Stats NZ has a new website.

For new releases go to

www.stats.govt.nz

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

Results are presented for each of the four variables in turn. We first assess the coverage of the data sources. For those in the linked Census-IDI dataset, we compare aggregate results, and then show a cross-classification of individual-level responses.

Māori ethnicity

Coverage

While the coverage of administrative sources varies depending on the agency, nearly everyone (98.6 percent) in the linked Census-IDI data has an ethnic code recorded from at least one source in the IDI. While there is some variation by age, sex, and ethnic group, ethnicity information is available for more than 95 percent at every single year of age, and for each main ethnic group.

Aggregate comparison

Table 2 shows the total responses for the Māori ethnic group in the administrative sources and the census. While almost 600,000 people reported being of Māori ethnicity in the census, table 2 includes only records available in the particular administrative source and that were linked to usual residents in the census. For example, of the birth registration records linked to the census, 188,700 are recorded as Māori in the census, and 187,600 have Māori ethnicity in the births data. The final column shows these total responses expressed as a ratio of the administrative source to the census (0.99 in this case) and provides a measure of the consistency of each source compared with the census.

Birth registrations have a ratio of almost 1, indicating very good agreement with the census at an aggregate level. All other single sources have a ratio less than 1, indicating fewer people are stating they are Māori than in the census.

Table 2
Total responses for Māori ethnic group for the census and administrative sources (linked Census-IDI source data only)

2013

Administrative source Total Māori ethnicity responses  Ratio Admin:Census 
Census Administrative source 
DIA Birth registrations  188,700  187,600  0.99 
Health NHI  561,200 440,700  0.79 
MoE Tertiary enrolments  261,400  241,700  0.92 
MoE School enrolments  227,300  202,300  0.89 
MSD  247,900  230,900  0.93 
ACC 227,000  167,300  0.74
Ever-recorded, all sources  569,600  678,400  1.19 
Ranked sources 566,700  522,500  0.92 

To provide ethnicity information for the whole population, we must combine the information from these data sources. The bottom two rows in table 2 show results from applying two methods of combining ethnicity from different administrative sources. The ‘ever-recorded’ ethnicity assigns a person to the Māori ethnic group if they are recorded as Māori in any contributing source, at any time. This is the method used in the IDI Personal Details table in 2015. The total responses ratio of 1.19 shows that the ever-recorded method inflates the Māori ethnicity count compared with the census by nearly 20 percent.

Another approach is to choose only one source of ethnicity for each person, but to choose the highest quality source available. In the ranking used here, birth registrations have the highest ranking, followed by Ministry of Education, Ministry of Health, and lastly Accident Compensation Corporation data. The ratio of 0.92 for this ranking method is closer to 1 than the ever-recorded ratio; however, Māori are still being undercounted compared with the census. Ranking the sources does provide almost full coverage, but it cannot fully make up for the lower number of Māori responses in many sources.

Similar patterns are seen for other ethnic groups when these methods are applied.

Individual-level comparison

The linked Census-IDI data allows us to compare responses at an individual level. Much of the discrepancy between administrative sources and the census is due to the lower reporting of multiple ethnicities in administrative data. Figure 1 shows agreement of the administrative sources with the ethnic profile reported in the census, for single Māori ethnicity or Māori in combination with another ethnic group. Reporting of sole Māori is reasonably consistent across all administrative sources with 80 to 90 percent of those reporting sole Māori in the census also reporting sole Māori in the administrative sources. Only birth registrations achieves a similar consistency for Māori in combination with another ethnic group. Other administrative sources have less than 50 percent reporting Māori in combination with another ethnicity when they do in the census. Similar patterns are found for other level 1 ethnic groups reported singly or in combination.

Figure 2

Graph, Percent agreement with census ethnic profile, by admin source, 2013.

The two methods that bring together all the data sources show opposite results. The ever-recorded method has only 50 percent agreement with the census for sole Māori, but almost 90 percent agreement for Māori combined with another group. In contrast, the ranking method retains the high agreement for sole Māori seen in the individual sources, but still has lower agreement than the census for the combined Māori ethnic group.

