Awareness and impact of casino responsible gambling/harm minimization measures among Canadian electronic gaming machine players

ABSTRACT

Responsible Gambling/Harm Minimization (RG/HM) measures have the potential to reduce population level gambling harms. The present study examines Canadian casino electronic gaming machine (EGM) players’ (n = 2808) awareness of a selection of the available RG/HM measures, and the impact they believed these measures had on their gambling expenditure and enjoyment. The results showed that Canadian casino EGM players were generally aware of most of the measures, with this awareness being significantly higher among at-risk and problem/pathological gamblers. However, these measures had very little perceived impact on their gambling expenditure or enjoyment. Subsequent multivariate analyzes found that a) increased levels of overall awareness of RG/HM measures were related to male gender, younger age, and higher importance of gambling as a leisure activity; b) that decreased perceived expenditure from RG/HM measures were related to younger age, being an at-risk gambler, and the frequency of limit setting in the context of gambling play; and c) increased perceived enjoyment from RG/HM measures were related to lower household incomes. Messages targeted at Canadian casino EGM players who appear sensitive to the potential and actual harms of gambling, and messages relating to limit setting frequency in the context of gambling play, may impact gambling perceptions.

KEYWORDS:

• Gambling
• harm-minimization
• harm-minimization
• public health
• electronic gaming machines

Introduction

The problem gambling prevalence rate in Canada has declined from 1.1% in 2002 to 0.6% in 2018, which represents a 45% decrease in 16 years (Williams et al., 2021). That said, this represents an estimated 156,000 people, with an additional 2.7% of the adult population being at-risk of future problem gambling (Williams et al., 2021). Hence, there continues to be a need to identify effective prevention and treatment techniques to reduce gambling-related harm.

Responsible Gambling/Harm Minimization (RG/HM) measures are industry and government policies and procedures ostensibly designed to reduce the harmful effects of gambling (Christensen, 2019). Typically, gamblers are aware of these measures (Jackson et al., 2016) which include restrictions on venue operation (e.g. limits on venue hours), EGM restrictions (e.g. limits on maximum bets), player selected limits (e.g. pre-set limits), restrictions on the gambler (e.g. not playing while intoxicated), responsible gambling messaging, and other measures intended to reduce the negative impacts of gambling (Christensen, 2019).

RG/HM is contentious for several reasons. Some researchers argue that responsible gambling is not a process but rather an outcome (Blaszczynski et al., 2022), others argue that the term ‘responsible gambling’ inappropriately puts the responsibility on the gambler rather than the provider (Livingstone & Rintoul, 2020), while others advocate for a greater focus on ‘positive play’ (Tabri et al., 2020). Further, the gambling industry and associated organizations have recently started using the term ‘safer gambling’ (Responsible Gambling Council, 2024), possibly to avoid their responsibility for gambling harms (Christensen, 2020).

Typically, many western jurisdictions have a mix of commercial and government goals that attempt to maximize gambling revenues and taxes (and sometimes supporting charitable organizations), while minimizing gambling harms, although there are notable exceptions (Christensen, 2020). These competing concerns have directed many jurisdictions to construct legal requirements for gambling provision whilst allowing consumers some choice in their play and protection (Blaszczynski et al., 2022). For example, self-exclusion is a responsible gambling policy available in many jurisdictions that is typically enacted voluntarily by gamblers (Long, 2023).

However, there are specific legal requirements placed on gambling operators. For example, Alberta Gaming, Liquor and Cannabis (AGLC) list 390 separate violations in 45 categories (215 gambling and money laundering specific violations, although many gambling venues provide liquor and gambling so these would also include the liquor provision requirements) across a three tiered system resulting in warnings, additional provision of services, reimbursements, fines, fines or temporary suspensions, hearings, suspensions, or license cancellations (Alberta Gaming, Liquor and Cannabis, 2024a). However, there are only three listed violations that relate to RG/HM (age checks, extending credit, and allowing those on self-exclusion to enter a venue). Further, these violations are infrequently penalized: violations of the self-exclusion regulations have resulted in only six decisions recorded in the past 10 years (Alberta Gaming, Liquor and Cannabis, 2024b).

