Community attitudes survey 2018: Methodology
Our 2018 community attitudes survey was conducted online with 1235 Victorians aged 18 years and older. It ran from 23 April to 16 May 2018.
Survey participants were drawn from an Online Research Unit database of around 55,000 Victorians. This database reflects Australian Bureau of Statistics (ABS) estimates for age, gender and location (metropolitan and regional) in Victoria. However, it has an over-representation of young people and an under-representation of people aged 70 years and older.
People on the database were randomly invited to participate in the survey, and quotas were set for age, gender and location. Invitation rates were adjusted over time to ensure a balanced response. Because so few people in the population experience problem gambling, this group and others who experience gambling harm were deliberately oversampled to allow for comparative analysis.
In total, 10,711 email invitations were sent to potential participants. Screening questions were used to screen out people who work for government, social research or advertising, non-Australian citizens and people aged under 18. There were 3243 responses to the screen questions. Once data collection was completed, responses were examined further to assess whether respondents had considered each question or whether they had answered randomly in order to receive a reward for participation. Following this, there was a total of 1235 valid responses. This equates to a response rate of 11.5 per cent.
Once collected, the final data was weighted to ABS figures for age, gender and location, to reflect the Victorian population. Data was also weighted to the prevalence of at-risk and problem gambling, based on findings from our 2014 Study of gambling and health in Victoria.
There are several limitations to research using online panels for participant recruitment. The most significant limitation is that participants are not selected through probability-based sampling methods, which give demographic groups equal probability of being selected. However, by using quotas and weighting the data to ABS figures for age, gender and location, it is assumed the results would be similar to those of an equivalent sample of the general population.
Online research panels also typically report a higher prevalence of problem gambling than would occur in the general population. In the current study, this has been accounted for by weighting the results to the prevalence of at-risk and problem gambling reported in our 2014 Study of gambling and health in Victoria.
Lastly, other research has shown that people tend to underestimate how much they spend on gambling. In particular, people who report higher losses, or use high frequency gambling products, are less likely to provide accurate estimates of their gambling expenditure. However, people tend to provide more accurate responses when asked about short-term losses (last three months) compared to long-term losses (last 12 months). For more on this, see Self-reported losses versus actual losses in online gambling: an empirical study. The data on gambling losses presented in the survey are likely to be an underestimate, particularly for those experiencing harm from their gambling.
Based on the survey, find out what Victorians think about gambling.