Gambling activity in Australia

Findings from wave 15 of the Household, Income and Labour Dynamics in Australia (HILDA) Survey
Research Report – November 2017

1. Introduction

Background

Gambling is a common activity in Australia. Most adults participate at least once a year. The most recent gambling surveys put the national annual participation rate at around 64% (Dowling et al., 2016; Hing et al., 2014) and between 55-74% across the states and territories (Office of Economic and Statistical Research, 2012; Davidson et al., 2016; The Social Research Centre, 2013; Stevens et al., 2017; Sproston et al., 2012; ACIL Allen Consulting et al., 2014; Hare, 2015).

The gambling activities that Australians prefer are changing. Compared to two decades ago, far fewer now participate in activities that emphasise chance, including lotteries, scratch tickets, keno and Electronic Gaming Machines (EGMs). Much greater numbers now participate in activities that emphasise skill and experience in predicting the outcome, including some casino table games, horse and dog racing and especially sports (Armstrong et al. 2017).

The amount spent is substantial. In 2014/15, Australians gambled $191 billion and lost $22.7 billion across the range of available activities. This equates to $1,242 spent for every person aged 18 and over (Queensland Government Statistician's Office, 2016).

It means that, per capita, Australian adults are the largest spenders on gambling in the world, at around double the average of other Western countries (The Economist online, 2014, 2011, 2017).

Many see and experience gambling as a form of leisure and recreation. However, gambling can have serious repercussions for individuals, their families and society as a whole (Abbott et al., 2015). Between 5-12% of Australians are estimated to experience one or more gambling problems annually (Dowling et al., 2015; Hing et al., 2014), with rates ranging similarly widely across the states and territories (Office of Economic and Statistical Research, 2012; Davidson et al., 2016; The Social Research Centre, 2013; Stevens et al., 2017; Sproston et al., 2012; ACIL Allen Consulting et al., 2014; Hare, 2015). The burden of harm associated with these problems has been estimated to be of a similar magnitude to depressive disorder and alcohol misuse and dependence (Browne et al., 2016).

While numerous cross-sectional state/territory and some national gambling surveys have been performed, there was until recently no ongoing population-level survey of Australian gambling activity and the socio-economic characteristics of gamblers and their families. Longitudinal data with which to measure change over time in gambling activity and effects on individuals and families was similarly absent.

Inclusion of a gambling activity module within the Household, Income and Labour Dynamics in Australia survey (HILDA) is intended to address this need. HILDA is a nationally representative longitudinal panel study of Australian households which commenced in 2001. It provides data on a wide range of aspects of life around family dynamics, economic and subjective well-being and labour market dynamics.

Gambling questions were included for the first time in wave 15, connecting gambling activity in 2015 to these broad areas of life. The module comprises two components. The first measures the amount of expenditure on 10 different gambling activities during a "typical month"1. The focus on a "typical month" ensures the focus is on regular gambling, and thus effectively excludes the occasional or once-off expenditure.

The second component comprises the Problem Gambling Severity Index (PGSI; Ferris & Wynne, 2001) This includes nine questions used to identify people whose gambling behaviour caused them problems or put them at risk of problems, on a continuum of increasing severity.

The 12th annual statistical report of the HILDA Survey (Wooden and Wilkens, 2017) presented some initial findings in the chapter dedicated to gambling. They found that 39% of Australian adults participated in gambling activities in a typical month, with an average expenditure of $115 per month. As with previous studies, the HILDA Survey shows males are not only more likely to gamble, but they spend more on average and are more likely to experience gambling-related problems. While lottery was the most common activity, problem gambling is more common among participants in poker, electronic gaming machine users, and race and sports bettors. Most notably, while higher income and full-time employment were significant indicators of participation in gambling and of higher expenditure, it is the unemployed who are at greatest risk of developing problem gambling behaviours.

The intention of this report is to build on this work and provide a more detailed overview of gambling activity in Australia in 2015, in terms of participation, expenditure, and gambling problems among regular gamblers, as drawn from the HILDA self-report survey. The report follows a format and style common to gambling prevalence studies conducted in Australia and elsewhere. As with those studies, the report is intended as a reference document. It is written primarily for researchers and government officials who have an interest in Australian gambling statistics.