The academic literature examining RG/HM measures suggests that they have varying degrees of effectiveness (Christensen, 2019; Ladouceur et al., 2017). Further, these measures are suggested to be most effective when combined, mandated, and policed (Christensen, 2019; Christensen et al., 2022b; Delfabbro & King, 2021; Livingstone & Rintoul, 2020). However, the perceived effectiveness of RG/HM can be a non-linear relationship with problem gambling severity. For example, in one jurisdiction, the proposed new RG/HM measures were perceived to be more effective for reducing gambling expenditure by low-risk gamblers than non-problem or moderate/problem gamblers (Jackson et al., 2016). Further, RG/HM influences on enjoyment are important for multiple audiences, including industry, and gamblers. Although, the evidence for RG/HM measures influencing enjoyment is less decisive (ibid).

Nevertheless, most Canadian RG/HM practices are focused on providing information to the public, training venue staff, and supplying resources to gamblers for self-moderated EGM play (Christensen et al., 2022b, 2022b). Consequently, the Canadian RG/HM focus is to educate gamblers, so they are aware of how to gamble responsibly (Alberta Gaming, Liquor and Cannabis, 2024c; Christensen et al., 2022a), rather than restrict gambling (Christensen, 2020). Similar practices exist in other liberal gambling jurisdictions (ibid).

The present study is a further investigation of this issue, with an additional focus on whether RG/HM measures are differentially impactful in relation to demographic characteristics and category of gambler (recreational, at-risk, problem/pathological). Currently, there are few studies examining these categories of player-specific effects, and often these are restricted to gender and age or type of gambling activity (Gainsbury et al., 2018; Turner et al., 2005). Moreover, examinations of the interaction between gambler characteristics, gambling play, and gambling motivations on the influence of RG/HM measures are lacking.

These gaps in the literature are surprising, as understanding who responds to these policies and procedures and under what conditions are essential for developing appropriate RG/HM messaging and policies. Consequently, this study examines the predictors of those who are responsive to RG/HM measures, and how these characteristics and related behaviors interact and influence these measures.

This is a combined hypothesis/exploratory study using online panel participants that were active Canadian gamblers who had gambled on EGMs within a casino in the past 12 months. This study is part of the Alberta Gambling Research Institute’s National Study on Gambling and Problem Gambling in Canada (https://www.ucalgary.ca/research/national-gamblingstudy/). Our hypotheses were that a) higher awareness of the RG/HM measures will be related to higher perceived impact of RG/HM measures by EGM gamblers; and b) at-risk gamblers will report that RG/HM measures have a greater impact on reducing gambling expenditure than other gambler categories. The exploratory analyzes examined whether there were interactions between RG/HM measures and demographic variables, gambling motivations, gambling fallacies, and gambling play on gambling awareness, expenditure, and enjoyment.

Method

Recruitment

The data were obtained from adult (18+) Canadian residents recruited from a pool of online panelists associated with the survey firm Leger360. This study was approved by the University of Lethbridge’s Human Ethics Review Board (Protocol#: 2018–063, approved on 19 June 2018). The study followed the Government of Canada’s Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans – TCPS 2 (2022). Participants were contacted by e-mail notifications alerting members to a new survey described as a ‘very important academic study’ without mentioning gambling as the focus. Eligibility was restricted to people who completed an initial screening question indicating that they had gambled on one or more gambling activities once a month or more in the past year. Repeated e-mail solicitations were sent out until at least 1,400 completed surveys were obtained from each province or region (i.e. 1,400 each from British Columbia, Alberta, Saskatchewan, Manitoba, Ontario, Quebec, and from the Atlantic region as a group). In addition, sixty-one participants were recruited from the territories. In total 10,199 participants were recruited from August 2018 to September 2018 (Hodgins et al., 2022). Participants completed a self-administered online questionnaire covering demographic questions and a range of topics related to gambling, substance use, and mental health. All participants consented to participate in this study.

