Content uploaded by Philip Newall
Author content
All content in this area was uploaded by Philip Newall on May 14, 2020
Content may be subject to copyright.
BRIEF REPORT
Percentage and Currency Framing of House-Edge
Gambling Warning Labels
Philip W. S. Newall
1
&Lukasz Walasek
2
&Elliot A. Ludvig
2
#Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract
All commercial gambling games are constructed so that the gamblers will on average lose
money over time. This fact is often communicated to gamblers on virtual gambling games
as the “return-to-player.”A return-to-player of 90% means that for every £100 bet, on
average £90 is paid back out in prizes. In previous work, gamblers were better informed,
and perceived a lower chance of winning, when this information was equivalently
reframed as a “house-edge”of 10%, whereby the game keeps 10% of all money bet on
average. This paper explores whether there are further risk communication advantages to
using currency framing for the house-edge format, by directly stating the amount kept as:
“This game keeps £10 for every £100 bet on average.”Online gamblers (N = 1,007)
reported their perceived chances of winning for hypothetical games with house-edges of
either 0.5%, 7.5%, or 15%, presented as either percentages or currency units. Gamblers’
perceived chances of winning were only minimally affected by this framing of house-
edge information.
Keywords Framing effect .Behavioral science .Risk communication .Betting
All commercial gambling games are constructed so that gamblers will on average lose money
over time. Some games, however, take a greater proportion of money wagered than others,
effectively meaning that these games are sold at a higher “price”for the enjoyment derived from
wagering a given amount of money (Harrigan and Dixon 2009;Woolleyetal.2013). Some
fraction of real-world gambling behavior might be influenced by the fact that gamblers are poorly
informed about the price of gambling (Eggert 2004). An issue facing gambling warning labels for
communicating the price of different gambling products is that the price of gambling is inherently
statistical, and that proper understanding therefore requires a degree of risk literacy (Cokely et al.
International Journal of Mental Health and Addiction
https://doi.org/10.1007/s11469-020-00286-0
*Philip W. S. Newall
p.newall@cqu.edu.au
1
Experimental Gambling Research Laboratory, School of Health, Medical and Applied Sciences,
CQUniversity, 120 Spencer St, Melbourne, VIC 3000, Australia
2
Department of Psychology, University of Warwick, Coventry CV4 7AL, UK
2012). Therefore, an important question is not only what information to show to gamblers but
how best to display that information (Gigerenzer and Edwards 2003). This question is important
for policy makers, because moves toward more informative labelling of gambling product risk
would be considered the most freedom-preserving way of intervening on the public health costs of
gambling (Gambling Commission 2019; Nuffield Council on Bioethics 2007).
Currently, gambling warning labels for virtual gambling games in jurisdictions such as the UK
and the Australian state of Victoria present the “price”of electronic gambling machines to
gamblers with what is known as the “return-to-player”percentage. For example, “This game
has an average percentage payout of 90%,”means that for every £100 bet on this game £90 is paid
out on average in prizes (Collins et al. 2014). This is a statistical average payout that occurs over
the lifetime of the machine and does not refer to every play or even to each session of play. This
information also means, indirectly, that the remaining £10 from the £100 bet is kept as profit for
the game operator. A number of previous studies have shown that many gamblers fail to correctly
understand what the return-to-player means, for example by thinking that the return-to-player
percentage refers to the percentage of winning gamblers, or the percentage of individual winning
plays (Beresford and Blaszczynski 2019; Collins et al. 2014;Harriganetal.2017).
This limitation of the return-to-player as a risk communication tool suggests that alternative
approaches for communicating the price are needed. Newall et al. (2020) investigated the effects of
“reframing”the return-to-player as an equivalent statement which puts the emphasis on how much
money the game operator keeps on average: e.g., “This game keeps 10% of all money bet on
average.”This statement, which is known as the “house-edge”percentage, is formally equivalent to
a return-to-player of 90% (Parke et al. 2016). The house-edge statement was, however, understood
correctly by more regular gamblers, and led to a lower perceived chance of winning, than the
equivalent return-to-player statement (Newall et al. 2020). These two factors combined suggest that
the house edge would make gamblers better informed and more aware of the price of gambling
products. This is an example of a “framing effect,”where the way risk is communicated can impact
judgment and decision making (Levin et al. 1985; Tversky and Kahneman 1981). Given that
equivalent information is not always processed equally, it is therefore important to explore potential
further improvements in the communication of the price of gambling.