Overview of study design

Sample and response

The HILDA Survey commenced in 2001 with a nationally-representative sample of Australian households (residents in private dwellings).2 Wave 1 included 11,693 households, sampled from 488 areas across Australia, with members of 7,682 households, or 13,969 individuals, completing interviews. In wave 11 (2011), the sample was topped up with an additional 2,153 households (5,477 individuals) to address the issue of recent arrivals to Australia being under-represented in the HILDA sample.

This report focuses on data from the wave 15 survey which included a gambling question module for the first time. The fieldwork for wave 15 occurred between 28 July 2015 and 7 February 2016, with 98% of fieldwork completed by 31 December 2015. Wave 15 comprised 8,865 fully responding or 9,631 fully and partially responding households, comprising 17,606 responding individuals.

The gambling module formed part of the Self-Completion Questionnaire (SCQ), a paper form administered to every member of each household aged 15 years and over. The SCQ includes questions the respondents may prefer to not disclose in the presence of an interviewer or other household members. The response rate for these was 88%, with 15,245 persons responding to the gambling module.3

Gambling module

The gambling module consisted of two components relating to participation and problems. Participants were first asked whether they spent money on 10 gambling activities in a typical month, and roughly how much on average they spent on each (Table 1.1). Respondents were considered to be "activity participants" if they responded "yes" to the question of whether they spent money on the activity, even if their expenditure estimate was missing. However, only those with valid expenditure responses were included in expenditure calculations. Missing responses for "any expenditure in a typical month" on a given activity were coded to "no" for those participants who had replied "yes" to other activities. Participants with missing responses on all gambling expenditure questions were excluded from analysis.

In a typical month, roughly how much do you spend on the following activities? This includes money spent on-line (on a computer, mobile/smart phone, iPad etc.). If you are unsure, please make your best guess.

Table 1.1: Gambling expenditure questions included in HILDA Survey wave 15
  Any expenditure on
a typical month?
How much per month?
(On average)
Instant scratch tickets ("scratchies") No/Yes $
Bingo No/Yes $
Lotto or lottery games, like Powerball or Oz Lotto No/Yes $
Keno No/Yes $
Private betting (e.g., playing cards or mah-jong with friends and family) No/Yes $
Poker No/Yes $
Casino table games (e.g., blackjack, roulette) No/Yes $
Poker machines ("pokies") or slot machines No/Yes $
Betting on horse or dog races (but not sweeps) No/Yes $
Betting on sports No/Yes $

The second component consisted of the Problem Gambling Severity Index (PGSI; Ferris and Wynne 2001). The PGSI consists of nine items that capture problematic gambling behaviour in the past 12 months, and the adverse consequences of gambling experienced in the past 12 months4. These items are shown in Table 1.2. Each item is rated on a 4-point scale, where 0 = Never and 3 = Almost always. Responses are summed to give a score between 0 and 27. The higher the score, the greater the problems or likelihood of problems.

Now thinking about the last 12 months …

Table 1.2: The Problem Gambling Severity Index
(Cross X one box on each line) Never Sometimes Most of
the time
Almost  
always
Have you bet more than you could really afford to lose?        
Have you needed to gamble with larger amounts of money to get the same feeling of excitement?        
When you gambled, did you go back another day to try and win back the money you lost?        
Have you borrowed money or sold anything to get money to gamble?        
Have you felt that you might have a problem with gambling?        
Has gambling caused you any health problems, including stress or anxiety?        
Have people criticized your betting or told you that you had a gambling problem, regardless of whether or not you thought it was true?        
Has your gambling caused any financial problems for you or your household?        
Have you felt guilty about the way you gamble or what happens when you gamble?        

