Australia’s attitudes and perceptions towards drugs by region, 2019

Definitions

Measures of support and approval

People were asked to indicate how strongly they would support or oppose specific policies, or whether they personally approve or disapprove, using a 5-point scale. Unless otherwise stated, proportions for ‘support’ include people who selected “Support” and “Strongly support” in the survey. Similarly, proportions for ‘oppose’ include responses of “Oppose” and “Strongly oppose”, and proportions for ‘approve’ include responses of “Approve” and “Strongly approve”.

Respondents who selected “Don’t know enough to say” have been removed from the denominator for all proportions. As such, proportions of support or opposition are expressed as proportions of those who believed they knew enough about the statement to express their level of support.

Measures with the largest difference or change

The measures with the largest difference to the National proportion were chosen by examining the difference between the result for that region and the National result, relative to the National proportion.

Similarly, the measure with the largest change since 2010 were chosen by examining the difference between the 2019 proportion and the 2010 proportion, relative to the 2010 proportion.

Geographic data

Household selection process

2019 survey

The survey sample was selected using stratified, multistage random sampling. There were 15 strata in total, including the capital city and ‘rest of state’ for each state and territory, apart from the Australian Capital Territory which operated as 1 stratum.

 For capital city strata, statistical areas level 1 (SA1s) were selected with probability proportional to the number of private households in the stratum. For the first time in 2019, the major regional centres of Illawarra, Newcastle and Lake Macquarie, Geelong, Cairns, Gold Coast, and Sunshine Coast also used this SA1 selection process, to reduce geographical clustering.

In all other areas in the ’rest of state’ strata, statistical areas level 2 (SA2s) were selected for the first stage, with probability proportional to the number of households within the stratum. From within each selected SA2, SA1s were selected with probability proportional to the number of private households calculated in the same way.

2010 survey

While the same 15 strata were used in selecting the sample for the 2010 National Drug Strategy Household Survey, geographic areas were instead defined according to the 2006 Australian Standard Geographical Classification (ASGC):

• For capital city strata, Census Collection Districts (CCDs) were selected with the probability proportional to the number of private households in the district.
• In all other areas, Statistical Local Areas (SLAs) were chosen as the first stage instead, with CCDs selected within each SLA proportional to the number of private households at the 2006 Census.

Geographic presentation in this release

This data visualisation presents data at the smallest geographic area possible. For all jurisdictions except the Australian Capital Territory, this has resulted in data being presented at the SA4 level. Due to the oversampling in the Australian Capital Territory, and the relatively low number of SA3s in the area, data were robust enough to be presented at the SA3 level.

Survey data were assigned to SA4/SA3 in the 2016 ASGS based on where the household was located:

• 2019: As SA1s are entirely located within SA3s and SA4s, survey responses were simply assigned to the relevant geographic area in 2019.
• 2010: As many CCDs cross SA3 and SA4 boundaries, survey responses were assigned to 2011 ASGS SA2s proportionally, based on how much of the CCD was in each SA2. Then 2011 SA2s were assigned proportionally to 2016 SA2s, which were then grouped to the SA3/SA4 level.

All correspondences between SAs (both 2011 and 2016) were sourced from the Australian Bureau of Statistics release 1270.0.55.001 - Australian Statistical Geography Standard (ASGS): Volume 1 - Main Structure and Greater Capital City Statistical Areas, July 2016. Correspondences between CCDs and 2011 SA2s were sourced from 1270.0.55.006 - Australian Statistical Geography Standard (ASGS): Correspondences, July 2011.

Geographic areas with missing or unreliable data

Due to the stratified random sampling approach, no data were collected in some statistical areas. This includes all statistical areas in Other Territories, those used for No Usual Address and those used for Migratory – Offshore – Shipping purposes. Additionally, the following areas had no data collected in 2019:

• Coffs Harbour – Grafton (104)
• Riverina (113)
• Western Australia – Outback (South) (511)
• South East (603)
• Canberra East (80103)
• Uriarra – Namadgi (80111)

Where data are not available for an area, postcodes have been removed from the data visualisation so that they cannot be selected.

Some statistical areas did appear in the survey collection, but the data were considered too unreliable to be reported on their own. This can occur when the sample size is too small, or when very few SA2s were sampled within an SA4 (meaning that results cannot be generalised to the broader SA4).

