DRAFT
relating to health and illness, health risks and health care utilisation together with
background information including social and demographic characteristics. The 2002 and
2004 surveys included a breast cancer screening module.
Weighting
In each year’s data collection, the survey samples 1,000 individuals from each of the
seventeen local health regions, referred to as Area Health Services (AHS) in NSW3, as
shown in Figure 1. Due to different population sizes, residents in smaller AHS have a
higher probability of selection than those residing in more populated AHS. The use of
CATI may impact on a household’s probability of being sampled, depending on the
number of telephone lines into the home. Finally, household size also affects sampling
probability, with individuals in larger households having a smaller probability of
selection than those in smaller households.
To compensate for disparities in the probabilities of being selected, the survey sample is
weighted according to the number of eligible respondents in the household, the number of
residential telephone lines for the household and a sampling fraction in each AHS. Using
a weighted sample in analyses of data produces results that are representative of the
population in both rural and metropolitan NSW. All results reported in this paper were
derived from the weighted sample.
Dependent variable
The 2002 and 2004 surveys asked women aged 40 to 79 questions about (1) whether they
had ever had a mammogram; (2) time since their last mammogram; and (3) the reason for
the last mammogram. Responses to these questions were used to create a dependent
categorical variable indicating screening behaviours; these have been named “never
screeners”, “regular screeners” and “irregular screeners”. In line with BreastScreen
Australia benchmarks, women who responded ‘no’ when asked whether they had ever
had a mammogram were categorized as ‘never screeners’; those who indicated that the
3 In 2004, Area Health Services were amalgamated to form eight local regions; for the purposes of this
analysis we have persisted with the old Area boundaries.