Data quality

The survey included features to identify inaccurate responses. Surveys that were completed within 10 minutes, indicating insufficient attention to the questions, and answering questions in a regular pattern, were removed. In addition, response options were randomized across respondents.

Measures

Demographics

The survey included numerous demographic variables. The analysis focused on the variables commonly reported in Canadian gambling surveys (e.g. gender, age, highest level of education, household debt, marital status, household income, and being born in Canada).

Gambler category

The Problem and Pathological Gambling Measure (PPGM) is used to identify individuals as a non-gambler, recreational gambler, at-risk gambler, problem gambler, or pathological gambler (Williams & Volberg, 2014). The PPGM is a 17-item measure and uses a past 12-month time frame. Items are grouped into ‘problems/harms’ (n = 7), ‘impaired control’ (n = 4), and ‘other issues’ (n = 3). Total scores range from 0 to 14. To be designated as a problem gambler the PPGM requires the person to report evidence of impaired control plus significant problems/harms because of impaired control. A ‘pathological gambler’ is a problem gambler with a total score of five or higher. The PPGM better captures the heterogeneity of problem gambling relative to other measures (Christensen et al., 2019), better differentiates distinct levels of severity (Molander & Wennberg, 2022), and has very high sensitivity and specificity rates (Williams & Volberg, 2014; above 98% for both measures). For the present analysis, the non-gambler category was excluded, and the problem and pathological gambler categories were combined to reduce the effect of small cell sizes in the significance tests.

Responsible Gambling/Harm Minimization (RG/HM)

Gamblers who reported playing EGMs in the past year were asked about 14 RG/HM measures available in Canadian casinos. These measures were: responsible gambling information terminals, limits on casino hours of operation, prohibition of casino employees gambling at the casino, smoking bans, prohibition of gambling while intoxicated, gambling activity statements, limits on cash payouts, maximum bet restrictions, maximum betting lines, ability to pre-set gambling limits, clocks on EGMs, responsible gambling messages on EGMs, self-exclusion, and other measures. Respondents were asked if they were aware of the measure (yes, no); and if yes, whether the measure influenced their personal gambling expenditure (increase, no impact, decrease), and enjoyment (increase, no impact, decrease) of gambling. Positive awareness responses were coded as 1 and negative responses as 0. The spending and enjoyment responses were recoded into a countable value so total scores were able to be computed. The recoding reflected greater RG/HM impact (i.e. decreases in spending, increases in enjoyment). Gambling expenditure responses were recoded as decrease = 1, increase = −1, and no impact 0, while enjoyment responses were recoded as decrease =-1, increase = 1, and no impact as 0. The total score ranges were awareness 0 to 14, expenditure −14 to 14, and enjoyment −14 to 14.

EGM play

The survey included several questions on EGM play. The current analysis included responses to three key questions: EGM spend per month, EGM hours played per month, and EGM play frequency. All EGM responses were recorded as continuous values except for EGM frequency, which was recorded as an ordinal variable, with seven options from ‘Never’ to ‘Four or more times a week’. In addition, the ratio of net monthly win/loss to household income was computed. For this specific measure, household income responses were recoded as the mid-point between intervals, where the final interval maximum was estimated as the average of the top 1% of income in 2020: $512,000 (Statistics Canada, 2022a).

Limit setting

Respondents were also asked about their limit setting behavior. The specific questions included in the analysis were: ‘How often did you limit the amount of time playing?’, ‘How often did you limit how often you play to control your gambling?’, ‘How often did you set a predetermined spending limit?’. Response options were never, rarely, sometimes, usually, and always.

Gambling motivations

Three general gambling motivation questions were also included: ‘What is your primary reason that you gamble?’ (nine response options were available, eight specific options and ‘other’; Allami et al., 2023), ‘How important to you is gambling as a leisure or recreational activity?’, and ‘How important is money to you?’. Responses for the latter two questions were not at all important, somewhat important, quite important, and very important.