In these cases, SA4s were grouped together, to ensure that all data were reported where possible. SA4s were grouped together if there was a data quality issue (e.g. small sample size or large sampling error), provided they were adjacent, the sampled SA2s were located close to the neighbouring SA4, and groupings did not cross State and Territory boundaries. The approach was applied for grouping SA3s in the ACT.

Data for the following statistical areas were combined to ensure all data could be presented reliably:

• Far West and Orana (105) and New England and North West (110)
• Hunter Valley exc Newcastle (106) and Newcastle and Lake Macquarie (111)
• Illawarra (107) and Southern Highlands and Shoalhaven (114)
• Ballarat (201) and Warrnambool and South West (217)
• Bendigo (202) and North West (215)
• Darling Downs – Maranoa (307) and Queensland – Outback (315)
• Mackay – Isaac – Whitsunday (312) and Townsville (318)
• Bunbury (501) and Mandurah (502)
• Belconnen (80101) and Molonglo (80110)
• Weston Creek (80108) and Woden Valley (80109)

Postcode selection

The data visualisation tool presents data at the SA4 or SA3 level, depending on where the selected postcode is found. Postcodes were assigned to the geographic region that they had the highest proportion of area in. Postcodes were mapped to SA2s using 1270.0.55.005 - Australian Statistical Geography Standard (ASGS): Volume 5 - Remoteness Structure, July 2016 and then aggregated to the SA3 or SA4 level.   

Postcodes that map to regions where data is not available (e.g., Riverina) and those that do not map to any reported SA4 or SA3 cannot be selected in the data visualisation.

Interviewing in remote Indigenous communities

In 2019, the stratified, multistage random sampling of SA1s across Australia returned 8 remote Indigenous communities. Despite a random sampling technique being applied, by chance, all 8 Indigenous communities selected were located in the Northern Territory – Outback SA4 (702). Residents within those communities primarily spoke Aboriginal and/or Torres Strait Islander languages.

Due to the differences in the sample selection process (detailed in the Technical information from the 2019 NDSHS), and the unique circumstances of each of the communities, the data collected from the 8 remote Indigenous SA1s is not comparable to the data collected from Indigenous people surveyed in non-remote Indigenous communities or the non-Indigenous sample.

As a result, smaller disaggregations, such as results pertaining to the Northern Territory – Outback SA4, are likely to be altered by the inclusion of data from remote Indigenous communities. To preserve comparisons between regions, data from remote Indigenous communities are excluded from results for the Northern Territory – Outback SA4, and state results for the Northern Territory, but are included in National results.

Non-response bias and non-sampling error

Survey estimates are subject to non-sampling errors that can arise from errors in reporting of responses (for example, failure of respondents’ memories, incorrect completion of the survey form), the unwillingness of respondents to reveal their true responses, and higher levels of non-response from certain subgroups of the population.

The estimation methods used for the 2019 results take into account non-response and adjust for any underrepresentation of population subgroups in an effort to reduce non-response bias.

A limitation of the survey is that the data are self-reported. Some behaviours may become less—or more—socially acceptable over time which may lead to an increase in socially desirable responses rather than accurate responses. Any potential changes in self-reported behaviours need to be considered when interpreting survey results over time.

Sampling error

All proportions that are calculated from survey data are estimates rather than true population proportions. This means they have a margin of error due to only a sample of the population being surveyed. This is called sampling error.

There are different ways of measuring sampling error associated with an estimate from a sample survey. The 2019 NDSHS uses both relative standard error and margin of error; these are included in the supplementary tables.

Relative standard error

The standard error (SE) is a measure of the dispersion of estimates calculated from all possible random samples from the same population. This can be estimated using the achieved single sample. The relative standard error (RSE) is the SE expressed as a percentage of the estimate, and provides an indication of the size of the SE relative to the size of the estimate.

Results subject to an RSE of between 25% and 50% should be considered with caution and those with an RSE greater than 50% should be considered unreliable for most practical purposes. Estimates that have an RSE of between 25% and 50% are marked in the supplementary tables with *; those with an RSE between 50% and 90% are marked with ** and those with an RSE greater than 90% have not been published. Only estimates with an RSE of less than 25% are considered sufficiently reliable for most purposes.

Margin of error

The Margin of Error (MoE) describes the distance from the population value that the sample estimate is likely to be within, at the 95% level of confidence. This means that the “true” proportion for the entire population would be within the margin of error around the reported estimate for 95% of possible samples.