Criteria for selecting indicators

Our selection criteria help us to derive a balanced and manageable set of indicators from the mass of statistics available. Indicators for the social report have been selected against the following criteria:

• relevant to the social outcome of interest – the indicator should be the most accurate statistic for measuring both the level and extent of change in the social outcome of interest, and it should adequately reflect what it is intended to measure (ie it should be valid)
• based on broad support – there should be wide support for the indicators chosen so they report on a broadly shared understanding of wellbeing
• grounded in research – there should be sound evidence on key influences and factors affecting outcomes
• able to be disaggregated – ideally, it should be possible to break the data down by age, sex, socio-economic status, ethnicity, family or household type and region, so we can compare outcomes for different population groups
• consistent over time – the indicator should be able to be defined and measured consistently over time to enable the accurate monitoring of trends
• statistically sound – the indicator uses high-quality data and the method used to construct it is statistically robust
• timely – data should be collected and reported regularly to ensure indicators are providing up-to-date information
• nationally significant – the indicator reflects progress at a national level and is not confined to particular areas
• internationally comparable – as well as reflecting the social goals of New Zealanders, indicators should be consistent with those used in international monitoring programmes so we can make comparisons.

Trade-offs between these criteria are sometimes required. For example, it may be necessary to choose an indicator where data is produced at long intervals to ensure a consistent time series is available.

In some outcome domains, such as Health, there is an abundance of good data from which to draw appropriate indicators. In other outcome domains, such as Cultural Identity and Leisure and Recreation, there is less good-quality, relevant data available, resulting in fewer indicators in these domains.