Reid and Gleisner (2016) provide more detailed results and for all level 1 ethnic groups.

Māori descent

Coverage

Māori descent information is available from two government sources, electoral enrolments and birth registrations, which provide complementary coverage of adults and children, respectively.

Adults from age 18 years are entitled to vote. The electoral roll’s Māori descent population in 2013 is 93 percent of the Māori descent estimated resident population (ERP) aged 18 years and over. Coverage is lower for younger adults, but from about age 25, electoral numbers hover around 100 percent of the ERP (figure 3). These aggregate results may conceal additional over-coverage and under-coverage compared with the ERP.

Figure 3

Birth registrations provide Māori descent for babies born in New Zealand since 1995. Completion of the Māori descent question is high, with about 0.5 percent missing responses and a further 2.9 percent coded ’don’t know’ or ’unknown’. Children of Māori descent not included are those born overseas to Māori parents who return to live in New Zealand. Of people stating they were of Māori descent in the 2013 Census, 97.9 percent were born in New Zealand and 2.1 percent were born overseas. Birth registrations therefore provide close to full coverage of those with Māori descent born since 1995.

Since 1995 birth registrations have also recorded Māori descent for the parents of these babies, which provides some coverage of the adult population. However, there is no information on Māori descent for children born overseas. While we might assume most are not of Māori descent, we cannot confirm this directly.

Birth registrations before 1995 collected information on the ‘degree of Māori blood’. While unsuitable as a measure of ethnicity, this may have potential as a partial source of descent information. We do not investigate this data here.

Over time, the combination of birth registrations since 1995 and electoral enrolment data is likely to provide high coverage of Māori descent for people living in New Zealand. However, change to the Electoral Act will be required to make full use of the electoral roll in a linked-data environment.

Aggregate comparison

For this analysis we use birth registrations data from 1998, when digitisation was introduced. Registrations from 5 March 1998 are used to coincide with children aged 0 to 14 years in the 2013 Census. Since we use the linked Census-IDI data to compare the individual responses, linkage rates may affect the analysis. The first row of table 3 shows linkage results for all census children 0–14 years, but includes substitute records that cannot be linked to the IDI and have missing values for Māori descent.

Of the 764,600 children aged 0–14 years in the census with non-missing values for Māori descent, 96 percent were linked to the IDI spine and 87 percent (662,000) to their birth registration. Most of those not linked to the New Zealand birth registrations are born overseas, and respond as ‘Māori descent = No’. Just 3 percent stated they were born overseas and with ‘Māori descent = Yes’.

For these children linked to their birth registration, 450,000 mothers and 421,700 fathers are present in the registration data. Parents may have more than one child registered, so we expect to see more children than parents. For parents to be included in this analysis, they must first be linked to the IDI spine. Lower linkage rates in the IDI between birth parents and the IDI spine reduce the effective coverage of the parent information in birth registration records, particularly for mothers. Of all the birth parents, 52 percent of mothers and 72 percent of fathers were linked back to their census record.

We note that this final percentage is calculated using different units: ‘number of individuals’ (linked to the census) are divided by the ‘number of records' (of parents in the birth registrations). The processes used in the IDI mean that mothers and fathers could be present more than once in the birth registrations where they have multiple children, but the parent records were not recognised as belonging to the same individual. Parents linked to the spine and then to the census can be assumed to be unique individuals.

Table 3
Linkage rates and missing data for census records linked to birth registrations
For children, mothers, and fathers, 2013

  Base number  Linked to IDI spine  In census-births linked data 
Number Percent of base  Number Percent of base
Census 0–14 years  865,600  795,300  92  715,600  83 
Census 0–14 years Māori descent Yes/ No  764,600  735,900  96  662,000  87 
Mothers of linked births  450,000  252,500  56  236,000  52 
Fathers of linked births  421,700  357,400  85  304,500  72 

Using the records linked between census and birth registrations as we did for ethnicity, we compare total responses for ‘Māori descent = Yes’ between census and birth registrations. Table 4 shows that at an aggregate level birth registrations are very similar to the census – the ratio of 0.97 for children shows births are slightly under-reporting those with Māori descent when compared with the census. Results are consistent across children, mothers, and fathers.

