Part 5 : Matters for public entities to consider

Using the United Nations' Madrid indicators to better understand our ageing population
… social and demographic changes have profound implications for policy makers, planners, administrators, and others interested in population ageing.
Statistics New Zealand

In this Part, we discuss some matters that we would like public entities to consider and, as needed, respond to. They are:

It may be possible to analyse existing indicators by rural and urban status to measure rural disadvantage. Bringing together data by age is more difficult when inconsistent age groups are used. We found that survey and sample sizes were sometimes too small to produce statistically reliable profiles for older people by sex and ethnicity, but that more use could be made of other data to fill in some gaps. We were able to bring together a report on the Madrid outcome indicators but the "shelf life" of data was not always clear.

Statistics New Zealand has told us that it is currently represented on an international task force to improve the availability, accessibility, and comparability of statistical data to support ageing-related policy-making. Addressing the issues that we discuss in this Part would also contribute to improved comparability, analysis, transparency, and cost-effective use of data.

Measuring rural disadvantage

The Ministry of Social Development reports that the data needed to create an indicator of disadvantage for rural people for the Positive Ageing Strategy is not available. With 86% of the population in urban areas, there are problems in ensuring the same access to services for the other 14%. We consider it unlikely that a useful single measure of rural disadvantage for older people is possible. Entities agreed with our view in their feedback on our draft report.

The question of whether differences in access to services between urban and rural communities have negative effects on rural people is important. Statistics New Zealand told us that there might be opportunities to analyse existing indicators of older people by rural and urban status to get a picture of how the circumstances of older people in rural areas compare to their urban counterparts.

If public entities have not already done so, we encourage them to consider how they can assess whether differences in access to services have any negative effects on rural people, and consider what can be done to lessen those effects.8

Defining age groups consistently

In our report, we have defined older people as people aged 65 years or older. In some reports, older people were considered to be aged 15 or more; in others, 40 or more; and in others, 65 or more. We understand that some flexibility is useful, but greater consistency could be even more useful.

Data tables were not always published, which made it difficult for us to reorganise the data to make the age groups more similar. This made it more difficult for us to report consistently on the Madrid indicators and makes it more difficult for anyone else to do something similar for other purposes.

Statistics New Zealand has published a statistical standard for age on its website, which includes recommended categories for reporting on age.9 We encourage public entities to use the standard in collecting and reporting data.

Statistics New Zealand told us that there is an increasing shift towards making official statistics available from online databases. This gives users flexibility to get reports on categories of data, such as age group, to meet their needs. We encourage other public entities to consider whether they could also make their statistical data available online.

Choosing sample sizes for national surveys

There is often a tension between the cost of not having data, having poor quality data, getting a large enough sample to produce reliable statistics, and getting statistics updated in a timely way. These tensions can be amplified in a country with a small population that is unevenly and widely dispersed and during times of economic restraint.

We tried to report on the Madrid indicators by age group, sex, and ethnicity. Most of the time, it was not possible to do this because sample sizes were not large enough for the older age group. From our discussions with public entities, there are no easy solutions to this issue and the cost-benefit of some potential solutions is questionable.

We discussed with public entities whether more use could be made of the census to complement the quantitative data that surveys collect, and whether sample sizes for some surveys could be increased. However, Statistics New Zealand told us that the response rate to household surveys has decreased over recent years. Imposing extra burden on respondents would be likely to increase this trend and adversely affect the quality of survey data.

Partly in response to this trend, Statistics New Zealand is considering how it might transform the census from the current model (where respondents are asked a range of questions about themselves and their dwelling) to a model where census information is obtained, as far as possible, from administrative sources.10 Statistics New Zealand is investigating alternative approaches and is planning to give its first progress report to the Government in late 2013.

Statistics New Zealand also told us that public entities have other methods potentially available to them for meeting the demand for more reliable data on older people. The options include:

  • making more use of administrative data;
  • combining data from several instances of the same survey (data pooling);
  • combining data from different sources and surveys (data linking); and
  • using special statistical techniques (such as small domain estimation techniques) to get more value from existing data.

Statistics New Zealand is placing emphasis on making more use of existing data and is working closely with public entities to achieve this.

Different reporting periods

In publishing our findings for each indicator online, we have brought together a range of data from a range of public entities at We hope that people will use the findings to compare updated data. More recent data is available for some of the outcome indicators, and data from the 2013 census will update many more.11

We were not able to identify trends for some indicators because data had not been collected often enough to make comparisons. In other cases, data collection practices had changed, which made it difficult to compare past and current data. Nevertheless, the findings give us a perspective on older people's circumstances and we briefly summarise the findings from the Madrid outcome indicators in Part 8.

Even though the data is reported for different periods, Statistics New Zealand told us that this is not necessarily a problem and it is usual practice to do this. The frequency with which surveys are run depends on how susceptible the phenomenon in question is to change. Some phenomena, such as disability status, change more slowly. Others, such as labour force participation, change quickly and are reported more often. The most recent population data available to us was from the 2006 census.

However, the "shelf life" of the data was not obvious to us when we looked for it. On their websites, public entities often state when the next survey is due and when data from it will be reported. We would have found it helpful if the reason for the frequency for collecting and reporting data was explained.

8: For example, in August 2010, we reported on whether district health boards had been able to ensure that primary health after-hours services were available within 60 minutes' drive for 95% of a district health board's population. Our report, District health boards: Availability and accessibility of after-hours services, is available at

9: The standard is available from under the Surveys and Methods tab. See section 4 (output) of the Age standard for options on reporting age.

10: Administrative data is the data that public entities collect routinely, such as crime statistics, projections based on survey data, data on hospital stays, and electoral roll data. It is not collected using a survey or census.

11: The schedule for releasing reports on the census data is available at the Statistics New Zealand website,

page top