All HILDA Survey participants were administered the PGSI, irrespective of whether they had gambled in a typical month. Each participant was assigned a gambling risk state according to the following risk thresholds:

Table 1.3: Problem Gambling Severity Index risk thresholds
PGSI Score Risk category
0 Non-problem gambler
1-2 Low-risk gambler
3-7 Moderate risk gambler
8 or over Problem gambler

Non-problem gamblers were those who did not engage in problematic gambling behaviour or experience adverse consequences in the past 12 months from gambling. Low-risk and moderate-risk gamblers were those who reported low or moderate level problematic behaviour and/or consequences. They are considered as being at low to moderate risk of becoming problem gamblers. Problem gamblers were those who reported high level problematic behaviour and/or consequences.

Statistical analysis

The bulk of the report presents basic descriptive statistics, such as means and percentage distributions. Tables are based on the responding sample for each individual question (i.e., item non-response is excluded), and as such baseline numbers may vary slightly between tables. The sample sizes for each of the gambling activities, and for the categories derived from the PGSI are presented in Appendix B.

Unless otherwise stated, analysis is conducted at an individual responding person level (rather than household) and based on individual's responses. As monetised gambling is illegal under the age of 18, respondents aged 15-17 were excluded from analysis. The exception to this approach is Chapter 6, which addresses gambling expenditure as part of the wider household budget and therefore uses variables constructed from all household members' responses, and includes some analysis conducted at a household level.

Gambling participation and expenditure is analysed according to a set of demographic variables that are expected to be related to gambling behaviours. Some demographic variable percentages may not add to 100% as those which had small categories which did not readily collapse into broader categories are not presented (e.g., participants born outside of Australia, Asia or Europe) but were retained while calculating percentages. For further detail regarding the construction of individual variables for analysis, see Appendix C.

To reduce the impact of outliers on estimates of gambling expenditure, estimates were Winsorised whereby values were capped at the top and bottom 1%. An individual's overall expenditure was calculated by summing the individually capped expenditure values from each of the ten activities.

Standard errors of statistics are not presented in this report, instead estimates which have a relative standard error of between 30% and 50% are marked with # to indicate unreliable estimates and where caution is required when interpreting these estimates. Estimates with a standard error >50% were suppressed, with np (not presented) appearing in place of the estimate.

Confidence intervals for means were calculated at the 95% level, using weighted standard errors. Tests of statistical significance used survey weighted chi-square and t-tests. Statistical analyses were conducted using STATA v14.2.

Weighting

In order to generalise findings to the Australian population, HILDA Survey data was weighted to reflect the probability of households and individuals being selected in the complex-cross sectional survey. This report uses the Self-Completion Questionnaire (SCQ) weighting values provided in the HILDA Survey dataset. Details about the weighting process can be found elsewhere (Watson 2012). Throughout the report, these population weights were attached to the 14,453 SCQ respondents aged 18 or above to derive estimates of the proportion and number of adults in the population within each group of interest. Standard errors for weighted data were calculated using the delete-a-group Jackknife method, using replicate weights provided in the HILDA Survey dataset. Details are available elsewhere (Hayes 2008).

Comparison of HILDA data to other gambling statistics

Appendix A provides comparisons of the estimates produced from the HILDA data to estimates produced from other national and state/territory sources. There is considerable variation across data sources in respect to estimates of gambling participation and gambling expenditure. The key contributing factor to those differences is the focus on gambling "in a typical month" in HILDA, and so excluding less regular participation and expenditure. There is also considerable variation in rates of gambling problems. This is because the HILDA Survey administered the PGSI to a population representative sample, whereas gambling studies have only administered it to people who gambled in the past year.

1 The design of these gambling questions was undertaken in collaboration with the Melbourne Institute of Applied Economic and Social Research at the University of Melbourne and Australian Government Department of Social Services.

2 The survey was approved by the Human Research Ethics Committee of the University of Melbourne (1647030).

3 Further details on the structure of the HILDA Survey sample, including complete questionnaires, are available online: <http://melbourneinstitute.unimelb.edu.au/hilda>.

4 Whereas PGSI behavioural items refer to actions in the past twelve months, the adverse consequences reported in the past twelve months may be the legacy of prior problematic gambling behaviours. E.g., gambling-related financial hardship experienced in the past 12 months may be a consequence of gambling behaviour in prior years.