Gambling fallacies

The Gambling Fallacies Measure (GFM) is a 10-item multiple-choice measure of susceptibility to all six identified gambling fallacious beliefs: The Monte Carlo fallacy, the hot hand fallacy, belief that luck is dispositional, the illusion of control, and the neglect or misunderstanding of key statistical principles (i.e. base rate neglect and the law of large numbers). Research has shown that the GFM has demonstrated convergent and discriminant validity, and adequate 1-month test – retest reliability (Leonard et al., 2015). The total score (0–10) was used to represent this measure, with higher scores indicating greater resistance.

Data analysis

Cleaning

The data were cleaned so that ‘I prefer not to answer’, ‘I prefer not say’, or ‘uncertain’ were excluded from the analyzes. The range of the excluded responses was from 1.0% (born in Canada), to 11.8% (household income). Further, the response option ‘I don’t have a primary reason’ to the question ‘What is the primary reason that you gamble?’ was recoded into the next natural number in the sequence of the other responses (i.e. 10). Gender was recoded so that the ‘Other’ response was excluded (.1%).

Analytical techniques

The data were filtered so that only respondents that indicated they gambled on EGMs in a casino in or out of province were analyzed (n = 2808). This was done so that respondents were likely to have experienced the range of RG/HM measures we were investigating. The analyzes were non-parametric tests of independence (chi-square) for the RG/HM measures across PPGM categories (k independent samples, Kruskal-Wallis) Mann-Whitney follow-up comparisons (two independent samples), odds ratios between recreational and at-risk and problem/pathological gambler categories, parametric one way Analysis of Variance Analyzes (ANOVA) between PPGM categories for awareness, gambling expenditure, and enjoyment total scores (mean plots), and Generalized Linear Models (GLM; simultaneous main effects and 2-way interactions) for categorial and continuous measures between the same RG/HM domain total scores. The variables investigated as main effects were gambling category, demographics, EGM play, limit setting, gambling motivations, and gambling fallacies. The 2-way interactions were chosen to investigate variables often associated with influencing RG/HM (Christensen et al., 2019). These were gambling category, preset spending limit, limit time playing, limit frequency, EGM frequency, win/loss per annual income, and EGM spend. The GLM analyzes used a linked identity with maximum likelihood estimation, robust covariances, type III wald analyzes, and a full wald likelihood. The GLM analyzes fixed the continuous variables (win/loss per annual income, EGM spend, gambling fallacies total score, and EGM hours) as constants in the analyzes. The GLM used estimated marginal means and pairwise contrasts with Bonferroni adjustments to calculate pairwise differences for the categorical variables. GLM parameter estimates were reviewed to indicate significant comparisons when estimated marginal mean comparisons were unavailable (e.g. for the continuous variables). Mean differences between the target variable and the comparator (estimated marginal means difference; Mdif) and beta scores (parameter estimates; β) indicate the direction of effect. Two significant tests of model effects from the gambling expenditure results had no corresponding significant parameter estimate results (limit frequency and win/loss per annual income, and limit frequency and EGM hours).¹ The GLM table reports Wald Chi-Square (WCS) estimates, degrees of freedom (df), and significance (p-values). All analyzes were conducted using SPSS version 28.0. The statistically significant cutoff was adjusted using a Bonferroni correction for multiple comparisons: p < .0025.

Results

Demographics

Table 1 shows the percentage and number of respondents by gambling category and demographic details for the EGM casino gamblers (n = 2808). Participants were primarily recreational gamblers (52.3%). In terms of demographics, they were most often female (54.3%), aged between 55 and 64 years (21.9%), hold a bachelor’s degree (20.9%), have no debt (30.1%), were married (63.1%), whose household income was $40,000–$59,000 (18.4%), and were born in Canada (86.7%).

Table 1. Gambling and demographic categories.

