Back to Project: Applying behavioural insights to online wagering
[Voiceover]
Betting on sports, racing and lottery tickets are the only ways to bet online in Australia.
Betting online makes it easy to gamble at any place and any time.
We designed and tested a way to alert people when their betting goes up over time.
People in our trial had these things to say.
[Text on screen]
“It’s scary to find out how much you have spent overall, because you have a win every now and then. You just assume you’re doing well”
“It was very visual, thought I was up, but was down by a lot!”
“It is quite daunting as to how much one has spent versus how much you’ve actually won”
“It puts me off betting to see how much I lost”
“This looks really great. The information is very useful, and I think it would make most casual gamblers think twice about how much they’re spending. It’s like a bank statement. There’s no escaping the figures”
Amelia Johnston: Hi, I’m Amelia Johnston, from BETA, the Behavioural Economics Team of the Australian Government.
We put human behaviour at the heart of government policy.
We improve the lives of Australians by generating and applying evidence from the behavioural and social sciences to find solutions to complex policy problems.
In this video, we will present a study in which we took on a global social issue the rapid growth of online wagering.
The project is a great example of the value of collaboration between government and academia.
It also shows the enormous power of rigorous evidence in driving good policy.
In this case we worked closely with Ben Newell from the University of New South Wales, Swee-Hoon Chuah from the University of Tasmania and Bob Slonim from the University of Technology, Sydney.
And the project led to significant impact—the team’s prototypes were ultimately adopted as a template in new regulations by Australian states and territories in July 2022.
I’ll now turn it over to Charis Anton who will introduce the problem and the possible solution we were testing.
Charis Anton: Online wagering is the fastest growing form of gambling in Australia.
It is accessible, convenient and anonymous.
No longer confined to a race track or TAB, people can be tempted to gamble anywhere, such as at work or when spending time with loved ones – it only takes a few moments to place a bet.
Because of this ease of access, it can be hard for gamblers to keep track of bets and many aren’t aware of how much they have spent.
To help reduce harm from online wagering, Commonwealth, state and territory governments launched the National Consumer Protection Framework for Online Wagering in 2018.
One of those measures was mandating activity statements to help gamblers keep track of their betting.
The Behavioural Economics Team of the Australian Government, or BETA, within the Department of the Prime Minister and Cabinet, was approached by the Department of Social Services to design and test activity statements to see what effect they’d have on online gambling.
Ben Newell: Our discussions with BETA focused on the types of thinking that people might engage in and how potential biases can influence the decision to gamble.
We wanted to design the activity statements in ways that would make clear how much people were betting and how much they were winning and losing.
Lots of behavioural research suggests that people are particularly sensitive to losses a loss of $100 is felt much more keenly than the equivalent gain of $100 – so we predicted that a statement that showed losses really clearly would decrease the tendency to gamble.
We also built on the idea that sometimes people perceive sequences in outcomes – the so called gambler’s fallacy whereby a win might be much more likely following a series of losses and vice-versa.
Again, by showing outcomes of bets aggregated across time we could highlight that sequences are irrelevant.
We also wanted to make wins and losses equally available to people – rather than allowing people to selectively remember wins more readily than losses – a tendency which is often observed.
We then distilled these ideas into concrete designs for different activity statements.
Swee-Hoon Chuah: Of course we needed to make sure these features did not result in unintended consequences.
For example, highlighting losses appeals to loss aversion, but may also result in loss chasing behaviour on the part of some gamblers.
When reminded of their losses, they may gamble more to try and recoup these.
So it is important that these prototypes are tested with real users.
We designed 6 activity statements and ran focus groups and interviews with online gamblers to get their feedback on the designs.
We used the results to create two activity statements that we brought forward into our framed field experiment.
Both statements showed people the amount they’d spent, their wins and losses and net result in a given period.
In addition one statement had a graph showing cumulative results over a period of time and the other showed cumulative results in a table.
Laura Bennetts Kneebone: Once we had our two final prototypes, we created an online game with the help of Lynxx Australia.
The game simulated betting on a series of online horse races.
Lab dollars were deposited in the ‘player account’ at the beginning and after every eight races.
The player was allocated a horse with a fun name and shown the odds, which varied at each race.
After placing the bet of their choice – or sitting out the bet – the player would watch six horses gallop along the racetrack.
After watching ‘Fluffy Dakota’ win or lose, the account balance would increase or decrease and the next race begin.
At the end of eight races, all players were given a short break before the next round.
While some players were only given time to rest, other players were shown a dynamic activity statement, which reflected their betting behaviour in the previous 8 bets, and showed whether they had lost money overall, and how much.
There were 8 rounds of 8 bets, so players who were randomly selected to see a statement, had an opportunity to learn from their earlier feedback.
All the races were won on computer generated probabilities, linked to the odds, so each player had a slightly different experience from every other player – some had winning streaks, some had losing streaks.
More wins were had on lower odds.
To increase player engagement, 3 players were randomly selected to receive their final balance, at a conversion rate of 20 lab dollars to 1 Australian Dollar, but all players received a flat fee for their participation as well.
Our final sample was 1,501 participants and we tested two statements (seen by the treatment groups) against a control group who saw no statements (there was around 500 people in each group).
We screened out high risk gamblers due to ethical reasons.
All players took part in a post-game survey.
Players who saw a statement were asked about how they used the activity statements in the game.
We chose to run the experiment in a simulated environment under controlled conditions where participants could only bet between 0 and 15 lab dollars per race.
This was done to minimise harm so gamblers weren’t betting their own money.
Despite the game not being ‘real’, the majority of participants – 86% - said it felt at least a little like real life.
Betting patterns also showed people behaved like they would in real life, for example, when the odds on a horse were higher, people bet lower amounts and vice versa.
We hypothesised that showing participants the activity statements would reduce the total amount bet, and reduce the number of bets placed.
Players who saw the activity statements during the game bet significantly less than those who did not see an activity statement - the control group.
Players who saw the statement with the graph bet 7.6% less than the control group and those who saw the statement with the table bet 4.9% less than the control group.
The difference between the two statement groups was not a significant difference.
hile the activity statements caused people to bet fewer dollars overall, they did not cause them to place fewer bets.
Those who saw the statement placed, on average, 44.8 bets and those in the control group placed, on average, 45.7 bets.
The activity statement with the graph was particularly useful for moderate-risk gamblers and people who held more false gambling beliefs.
It significantly reduced the amount they bet.
We also found that the statements did not worsen loss chasing behaviour, they actually reduced it.
The majority of respondents - 91% - said they would like to receive the activity statements in real life.
Charis Anton: We published the results on our website and provided advice to state and territory governments, recommending they adopt our activity statement design.
We proposed that online wagering operators should provide monthly statements showing gambling wins and losses over the last 6 months.
In July 2022, state and territory Governments made it mandatory for online wagering companies to provide monthly
activity statements to their customers using our design as a template.
Ben Newell: This project showcases the value of government combining with academics in applying basic insights from behavioural science to help improve people’s welfare.
Bob Slonim: This project exemplifies just one of many exciting, innovative and impactful projects that experimenters have been involved in by working with public and private sector partners, including BETA.
BETA’s seven member Academic Advisory Committee is currently engaged in several projects.
We believe the ongoing working relationship between BETA and experimenters has established a role model for academics to collaborate with private and public sector partners.
This will extend behavioral academic research insights to address a broad range of policy issues.
Thank you for listening and please watch this space for new videos.