4.31. Comprehensive statistical records on overseas arrivals and departures have been maintained since the colonial days. This has been made possible by the relative isolation of Australia, the absence of direct land links with other countries, and the limited number of ports of entry. These natural advantages have been reinforced by government control of arrivals and departures.
4.32. Early migration statistics were derived from passenger lists (manifests) which masters of ships were required to submit to the authorities on arrival or departure from Australia. Migration statistics were published in varying detail in Statistical Registers or Year Books of the Colonies, and after 1901 also in the Commonwealth Year Book and statistical bulletins.
4.33. With the advent of air travel, people arriving in or departing from Australia lodged a passenger card containing information identical to that previously provided on shipping manifests.
4.34. In 1965, the use of shipping manifests for statistical purposes was abandoned. The control of all passenger movement became the responsibility of the immigration authorities and incoming and outgoing passenger cards were prescribed for use by all sea and air passengers. From January 1974 responsibility for processing passenger cards was transferred from the ABS to the Department of Immigration and Multicultural Affairs as they are now known.
4.35. More information on the history of overseas migration statistics is contained in 'Immigration and Ethnicity' by Charles A. Price and 'Trans-Tasman Migration: Trends, Causes and Consequences' edited by Gordon A. Carmichael.
4.36. Since 1 July 1998, information has also been collected from alternate sources and matched to the corresponding passenger cards. As such, this information is no longer requested on passenger cards. Information on age, sex and country of birth is obtained through matching to passport and visa data, while information on time spent in Australia (for visitors and temporary entrants) and time spent abroad (for Australia residents) is calculated through comparison of arrival and departure dates which are accessed through the movements database, maintained by the Department of Immigration and Multicultural Affairs. Information on marital status is only available from some visa data and, as such, is not available at all for Australian and New Zealand citizens.
4.37. Information on intended duration of stay, for people commencing their journey, and actual duration at the completion is used for the classification of movements into three categories: short-term (less than twelve months), long-term (twelve months or more) and permanent. For national and State population estimation purposes, permanent and long-term arrivals are added into the population and permanent and long-term departures are subtracted.
4.38. There are five possible causes of error in the overseas migration data which can contribute to inaccuracies in population estimates:
- Mis-statement of the State of usual residence by permanent and long-term departures.
- Mis-statement of State of intended residence by permanent and long-term arrivals.
- Errors in the estimates of 'category jumping' (see Appendix 3).
- Failures to match passenger cards to other data sources.
- Errors from unrecorded arrivals and departures.
Mis-statement of the State of usual residence by permanent and long-term departures
4.39. The extent of this type of error is believed to be small because such people are, or have been permanent residents and can therefore be expected to report their address correctly.
Mis-statement of State of intended residence by permanent and long-term arrivals
4.40. The magnitude of this type of error is illustrated in Table 4.8 which compares the distribution by State of permanent and long-term arrivals for 1980-81, 1985-86, 1990-91 and 1995-96 with the census distribution of Australian residents whose usual residence one year ago was overseas.
4.41. Apart from the disparity between the definition of permanent and long-term residence and 'usual residence' in the Census, the only differences which may affect comparability between these two data sources arise from:
- the possibility that permanent and long-term arrivals in the year prior to the census may depart or die before the census is taken. The numbers of such cases are unlikely to be large enough to affect comparisons; and
- the potential disparity between the definition of 'intended address' on the passenger card and 'place of usual residence' on the census form. There are no data on the effects of this disparity though the potential for error is greatest for settlers and visitors since they are more likely to change address than are Australian residents returning; and
4.42. Notwithstanding these differences, the comparison in Table 4.8 illustrates important differences between the two distributions. Overseas migration statistics show New South Wales and Victoria having a larger share of permanent and long-term arrivals than is shown in census data, while most other States have a smaller proportion than that shown by the census. Queensland in particular shows a significantly higher proportion of permanent and long-term arrivals in the census as compared with passenger card data. This indicates that there is a proportion of migrants who record New South Wales and Victoria as their State of arrival on the passenger card, but move interstate during their first year in Australia.
4.43. These differences between overseas migration data and census data do not affect population estimates as long as the interstate migration estimates reflect the interstate moves of migrants in their first year in Australia. However, accurate data are not available to measure the extent to which interstate migration estimates cover the movements of migrants within their first year of settlement.
- interstate migration subsequent to arriving in Australia.
Errors in the estimates of 'category jumping'
4.44. 'Category jumping' arises, in the context of overseas migration statistics, when the duration of a person's journey differs from that originally indicated (on the arrival/departure card at the beginning of the journey) in such a way as to affect his/her categorisation. For example, an Australian resident departing for a short-term visit overseas (with a stated intention to stay abroad for less than twelve months) who in fact stays more than twelve months, thereby changing from short-term to long-term. Changes such as this (ie. between short-term and other categories and vice versa) would lead to errors in post-censal updates of the population. To avoid these errors an adjustment is made to net permanent and long-term migration, when updating population estimates, to compensate for category jumping. However this adjustment cannot be compiled from individual records because of the difficulties attempting to compare arrival and departure records for each individual traveller and because of the complications of lags arising from incomplete journeys. Category jumping must therefore be estimated and is consequently subject to a degree of error. For an outline of the method used in estimating the extent of category jumping see Appendix 3.
4.45. As category jumping estimates are based on short-term overseas arrivals and departures data, which are derived from a sample, sampling error must be taken into account when assessing the extent of error in the estimates. The sample size, however, is sufficiently large that any contribution arising from sampling error to errors in the estimate of category jumping can be expected to be small. For example, estimates of 10,000 or more have a relative standard error of about 6 per cent.
Failures to match passenger cards to other data sources
4.46. As detailed in 4.36, data on age, sex and country of birth are obtained through matching to passport or visa records. This matching is done on the basis of the passport number and the citizenship and is done for all fully keyed passenger cards (see Appendix 3). In cases where no match is possible, usually due to an incorrect passport number, no information on age, sex and country of birth is available for that movement and an alternative passenger card of the same citizenship and category of travel is fully keyed. On average, 3,000 to 3,500 alternative passenger cards are selected each month.
4.47. Similarly, no information on actual length of stay in Australia (for visitors) or actual time spent abroad (for Australian residents) is available if it is not possible to match the arrival and departure dates. This is estimated to occur for less than 1% of all fully keyed passenger cards.
4.48. Problems can also occur due to non-response to the category of travel (ie. migrating permanently to Australia, visitor or temporary resident, resident returning to Australia for arrival cards and visitor or temporary resident departing, Australian resident departing temporarily and Australian resident departing permanently for departure cards). In such cases, a category of travel is imputed on the basis of citizenship, migration visa, intended length of stay and whether they intend to live in Australia for the next 12 months (which is asked on the front of the arrival card). Non-response to category of travel is estimated to be higher on the arrival cards (up to 1.5%) compared to the departure card (up to 0.7%).
Errors from unrecorded arrivals and departures
4.49. These are movements which have not been recorded by any immigration control mechanism and are different from 'category jumping' identified in paragraph 4.44. The numbers are believed to be insignificant although it is not possible to fully quantify them.