Category		%	N
PPGM	Recreational	52.3	1468
	At Risk	22.7	638
	Problem/Pathological	25.0	702
Gender	Male	45.7	1283
	Female	54.3	1525
Age	Between 18 and 24	5.0	141
	Between 25 and 34	15.2	428
	Between 35 and 44	15.8	443
	Between 45 and 54	19.2	539
	Between 55 and 64	21.9	615
	Between 65 and 74	17.6	493
	75 or older	5.3	149
Highest Education	Primary	3.1	84
	Some Secondary	5.0	138
	Completion of Secondary School	15.8	436
	Some Vocational Training	3.1	86
	Completion of Vocational Training	7.9	218
	Some Post Secondary	16.2	445
	A post-secondary Certificate	18.5	510
	Bachelor’s Degree	20.9	574
	Professional Degree	3.4	93
	Master’s or Doctorate	6.1	168
Household Debt	No debt	30.1	739
	Less than $39,000	29.0	712
	$40,000 - $79,000	10.1	248
	$80,000 - $119,000	6.6	161
	$120,000 - $159,000	5.7	140
	$160,000 - $199,000	4.8	117
	$200,000 - $239,000	3.8	94
	$240,000 - $299,000	3.6	88
	>$300,000	6.5	159
Marital Status	Single	21.0	581
	Married	63.1	1741
	Separated	2.5	69
	Divorced	7.8	216
	Widowed	5.6	154
Household Income	Less than $20,000	6.8	168
	$20,000 - $39,000	15.7	388
	$40,000 - $59,000	18.4	457
	$60,000 - $79,000	17.3	430
	$80,000 - $99,000	13.4	333
	$100,000 - $119,000	10.1	251
	$120,000 - $139,000	6.7	165
	>$140,000	11.6	287
Born in Canada	No	13.3	370
	Yes	86.7	2407

Awareness, gambling expenditure, and enjoyment

Table 2 shows the proportion of EGM gamblers who were aware of individual RG/HM measures across PPGM categories. The recreational category had statistically significant lower scores compared to the at-risk subgroup for responsible gambling information terminals, limits on casino hours, ability to receive statements, limits on cash payouts, limit on maximum bets, limit on maximum lines, ability to pre-set limits, clocks, responsible gambling messaging, and casino self-exclusion. The recreational category had statistically significant lower scores compared to the problem/pathological category for all RG/HM measures except for smoking bans, which was higher. There were also statistically significant differences between the at-risk and problem/pathological categories. The at-risk category had a higher score for smoking bans, but lower scores for prohibition of gambling while intoxicated, limit of maximum number of lines, ability to preset limits, and clocks.

Table 2. Awareness.

	PPGM recoded into three categories
	Recreational				At Risk				Problem/Pathological
	Average	SD	Y	N	Average	SD	Y	N	Average	SD	Y	N
‘Responsible gambling information terminals’ and/or ‘gambling information centres’	0.55	0.50	801	665	0.67	0.47	426	210	0.68	0.47	475	226
Limits on casino hours of operation	0.57	0.50	831	635	0.65	0.48	412	224	0.69	0.46	484	217
Prohibition of casino employees from gambling at the casino	0.62	0.48	913	553	0.68	0.47	433	203	0.69	0.46	485	216
Smoking bans	0.88	0.33	1290	176	0.87	0.34	552	84	0.76	0.43	533	168
Prohibition of gambling while intoxicated	0.42	0.49	623	843	0.46	0.50	292	344	0.55	0.50	383	318
Ability to receive statements of gambling spending and frequency for reward/loyalty card members	0.28	0.45	414	1052	0.39	0.49	249	387	0.44	0.50	307	394
Limits on cash payouts for winnings (i.e. receiving a cash voucher or cheque instead)	0.43	0.49	626	840	0.53	0.50	340	296	0.56	0.50	391	310
A limit on the maximum bet on electronic gambling machines	0.50	0.50	734	732	0.61	0.49	390	246	0.65	0.48	459	242
A limit on the maximum number of betting lines on slots or VLTs	0.41	0.49	606	860	0.53	0.50	335	301	0.61	0.49	430	271
Ability to put preset limits on time, money or frequency of slot or VLT play	0.27	0.45	399	1067	0.39	0.49	246	390	0.51	0.50	358	343
Clocks on electronic gambling machines	0.26	0.44	379	1087	0.40	0.49	252	384	0.55	0.50	386	315
Responsible gambling messaging on electronic gambling machines	0.62	0.49	902	564	0.69	0.46	442	194	0.69	0.46	483	218
Casino self-exclusion (ability to ban yourself from the casino)	0.58	0.49	845	621	0.65	0.48	416	220	0.66	0.47	466	235
Other	0.01	0.08	10	1456	0.02	0.14	13	623	0.03	0.16	19	682

