School of Population Health

Deprivation and Health Geography within NZ

Data zones


Historically in New Zealand, researchers and policy analysts have conveniently used the geographical boundaries constructed for the national census (i.e. meshblocks and Census Area Units) to report small-area or ‘neighbourhood’ statistics. However, there is considerable variation in the population distributions and this can be problematic for small area research. Many meshblocks are too small to publish statistical analyses reliably and CAUs with large populations may skew distributions and/or results.

In response to the need for a standard neighbourhood level geography, a customised geographical base called data zones was developed by combining Census meshblocks. Of the 46,629 meshblocks in 2013, 640 oceanic or coastal inlet meshblocks were excluded and the remaining 45,989 were used to build 5,958 data zones with a mean population of 712 (approximately 8 meshblocks per data zone). With the exception of one small data zone representing all of Stewart Island (total population of 384) and 10 large data zones with populations between 1,381 and 1,899 (mostly comprising a single meshblock) the population of the data zones ranges from 500 to 1000.

Following consultation with Statistics New Zealand’s geospatial team, the data zones were nested within CAUs wherever possible (93.3%). In addition, the data zones nest within higher geographical units such as GEDs, Territorial Authorities, District Health Boards (DHB) and Regions. One strength of these data zones is that they are independent of the administrative units used by different government agencies (e.g. Police Districts, School Zones) which may change over time and thus represent a neutral geographical basis for facilitating data sharing.

Data zones were not intended to reflect the true extent of actual communities; rather they are an intermediate geography between meshblock and Census Area Units that facilitates small-area analyses of health and social data at a scale small enough to be statistically robust while also conveying a sense of neighbourhood.

Our approach to zone design was semi-automated. We used ArcGIS Districting software, expert local knowledge and visual interpretation, and we considered six key zone design criteria: Compactness, population equality, respecting higher geographical boundaries, respecting physical and social environments, and internal homogeneity. It is not practical or possible to satisfy all six of these criteria in some cases, so priorities for the criteria were set and compromises achieved. Criteria were categorised as either hard constraints or soft constraints, where the former have higher priority than the latter and hard constraints must be met under normal circumstances. When there are conflicts, a lower priority criterion gives way to a higher priority criterion.

Developing intermediate zones for analysing the social geography of Auckland, New Zealand is a 2016 article describing the creation of two new geographical boundary files, known respectively as lower zones and upper zones, for investigating social phenomena and visualising disparities in health outcomes in the Auckland Region. The same methodology was used to develop 5,958 data zones for the whole of New Zealand.