Table 4
Total responses for ‘Māori descent=Yes’ for census and birth registration records (linked data only)
2013

  Census Birth registrations  Ratio
Births:Census
 
Children  205,000 197,900  0.97 
Mothers  50,900  48,600  0.95 
Fathers  46,700  45,500  0.97

Individual-level comparison

We compare Māori descent responses for birth registrations with 2013 Census responses for the same individuals, now also excluding records with missing responses in the birth registrations. Tables 5a, 5b, and 5c show the cross-classification of ‘yes’ and ‘no’ Māori descent responses, and the percent of the total in each cell.

For the children, 96 percent of individuals provide the same 'yes', or 'no' Māori descent response for births and census. The births-to-census ratio of slightly less than 1 for Māori descent = 'yes' is a result of ‘no’ responses in the births data where the census response was 'yes', being about twice the number that were misclassified in the opposite direction. A similar pattern is seen for mothers and fathers.

Table 5a
Comparison of Māori descent response for children in birth records and census
Counts and percentage of total (linked data only), 2013

Births: children Māori descent response 
Yes  No  Total 
Number % Number Number
Census Māori descent response Yes  177,900 28 17,400  3 195,300  30 
No  8,000 1 439,800 68 447,800 70
Total  185,800 29 457,300 71 643,100 100

Table 5b
Comparison of Māori descent response for mothers in birth records and census
Counts and percentage of total (linked data only), 2013 

Births: mothers Māori descent response 
Yes  No  Total 
Number  % Number Number
Census Māori descent response Yes  45,500 21 3,200 1 48,700 22 
No 1,100 1 169,200 77  170,300 78 
Total  46,600 21  172,400 79  219,000  100 

Table 5c
Comparison of Māori descent response for fathers in birth records and census
Counts and percentage of total (linked data only), 2013

Births: fathers Māori descent response 
Yes  No  Total 
Number  % Number Number
Census Māori descent response Yes  41,000 15 3,600 1 44,600 16
No 1,700 1 231,800 83  233,600 84 
Total  42,800 15 235,400 85  278,200 100 

Iwi affiliation

The starting point for iwi results is the New Zealand residents who answered ‘yes’ to Māori descent in the 2013 Census, and were linked to MoE data through either the school roll returns (schools) or tertiary qualification enrolments (tertiary).

Coverage

Coverage of the MoE information is restricted to younger ages because students are usually children or younger adults, and because of the fairly recent start to data collection. The schools data used in this analysis is mainly for those aged 5 to 25 years in 2013; tertiary data are mainly for people aged 18 years and over, with most individuals being under 40 years.

The census obtains valid iwi responses from 80 percent of those indicating they are of Māori descent. The MoE response rates for iwi are lower, with 68 percent of tertiary data providing a valid iwi when they identified as Māori descent in the census, and just 34 percent for schools (table 6).

Table 6
Valid iwi responses in the census and linked Ministry of Education data
2013

Base population group  Age range covered  Māori descent in Census  Valid iwi responses 
Number  Percent
Census usual residents  All ages  668,700  535,900  80
Census linked to IDI  All ages  640,000  513,600  80 
MoE schools linked to census  5–25 years  253,100  86,700  34 
MoE tertiary linked to census  18 years + 291,100  198,500 68

The combination of coverage limitations and missing data mean that combining MoE data from either source provides valid iwi responses for 42 percent of the 2013 Census Māori descent population.

Aggregate comparison

As with ethnicity, people provide multiple responses for iwi. Table 7 shows the ratio of total responses for an iwi in the schools and tertiary education sources compared with the census.

Because the classification has 128 iwi, we present results for the 12 largest iwi only. The first column shows the total responses from the full 2013 Census dataset for comparison. The ‘Linked Census iwi responses’ columns show the total number of responses received for each iwi in the census, restricted to individuals who linked to the education datasets. The schools and tertiary iwi responses columns are the number of responses for that iwi in the education data linked to the census. The ratio is the education total divided by the linked census total. As before, a ratio close to 1 indicates high consistency between the census and education data.