Average = average score.

SD = Standard deviation.

Y/N = Number of respondents with a ‘Yes’ (1) or ‘No’ (0) aware endorsement.

Bold indicates significant chi-square test of independence, the lowest proportion across categories, p<0.0025.

Number of significant differences: at risk vs. problem pathological (n=5).

recreational vs. at risk (n=10), recreational vs. problem/pathological (n=14)

Table 3 shows the average gambling expenditure scores for individual RG/HM measures across PPGM categories. Table 3 shows very low average scores across PPGM categories and only a few significant differences between categories. There were no significant differences between recreational and at-risk categories for any RG/HM measure. However, the recreational category had statistically significant higher scores than the problem/pathological category for the ability to pre-set limits, and clocks. There was only one statistically significant difference between the at-risk group and the problem/pathological categories, the at-risk category had a higher score than the problem/pathological category for clocks.

Table 3. Gambling expenditure.

	PPGM recoded into three categories
	Recreation					At Risk					Problem/Pathological
	Average	SD	Inc	No	Dc	Average	SD	Inc	No	Dc	Average	SD	Inc	No	Dc
‘Responsible gambling information terminals’ and/or ‘gambling information centres’	0.07	0.30	8	725	68	0.10	0.41	16	350	60	0.01	0.65	97	277	101
Limits on casino hours of operation	0.09	0.30	5	750	76	0.13	0.46	20	320	72	0.03	0.70	112	244	128
Prohibition of casino employees from gambling at the casino	0.05	0.26	7	851	55	0.05	0.37	19	374	40	−0.01	0.60	90	309	86
Smoking bans	0.04	0.35	53	1133	104	0.04	0.40	34	461	57	0.04	0.65	101	308	124
Prohibition of gambling while intoxicated	0.08	0.31	8	559	56	0.08	0.49	24	221	47	−0.02	0.75	113	166	104
Ability to receive statements of gambling spending and frequency for reward/loyalty card members	0.06	0.33	10	368	36	0.03	0.49	26	189	34	0.01	0.77	89	127	91
Limits on cash payouts for winnings (i.e. receiving a cash voucher or cheque instead)	0.08	0.30	6	565	55	0.07	0.45	23	270	47	0.01	0.70	94	199	98
A limit on the maximum bet on electronic gambling machines	0.11	0.34	6	642	86	0.17	0.49	19	284	87	0.07	0.71	100	229	130
A limit on the maximum number of betting lines on slots or VLTs	0.13	0.38	9	510	87	0.15	0.48	18	250	67	0.01	0.71	104	216	110
Ability to put preset limits on time, money or frequency of slot or VLT play	^a0.14	0.41	10	324	65	0.14	0.53	20	171	55	^a−0.02	0.74	100	164	94
Clocks on electronic gambling machines	^a0.15	0.40	7	310	62	^b0.18	0.51	14	178	60	^a,b0.00	0.72	100	187	99
Responsible gambling messaging on electronic gambling machines	0.07	0.29	8	821	73	0.11	0.43	20	355	67	0.01	0.65	100	279	104
Casino self-exclusion (ability to ban yourself from the casino)	0.06	0.27	6	779	60	0.09	0.38	13	353	50	0.01	0.66	97	266	103
Other*	0.06	0.24	0	17	1	0.35	0.75	3	7	10	−0.06	0.79	17	20	14

Average = average score.

