3405.0.55.001 - Discussion Paper: Assessment of Methods for Developing Experimental Historical Estimates for Regional Internal Migration, Dec 2011
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 22/12/2011 First Issue
Page tools:
 Print Page
 Print All
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
The Census-based internal migration data have been adjusted in the same way that Census population counts are adjusted to create the Estimated Resident Population (ERP). More detail is provided in the explanatory notes of the ABS publication Regional Population Growth, Australia (cat. no. 3218.0). Arrivals and departures A table of the number of arrivals and departures by Statistical Division is available in Appendix 4. These estimates are SLA movements aggregated to SDs and show the difference between the adjusted 2005/06 Medicare arrivals/departures and the 2006 Census-based internal migration estimates. Overall, there was a small percentage difference in most SDs and the total difference was 7 per cent of arrivals and 6 per cent of departures. Geographic level The finest level of detail available for dissemination of the data is by SLA. However, the limitations of converting Medicare data at the postcode level to SLA have already been outlined and discussed in this paper. If required, it would also be possible to release the data at broader regions (aggregated from SLA), eg. Statistical Division (SD) or Statistical Sub-division (SSD). Listings and descriptions of all ASGC regions in Australia are available in the ABS publication Australian Standard Geographical Classification (ASGC) (cat. no. 1216.0). A comparison of the 2005-06 Medicare data and the 2006 Census data by SD shows that there are a larger number of movers in the Medicare data in most capital city SDs, apart from the Brisbane SD and the Canberra SD. This may be due to multiple movers in the Medicare data because they are collated on a quarterly basis and could potentially capture up to four moves per individual, whereas the Census data relate to comparing current address of usual residence with that of one year earlier and hence capturing one move per person. Age/sex analysis This paper has only considered experimental estimates of internal migration of the total population. However, based on the available data, these migration figures can be broken down by age and sex, although it is not anticipated that this level of data will be released at this stage due to potential quality concerns. Analysis of the internal migration data by age and sex in a time series can indicate whether the age and/or sex structure of migrants is changing for regions. The following graph shows Medicare arrivals as a ratio to Estimated Resident Population by age group. In 2006, the overall ratio of movers in Australia aged between 15 and 44 years was higher than the ratio aged 45 and above. Some reasons that younger people are more mobile may be related to employment, for example, they may be more likely to move locations for employment or education-related opportunities. A slightly higher ratio of females than males aged 15 to 34 years moved location. As an example, in the Adelaide (C) SLA in 2005-06, the proportion of Medicare arrivals was quite low for children aged 0-14 years (9%). This also corresponds with the adult data showing that there is a lower proportion of people aged 35-44 (14%), who could potentially be parents of children in this age group. The higher proportions of arrivals aged 15-24 (22%) and 25-34 (28%) may be partly due to the high numbers of university students living in this SLA. This SLA is largely high density living including many apartment buildings, townhouses and row cottages. There are a higher proportion of male arrivals in the age groups most likely to be in the workforce, including those aged 25-34 (29% of males, 27% of females), aged 35-44 (16% of males, 11% of females) and aged 45-64 (23% of males, 21% of females). On the other hand, there was a higher proportion of female arrivals aged 15-24 (24% of females, 20% of males). In 2005-06, in the Mundaring (S) SLA, east of Perth, the proportion of Medicare departures was highest for young adults aged 15-24 (24%), 25-34 (20%) and children aged 0-14 (19%). The proportion of male departures in the 25-34 year age group was higher (21%) than the proportion of females in the same age bracket (18%). Time series SLA level data are available from 1985-86 to 2005-06 based on the 2006 ASGC. It would be possible to perform analysis on the historical numbers and validate against previous Census numbers, however, this would be quite a large undertaking as there are several years of data to look at. A full investigation of the time series data could be the focus of a separate project. A full time series from 1985-86 to 2009-10 can be extracted at the Australia level or the state/territory level, however more correspondence work would be required to calculate the SLA data based on 2006 ASGC for the years beyond 2005-06. The Australian totals for the Medicare estimates are illustrated in the graph below. Missed Movers Index (MMI) Based on the 2006 Census, it is estimated that the postcode-based Medicare data are missing around 75,000 inter-SLA moves across the country, or 3.6 per cent of all inter-SLA moves in 2005-06. By state, this ranges from 1.3 per cent missed arrivals in New South Wales, up to 11.6 per cent for Australian Capital Territory. The MMI is higher for the Australian Capital Territory, Northern Territory and Queensland because these states and territories have several small SLAs. Where there are larger postcode boundaries than SLA boundaries, the moves between small SLAs within the same postcode are more likely to be missed. The number of missed movers can be reduced by applying the MMI, which is outlined in Appendix 2.
Concordance Confidence Index (CCI) The quality of postcode-based data which has been converted to ASGC using a correspondence can be assessed using a Concordance Confidence Index (CCI). This index provides an indication of how closely the postcode-based data aligns with the SLA. More information about the CCI calculations is provided in Appendix 3. The postcode-based Medicare data can be improved by grouping SLAs with a low confidence index with neighbouring SLAs. For example, Northern Territory has a very low confidence index (25.8%), but this can be improved markedly (by 49.1%) by grouping selected SLAs.
(a) The overall CCI has been calculated for each state/territory using individual SLA confidence indexes. (b) The overall CCI has been recalculated for each state/territory after some SLAs were combined with neighbouring SLAs. Reconciling to state/territory estimates The final step in creating a set of experimental regional migration estimates is to ensure that the sum of estimated arrivals and departures between the states and territories are constrained to the estimates of interstate migration as published by the ABS in Australian Demographic Statistics (cat. no. 3101.0) and Migration, Australia (cat. no. 3412.0), using an Iterative Proportional Fitting (IPF) procedure. Data confidentiality To ensure data confidentiality, the geographic level to be released in outputs will need to be investigated in addition to whether the data can be cross-tabulated with any other variables, e.g. age and sex breakdowns. Before release of any tables of estimates, standard ABS confidentiality procedures have been followed to ensure the confidentiality of individual unit records. Document Selection These documents will be presented in a new window.
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