Previous research suggests that percentages are an imperfect risk communication tool (Chen
and Rao 2007; Gigerenzer and Hoffrage 1995). For example, reframing the percentage manage-
ment fees charged by mutual funds as corresponding currency equivalents can help nudge
investors toward the rational strategy of choosing a low-fee fund (Choi et al. 2010; Hastings
and Tejeda-Ashton 2008). Participants in those studies with a hypothetical portfolio of $10,000
put more weight on a management fee of $100/year than a fee of 1% a year, even though the
information conveyed by the two formats is identical. In the present context, we predict that, “This
game keeps £10 for every £100 bet on average,”helps gamblers to be more wary of the price of a
gambling game than equivalently saying, “This game keeps 10% of all money bet on average.”
The benefits of currency over equivalent percentage framing, however, are not always
uniform. In general, people are more risk-seeking for small amounts of money, which is known
as the “peanuts”effect (Weber and Chapman 2005). For example, for a small investor whose
mutual fund management fees correspond to $10–$15 a year, the currency framing actually
makes them less likely than percentage framing to choose a low-fee fund (Newall & Love
2015).
Such a combination of effects of converting percentages into currency amounts could be
useful in the current context, because this would make gamblers more wary of high house-
International Journal of Mental Health and Addiction
edge games, while increasing the relative attractiveness of low house-edge games. This would
effectively increase gamblers’sensitivity to the price of different gambling products.
Therefore, the present research explored the impact of percentage and currency framing for
house-edge warning labels on gamblers’perceived chances of winning across a wide range of
values (0.5 to 15%). This range of values was chosen because 0.5% is about the lowest house
edge possible to allow an operator to recoup the cost of providing a game, while 15% is the top
end for the house edge found previously in Canada (Harrigan and Dixon 2009) and Australia
(Woolley et al. 2013).
Our preregistered hypothesis was that there would be an interaction between label framing
and house-edge value. Specifically, we expected the dependent variable (a gambler’s perceived
chance of winning), to vary more under currency than percentage framing. That is, we
expected participants to be more sensitive to variations in a gambling product’s“price”with
currency framing. Data, materials, analysis code, and the preregistration document can be
accessed from https://osf.io/9ckph/.
Method
Participants
A total of 1007 participants were recruited on Prolific Academic and were paid £0.50 each.
Participants took an average of 3.6 min to complete the study, so this translated to an average
payment of £8.33/h. Prolific Academic is a crowdsourcing platform similar to Amazon Mechanical
Turk, where researchers post experiments for a pool of registered potential participants to complete
(Palan and Schitter 2018). Prolific Academic has the benefit compared with Mechanical Turk of
various pre-screening filters that can be set by the experimenter to ensure that only a relevant subset
of the participant pool can take part. In this case, participants were pre-screened to be aged 18 or
older, UK residents, and to have played at least one online luck-based casino gambling game (i.e.,
one or more of Baccarat, Craps, Pachinko, Roulette, Slots, Video poker, and Virtual sports betting).
Participants were 54.3% female (0.1% other), and had a mean age of 35.9 years (SD = 10.1).
Occupation was reported as follows: student (5.6%), in work (80.8%), unemployed (7.8%), retired
(15.9%), other (4.2%). Education was reported as follows: secondary school (14.2%), college
(35.2%), undergraduate (36.1%), and postgraduate (14.5%). Ethical approval was obtained from
the University of Warwick human ethics committee prior to the study commencing.
Design and Materials
Participants were randomly assigned to either receive percentage or currency framing (be-
tween-participants). Participants then completed three trials in random order, corresponding to
a house edge of 0.5%, 7.5%, and 15%. On each trial participants were presented with some
short introductory text about online gambling and then a warning label. Figure 1shows an
example from the percentage condition. The exact wordings used were as follows:
This game keeps 0.5%/7.5%/15% of all money bet on average.
