ABOUT THIS RELEASE
This paper was presented to the Australian Population Association Conference, 15 - 17 September 2004.
The objective of the Census collection operation is to achieve a high quality census count that obtains maximum coverage of the population in a cost-effective manner.
The cost of coding the "Main Language Other than English Spoken at Home" topic in the 2006 Census and alternate censuses could be reduced if it can be shown that modelling language data from other census responses can provide detailed language data of an acceptable reliability.
This paper presents the results from multivariate modelling analysis of non-English languages. The research aimed to construct a predictive model of languages spoken using other census responses such as Ancestry, Birthplace, and Religious Affiliation; to estimate the number of people who speak a language other than English at home in each statistical local area (SLA), using the model above; and to evaluate the models and assess the feasibility of reducing the language question.
This study finds that ancestry, birthplace and religous affiliation are reasonably good predictors of language. The predictions from the models for SLAs were reasonably accurate for most languages selected.