SD = Standard deviation.

Number of respondents scoring: Inc = increase (−1), No = no change (0), Dc = decrease (1)

Bold indicates significant chi-square test of independence, the lowest average score across categories, p< 0.0025.

Superscript indicates significant differences between pairs, p<0.0025.

*The fewer responses to this question likely resulted in the non-significant results (n = 89).

Number of significant differences:recreational vs. at risk (n = 0), recreational vs. problem/pathological (n = 2), at risk vs. problem/pathological (n = 1).

Table 4 shows the average enjoyment scores for individual RG/HM measures across PPGM categories. The only statistically significant difference was that the recreational category had a lower score than the problem/pathological category for responsible gambling messaging.

Table 4. Enjoyment.

	PPGM recoded into three categories
	Recreational					At Risk					Problem/Pathological
	Average	SD	Dc	No	Inc	Average	SD	Dc	No	Inc	Average	SD	Dc	No	Inc
‘Responsible gambling information terminals’ and/or ‘gambling information centres’	0.00	0.29	33	736	32	0.02	0.40	30	357	39	0.03	0.63	89	285	101
Limits on casino hours of operation	−0.02	0.29	43	759	29	0.00	0.42	35	341	36	0.05	0.69	104	250	130
Prohibition of casino employees from gambling at the casino	0.00	0.26	33	849	31	0.02	0.35	23	379	31	0.07	0.62	77	298	110
Smoking bans	0.13	0.51	93	930	267	0.11	0.57	62	370	120	0.05	0.72	125	259	149
Prohibition of gambling while intoxicated	0.03	0.38	35	534	54	0.09	0.49	23	219	50	0.09	0.74	88	172	123
Ability to receive statements of gambling spending and frequency for reward/loyalty card members	0.03	0.35	19	362	33	0.08	0.49	20	189	40	0.12	0.76	72	126	109
Limits on cash payouts for winnings (i.e. receiving a cash voucher or cheque instead)	−0.04	0.31	42	564	20	−0.03	0.48	44	262	34	−0.02	0.73	108	184	99
A limit on the maximum bet on electronic gambling machines	0.01	0.30	29	669	36	0.01	0.46	39	308	43	0.01	0.68	104	246	109
A limit on the maximum number of betting lines on slots or VLTs	−0.01	0.31	34	546	26	−0.01	0.47	38	262	35	0.03	0.70	98	222	110
Ability to put preset limits on time, money or frequency of slot or VLT play	−0.01	0.35	25	351	23	0.02	0.50	28	185	33	0.06	0.74	88	160	110
Clocks on electronic gambling machines	−0.01	0.35	25	332	22	0.00	0.48	28	195	29	0.06	0.70	84	194	108
Responsible gambling messaging on electronic gambling machines	^a0.00	0.28	35	831	36	0.00	0.44	43	358	41	^a0.09	0.64	81	279	123
Casino self-exclusion (ability to ban yourself from the casino)	0.00	0.25	25	793	27	0.01	0.35	24	364	28	0.07	0.63	78	277	111
Other*	0.00	0.34	1	16	1	−0.15	0.67	6	11	3	0.22	0.83	13	14	24

Average = average score.

SD = Standard DeviationNumber of respondents scoring: Dc = decrease (−1), No = no change (0), Inc = increase (1)

Bold indicates significant chi-square test of independence, the lowest average score across categories, p<0.0025.

Superscript indicates significant differences between pairs, p<0.0025.

*The fewer responses to this question likely resulted in the non-significant results (n=89).

Number of significant differences:recreational vs. problem/pathological (n=1),recreational vs. at risk (n=0), at risk vs. problem/pathological (n=0).