Note: the education datasets in the IDI can contain multiple records for an individual, corresponding to distinct enrolments across different institutions or years. Iwi information in each record can vary. For this paper, we use the record containing the most iwi responses, and have not attempted to reconcile or combine responses from different records.

Table 7
Total responses for 12 largest iwi
Census and Ministry of Education (linked data only)

Iwi name   2013 Census  Schools Tertiary
Total iwi responses  Linked Census iwi responses Iwi responses Ratio admin:census Linked Census iwi responses Iwi responses Ratio admin:census
Ngāpuhi 125,600  49,900  22,300  0.45  53,800  46,600  0.87
Ngāti Porou  71,000  28,300  12,600  0.44  32,500  29,800  0.92 
Ngāi Tahu 54,800  20,900 6,300  0.30  24,000  16,300 0.68 
Waikato  40,100  16,000  1,700  0.11  18,000  5,900  0.32 
Ngāti Tuwharetoa  35,900  14,400  5,200  0.36  15,900  12,400  0.78
Ngāti Maniapoto 35,400  13,800  4,700  0.34  15,900 11,500  0.72 
Tuhoe  34,900  14,400  7,000  0.49  15,200  13,900  0.91
Ngāti Kahungunu ki Te Wairoa  21,100  8,200 800  0.10  10,000  3,600 0.36 
Te Arawa  19,700  8,000 5,000  0.63  8,900  11,100  1.24 
Ngāti Kahungunu, reg unspecified  18,300 7,500  7,000  0.94 7,900  14,100  1.78 
Te Rawara  16,500  6,200  1,900  0.30  7,800  5,400  0.69 
Ngāti Awa  16,200  6,300  2,800  0.45  7,600  6,600  0.87 

All but two of the ratios are less than 1, indicating that the education data consistently undercounts iwi affiliation when compared with the census. Ratios for schools data, being mainly less than 0.5, are lower than tertiary institutions – where most are between 0.7 and 0.9. There is also considerable variation among iwi. Tuhoe, Ngāti Porou, Ngāpuhi, and Ngāti Awa show ratios closest to 1 in both schools and tertiary data, while Waikato has the lowest ratio in both sources. Differences in people’s responses and coding practices are likely to be affecting results for the two Ngāti Kahungunu categories.

We note that MoE reports their results on iwi regional groupings, rather than individual iwi. This may improve some results, particularly where there are coding difficulties.

Individual-level comparison

For people who provide an iwi in the education data, we can also look at how consistent the reported iwi affiliation is with the iwi recorded for the same person in the census. Results in figure 3 are for the 12 largest iwi. Of individuals who reported a specific iwi in the census, 20 to 40 percent reported belonging to the same iwi in the schools data. The tertiary results are better, with 30 to 60 percent belonging to the same iwi as stated in the census. The same four iwi perform best in this individual-level analysis, and Waikato again shows the lowest consistency between census and MoE data.

Figure 4

Summary

Table 8 summarises results by coverage for the agency, and quality of the data collected for each variable. The high, medium, or low quality ratings are somewhat subjective, but are an attempt to provide a reasonable assessment of the findings reported here.

Table 8
Summary of quality measures for administrative sources for four essential variables

Source  Coverage of source Quality of administrative responses 
Māori ethnicity Māori descent  Iwi  Te reo 
DIA
Registered births
  • 0–14 years
  • All NZ born 
HIGH HIGH 
Ministry of Health 
  • All ages
  • Most NZ residents 
MEDIUM … 
MoE schools 
  • 5–25 years
  • Most NZ residents 
MEDIUM  …  LOW  LOW 
MoE tertiary
  • 18 to under 40
  • Students 
MEDIUM  …  MEDIUM  LOW 
Ministry of Social Development 
  • 15 years +
  • Working-age benefit receipt 
MEDIUM  …  … 
ACC 
  • 15 years +
  • Injury claimants 
LOW  …  … 
Symbol: ... not applicable
  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+
Top
  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+