CAUTION - The data presented are experimental estimates drawn from a de-identified linked Settlement Database_Census sample file. The following points must be considered when analysing the data:
The proportions are affected by both sampling and non-sampling error.
Proportions with a higher level of sampling error have been flagged.
Non-sampling error has resulted primarily from error associated with the linking process and may occur due to false or missed links. For the estimates presented in these tables, the major source of non-sampling error is from missed (unlinked) records. In particular, approximately 34% of the SDB records were missed (or unlinked), and thus in producing proportions from the linked data, it is assumed that relationships between variables observed in the linked data, are the same for those in the unlinked data. In an attempt to mitigate potential bias, the records that were linked were calibrated (weighted) to sum to known totals from the Settlements Database, however it is unlikely that this fully accounts for bias introduced by missed links (unlinked records). Given that this source of non-sampling error is very difficult to quantify, cells affected by this non-sampling error are not specifically flagged in these tables of estimates. Therefore, in assessing the reliability of these estimates, a user should consider this general caution in regards to non-sampling error, in conjunction with the standard flags used to highlight estimates with a higher level of sampling error.
To gain further insight into this issue of non-sampling error, a user of the data should read Section 5 'Evaluation of the Linkage' in Research Paper: Assessing the Quality of Linking Migrant Settlement Records to Census Data, August 2009 (cat. no. 1351.0.55.027).