we used an experimental model that enabled us to: 

  • test for differences in self-response rates (the number of households that respond before a field officer visits to follow up), in particular, whether call-to-action letters would be more effective if delivered one week before test day or two weeks before test day
  • find out how people responded to different types of reminders (eg postcards and letters)
  • find out more about the best times to deliver paper forms
  • understand the impact of the new census model on the coverage of dwellings (ie how well we count them).

The test had two phases. In the first phase from November to December 2015, we updated 33,000 addresses for the distribution of census material.

From February to May 2016, we ran the second phase of the test. We contacted our sample of 22,000 households and tested our reminder and visit processes. Our response rate for the test was 59 percent.

Overall, the test was successful and we met all our high-level test objectives to understand how the new collection model:

  1. affects the coverage of dwellings
  2. affects the response rates
  3. behaves in a field environment.

The next section provides the key results from the test.

Test results

This section presents the key results from the Census Test (March 2016).

Self-response rate

The self-response rate, or the proportion of households that responded before a field officer visited to follow up, was 32 percent. We anticipated a response of 35 percent.

Varying the timing of the call-to-action letter had no impact on the self-response rate (see figure 1).

Figure 1

Graph, Self-response rate, for respondents who were sent materials, Census Test March 2016.


Providing paper forms in areas where field officers delivered materials was as effective as the postal reminders (see figure 2).

Figure 2

Graph, Self-response rate by type of contact and materials, Census Test March 2016.


Following up with those who had not responded

Non-response follow-up (NRFU) refers to the different approaches we used to follow up with those who had not responded. These approaches were made in a number of waves. On average, we saw a 30 percent increased response rate across all waves (see figure 3).

Figure 3

Graph, Non-response follow-up – response rates, Census Test March 2016.


Overall response rate

Overall, 59 percent of our test households responded. We were anticipating a response rate of 40 percent.

Use of online forms by individuals

A total of 72 percent of respondents used the online form. Table 1 shows the number of online respondents by region, age group, sex, and ethnicity.

Table 1

 Number of online respondents by region, age group, sex, and ethnicity
   Number of online respondents  Total number of respondents  % of total respondents
 Auckland  7,804  10,276  75.9
 Waikato  1,761  2,961  59.5
 Wellington  3,811  5,161  73.8
 Canterbury  2,832  4,083  69.4
 Age group      
 0–20  4,751  5,714  83.1
 21–40  4,178  5,325  78.5
 41–60  4,722  6,452  73.2
 61+  2,685  5,222  51.4
 Female  8,407  11,729  71.7
 Male  7,929  10,828  73.2
 European  11,937  16,546  72.1
 Māori  1,295  2,059  62.9
 Pacific people  843  1,313  64.2
 Asian  3,113  3,873  80.4
 MELAA  233  288  80.9
 Other  89  123  72.4
Note: Total number of respondents for each demographic isn’t the same because of item non-response, multiple response, and other issues.
MELAA – Middle Eastern / Latin American / African

Results demonstrate we are going in the right direction to reach our online target of 70 percent of individuals completing the online census form in 2018.

Use of online forms by households

A total of 64 percent of households used the online form.

We also compared the differences in online uptake from people if we hand-delivered material to their house and left them with paper forms or if we sent them a letter with an access code. Results showed that providing paper forms to respondents will prevent us from achieving a high online uptake.

The results indicate that an appropriate balance can be struck so that both self-response rates and online uptake targets can be achieved, but further work is needed (see figure 4).

Figure 4

Graph, Online uptake by bype of contact and materials given, Census Test March 2016.

How people completed online forms

We are designing the 2018 Census so that it is easy for people to complete their census forms online by mobile, tablet, or desktop computer. Figure 5 shows which devices respondents used to complete the online form.

Figure 5

Pie chart, Device people used to complete online form, Census Test March 2016.

Figure 6 shows the top five browsers respondents used to submit their online form.

Figure 6

Graph, Top 5 browsers used to submit online form, Census Test (March 2016).

Figure 7 shows the types of mobile devices used by respondents to submit their online forms.

Figure 7

Graph, Mobile device used to submit online form, Census Test March 2016.

Figure 8 shows the top three reasons respondents didn’t complete the forms online.

Figure 8

Graph, Top 3 reasons for not completing form online, Census Test March 2016.

Next steps

We have developed a number of recommendations and have prepared action plans as a result of the 2016 Census Test.

We will incorporate these into our planning for the next Census Test in 2017.

We have also made significant progress in understanding how the new collection model works for our people in the field, particularly the challenges and opportunities of using new technology.

The programme team will also be conducting a series of smaller, independent process and systems tests before the 2018 Census.

For more information regarding the Census Test (March 2016) email

ISBN 978-0-908350-59-9 (online)
Published 29 July 2016

  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+
  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+