This game keeps 50p/£7.50/£15 for every £100 bet on average.
International Journal of Mental Health and Addiction
On each trial participants gave their perceived chance of winning using a 7-point Likert
scale, which can also be seen in Fig. 1.
Procedure
After these three trials, participants completed an attention-check trial corresponding to an
implausibly high house edge of 95%, using the same framing that they had received over the
previous trials. As the first exclusion criterion, any participant who gave a higher perceived
chance of winning on this trial than on any previous trial was to be excluded, for reasons of
potential inattentiveness. The second exclusion criterion was to remove any participants who
gave a higher perceived chance of winning for a higher house-edge game. For example, if
participants rated a higher chance of winning with a house edge of 15% than with a house edge
of 7.5%, then they were excluded from the analysis, for an apparent failure to understand the
Fig. 1 Example of the main stimulus screen (percentage condition)
International Journal of Mental Health and Addiction
statistical nature of the house edge in gambling (which may well be due to participant
inattentiveness, in this experimental setup).
After the attention-check trial, participants completed the two individual difference scales
described below and provided demographic information.
Measures
The dependent variable was the gambler’s perceived chances of winning, as measured by a 7-
point Likert scale (see Fig. 1). Participants also completed the Problem Gambling Severity
Index (Ferris and Wynne 2001), which directly measures behavioral dependence and gambling
harm, and the Consumption Screen for Problem Gambling (Rockloff 2012), a brief three-item
screen which measures gambling consumption. The latter screen has been shown to also be an
efficient method of detecting problem gamblers, as those who gamble the most frequently are
also the most likely to have gambling problems.
Results
Participants had a mean problem gambling severity index of 3.1 (SD = 4.3), and a mean
gambling consumption screen score of 3.3 (SD = 2.7). The results of the two exclusion criteria
were as follows. The first exclusion criterion (95% house-edge catch trial) saw a loss of 6.9% of
participants in the currency condition and 3.4% in the percentage condition. The second
exclusion criterion (mistaken perceived chances of winning) saw a loss of 16.2% of participants
in the currency condition and 14.1% in the percentage condition. Across both exclusion criteria,
17.6% of participants were lost in the currency condition and 14.5% in the percentage condition.
This difference was not significantly different, as measured by logistic regression (z=−1.33,
p= .184). Because this meant that attrition was not significantly different between the two
conditions, analysis could proceed on the remainder of the sample (N= 845) as preregistered.
Data were analyzed using a mixed-effects model, to account for the shared variance across
participants’three perceived chances of winning. Perceived chances of winning were regressed on
the independent variables of framing (two levels, between-participants) and magnitude (three
levels, within-participants), and their interaction. In addition, a random intercept for participants
was included. This was performed with the afex package in R (Singmann et al. 2015).
Figure 2shows a plot of the results. There was a significant effect of magnitude (F(2,
1686) = 1557.12, p< .001), meaning that participants correctly perceived a lower chance of
winning with higher values of the house edge. There was no significant effect of condition
(F(1, 843) = 3.01, p= .08). However, an inspection of the marginal means shows there was a
trend for every level of the house edge for participants to give a higher perceived chance of
winning in the currency than percentage condition. Additionally, our hypothesis of an inter-
action between condition and magnitude was not supported (F(2, 1686) = 0.35, p= .71).
Participants’perceived chances of winning were equally responsive to variations in the “price”
of gambling, across both conditions. As this interaction effect was non-significant, no further
analyses were performed, following the preregistered analysis plan.
International Journal of Mental Health and Addiction
Discussion
Overall, there was no reliable effect of percentage or currency framing of house-edge
warning labels, with respect to gamblers’responsiveness to variations in the price of
gambling. Although house-edge labels appear better than the return-to-player labels that
are currently in use (Newall et al. 2020), reframing the house edge as a currency amount
instead of a percentage appears limited in terms of additional improvement. There was a
weak trend toward gamblers perceiving a higher chance of winning with currency than
percentage framing, although this potential effect requires replication. However, if found,
any such effect would not say that either percentage or currency house-edge labels are
more effective than the other at communicating the price of gambling, only that they
should not be used interchangeably.
