Appendix 2: How we analysed the data

Our intention was to get most of the data we needed from published sources. This was partly to reduce the workload on public entities, and partly to check whether the information was accessible to the public. We assessed "accessibility" in two ways: could we find the information, and could we understand the information? Sometimes we asked public entities to give us data when we could not find it online. On other occasions, we did not do this but said in our indicator report that the data was available from the public entity.

We adapted the Madrid indicators to collect data by ethnicity as well as age and sex. However, it was not always possible to report on an indicator by age, sex, and ethnicity because sample sizes were too small to produce reliable statistics for two or three variables at a time. For example, it was common for data to be reported by ethnicity alone or by ethnicity and sex. We found it was less common for data to be reported by age and ethnicity together or by age, sex, and ethnicity together.

In the findings already published online, we rounded numbers and percentages to no more than two decimal points. We published the supporting data for each graph and gave references to any published sources so that people could find the data we used and any extra information they might be interested in.

One of our initial objectives for this work was to find out how public entities used the data that was collected. We were not able to answer this question easily for most Madrid indicators because sometimes public entities collect data for their own use and at other times they collect data for use by others. Each indicator finding published online has a brief comment about why the data was collected and its use. As part of its work to advise the Government on options for collecting official statistics (see Part 5), Statistics New Zealand told us that it is doing work to find out how census data is used and by whom.