This study only used an online questionnaire about a hypothetical gamble, but bigger
differences may be found in a more ecologically valid task. In addition, participants here
only gave subjective perceived chances of winning. Future studies should investigate
whether, for example, wishful thinking may contribute to some gamblers thinking they
can “beat the odds”and have a higher overall chance of winning than is communicated
through the warning label. Actual gambling behavior may also be more responsive to
changes in warning label framing than the subjective perceived chances of winning
measured here. Research should also continue to explore other potential avenues for risk
communication improvement in gambling warning labels (Ginley et al. 2017; McGivern
et al. 2019;Walkeretal.2019). For example, many electronic machine gamblers appear
confused about the return-to-player, misunderstanding that this single-play statistic does
not correspond to their expected return after gambling an initial stake repeatedly
(Harrigan et al. 2017). The currency format of house-edge warning labels may be most
effective when combined with a running total of a gambler’s total amount bet, as a
potential correction for this misunderstanding surrounding repeated gambling. It might
also be that presenting house-edge information graphically is more effective than using
text (Garcia-Retamero and Cokely 2017).
Fig. 2 Experimental results. Perceived chances of winning: 7 = Very high chance of coming out ahead, 4 =
Neither high nor low chance of coming out ahead, 1 = Very low chance of coming out ahead. Error bars represent
95% confidence intervals
International Journal of Mental Health and Addiction
Gambling is increasingly seen as a public health issue (van Schalkwyk et al. 2019;Wardle
et al. 2019). The design of more effective warning labels is just one avenue that research
should explore to attempt to lessen gambling’s public health impact (Nuffield Council on
Bioethics 2007).
Compliance with Ethical Standards
Conflict of Interest Philip Newall was in 2018 included as a named researcher on a grant funded by
GambleAware, and in 2019 received travel and accommodation funding from the Spanish Federation of
Rehabilitated Gamblers. Lukasz Walasek declares no conflict of interest. Elliot Ludvig was co-investigator on
a grant funded by the Alberta Gambling Research Institute that ended in February 2019.
Informed Consent All procedures followed were in accordance with the ethical standards of the responsible
committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as
revised in 2000 (5). Informed consent was obtained from all patients for being included in the study.
References
Beresford, K., & Blaszczynski, A. (2019). Return-to-player percentage in gaming machines: impact of informa-
tive materials on player understanding. Journal of Gambling Studies, 36,51–67. https://doi.org/10.1007
/s10899-019-09854-z.
Chen, H., & Rao, A. R. (2007). When two plus two is not equal to four: errors in processing multiple percentage
changes. Journal of Consumer Research, 34(3), 327–340.
Choi, J. J., Laibson, D., & Madrian, B. C. (2010). Why does the law of one price fail? An experiment on index
mutual funds. The Review of Financial Studies, 23(4), 1405–1432.
Cokely, E. T., Galesic, M., Schulz, E., Ghazal, S., & Garcia-Retamero, R. (2012). Measuring risk literacy: the
Berlin numeracy test. Judgment and Decision making, 7(1), 25–47.
Collins, D., Green, S., d'Ardenne, J., Wardle, H., & Williams, S. (2014). Understanding of return to player
messages: findings from user testing. London: NatCen Social Research.
Eggert, K. (2004). Truth in gaming: toward consumer protection in the gambling industry. Maryland Law
Review, 63(2), 217.
Ferris, J., & Wynne, H. J. (2001). The Canadian problem gambling index: final report. Ottawa: Canadian Centre
on Substance Abuse.
Gambling Commission. (2019). National strategy to reduce gambling harms. Retrieved from http://www.
reducinggamblingharms.org/asset-library/national-strategy-to-reduce-gambling-harms.pdf
Garcia-Retamero, R., & Cokely, E. T. (2017). Designing visual aids that promote risk literacy: a systematic
review of health research and evidence-based design heuristics. Human Factors, 59(4), 582–627.
Gigerenzer, G., & Edwards, A. (2003). Simple tools for understanding risks: from innumeracy to insight. BMJ
(Clinical Research Ed.), 327(7417), 741–744. https://doi.org/10.1136/bmj.327.7417.741.
Gigerenzer, G., & Hoffrage, U. (1995). How to improve bayesian reasoning without instruction: frequency
formats. Psychological Review, 102(4), 684–704.
Ginley, M. K., Whelan, J. P., Pfund, R. A., Peter, S. C., & Meyers, A. W. (2017). Warning messages for electronic
gambling machines: evidence for regulatory policies. Addiction Research & Theory, 25(6), 495–504.
https://doi.org/10.1080/16066359.2017.1321740.
Harrigan, K. A., Brown, D., & Barton, K.R (2017). Classification of slot machines in Ontario: providing relevant
information to players.Guelph: Gambling Research Exchange Ontario.
Harrigan, K. A., & Dixon, M. (2009). PAR sheets, probabilities, and slot machine play: implications for problem
and non-problem gambling. Journal of Gambling Issues, 23,81–110.
Hastings, J. S., & Tejeda-Ashton, L. (2008). Financial literacy, information, and demand elasticity: survey and
experimental evidence from Mexico (working paper no. w14538). Retrieved from National Bureau of
economic research website: http://www.nber.org/papers/w14538;;
Levin, I. P., Johnson, R. D., Russo, C. P., & Deldin, P. J. (1985). Framing effects in judgment tasks with varying
amounts of information. Organizational Behavior and Human Decision Processes, 36(3), 362–377.
McGivern, P., Hussain, Z., Lipka, S., & Stupple, E. (2019). The impact of pop-up warning messages of losses on
expenditure in a simulated game of online roulette: a pilot study. BMC Public Health, 19(1), 822.
Newall, P. W. S., Love, B., & C. (2015). Nudging investors big and small toward better decisions. Decision, 2(4),
319–326.
International Journal of Mental Health and Addiction
Newall, P. W. S., Walasek, L., & Ludvig, E. A. (2020). Equivalent gambling warning labels are perceived
differently. Addiction.https://doi.org/10.1111/add.14954.
Nuffield Council on Bioethics. (2007). Public health: ethical issues. Cambridge: Cambridge Publishers.
Palan, S., & Schitter, C. (2018). Prolific. ac—a subject pool for online experiments. Journal of Behavioral and
Experimental Finance, 17,22–27.
Parke, J., Parke, A., & Blaszczynski, A. (2016). Key issues in product-based harm minimisation: examining
theory, evidence and policy issues relevant in great britain. London: Responsible Gambling Trust.
Rockloff, M. J. (2012). Validation of the consumption screen for problem gambling (CSPG). Journal of
Gambling Studies, 28(2), 207–216.
Singmann, H., Bolker, B., Westfall, J., & Aust, F. (2015). Afex: Analysis of factorial experiments. R Package
Ve r si o n 0.14–2,
Tversky, A., & Kahneman, D. (1981). The framing of decisions and the psychology of choice. Science,
211(4481), 453–458.
van Schalkwyk, M. C. I., Cassidy, R., McKee, M., & Petticrew, M. (2019). Gambling control: in support of a
public health response to gambling. Lancet (London, England), 393(10182), 1680–1681. https://doi.
org/10.1016/S0140-6736(19)30704-4.
Walker, A. C., Stange, M., Dixon, M. J., Koehler, D. J., & Fugelsang, J. A. (2019). Graphical depiction of
statistical information improves gambling-related judgments. Journal of Gambling Studies, 35, 945–968.
https://doi.org/10.1007/s10899-019-09860-1.
Wardle, H., Reith, G., Langham, E., & Rogers, R. D. (2019). Gambling and public health: we need policy action
to prevent harm. Bmj, 365. https://doi.org/10.1136/bmj.l1807.
Weber, B. J., & Chapman, G. B. (2005). Playing for peanuts: why is risk seeking more common for low-stakes
gambles? Organizational Behavior and Human Decision Processes, 97(1), 31–46.
Woolley, R., Livingstone, C., Harrigan, K., & Rintoul, A. (2013). House edge: hold percentage and the cost of
EGM gambling. International Gambling Studies, 13(3), 388–402.
Publisher’sNote Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
International Journal of Mental Health and Addiction