Odds ratios

Table 5 shows the odds ratios comparing at-risk and problem/pathological categories with the reference recreational category. In general, these results mirror the previous chi-square analyzes. The at-risk and problem/pathological categories had higher RG/HM awareness scores than the recreational category, while the gambling expenditure and enjoyment odds ratios were closer to 1.0 indicating fewer differences between PPGM categories (although the ‘other’ RG/HM measure for the at-risk category was twice as strong as the recreation category, the ‘other’ RG/HM category had fewer responses and was not significant in the chi-square analyzes).² Notably, the strongest results were that the problem/pathological category had a lower gambling expenditure score for the ability to preset limits (the same as the chi-square results) and had higher enjoyment scores for responsible gambling messaging (again, the same as the chi-square results) and prohibition of casino employees gambling at the casino than the recreational category.

Table 5. Odds ratios.

	Awareness			Expenditure			Enjoyment
	AR vs R	PP vs R	N	AR vs R	PP vs R	N	AR vs R	PP vs R	N
‘Responsible gambling information terminals’ and/or ‘gambling information centres’	1.684	1.745	2803	1.180	0.710	1702	1.136	1.189	1702
Limits on casino hours of operation	1.405	1.704	2803	1.239	0.821	1727	1.098	1.500	1727
Prohibition of casino employees from gambling at the casino	1.292	1.360	2803	0.986	0.666	1831	1.144	1.676	1831
Smoking bans	0.897	0.433	2803	1.017	1.065	2375	0.914	0.752	2375
Prohibition of gambling while intoxicated	1.149	1.630	2803	1.031	0.652	1298	1.291	1.384	1298
Ability to receive statements of gambling spending and frequency for reward/loyalty card members	1.635	1.980	2803	0.903	0.823	970	1.205	1.561	970
Limits on cash payouts for winnings (i.e. receiving a cash voucher or cheque instead)	1.541	1.692	2803	0.985	0.746	1357	1.015	1.022	1357
A limit on the maximum bet on electronic gambling machines	1.581	1.892	2803	1.362	0.894	1583	1.007	1.019	1583
A limit on the maximum number of betting lines on slots or VLTs	1.579	2.252	2803	1.099	0.647	1371	1.016	1.226	1371
Ability to put preset limits on time, money or frequency of slot or VLT play	1.687	2.791	2803	1.046	0.592	1003	1.098	1.343	1003
Clocks on electronic gambling machines	1.882	3.515	2803	1.171	0.605	1017	1.046	1.369	1017
Responsible gambling messaging on electronic gambling machines	1.425	1.385	2803	1.226	0.718	1827	0.968	1.732	1827
Casino self-exclusion (ability to ban yourself from the casino)	1.390	1.457	2803	1.165	0.763	1727	1.048	1.656	1727
Other	3.038	4.056	2803	2.486	0.747	89	0.689	1.905	89

R= Recreational, AR = At Risk, PP=Problem/Pathological.

N = Number in analysis.

Mean plots

Figure 1 shows the average RG/HM total scores (N = 2808; all plots) and standard error bars (these were very small: .04 to .15) for awareness, expenditure, and enjoyment. Figure 1 shows higher scores for awareness than for expenditure or enjoyment. All three plots show relatively positive linear relationships as PPGM severity increases. However, expenditure had slightly higher total scores for the at-risk category compared to the problem/pathological category.

Figure 1. Average plots of awareness, gambling expenditure, enjoyment total scores.

Generalized linear models

Table 6 shows the GLM main effects and 2-way interaction models for the total scores of awareness, gambling expenditure, and enjoyment for the variable groups problem gambling severity, demographics, EGM play, limit setting, gambling motivations, and gambling fallacies. The GLM analyzes for the total awareness, gambling expenditure, and enjoyment scores reported on 81.9% (n = 2300) of the filtered data (n = 2808).

Table 6. GLM main effects and 2-way interactions.

		Awareness			Expenditure			Enjoyment
Main Effects		WCS	df	Sig.	WCS	df	Sig.	WCS	df	Sig.
	(Intercept)	525.568	1	0.000	0.654	1	0.419	5.972	1	0.015
Problem Gambling Severity	PPGM	7.498	2	0.024	14.741	2	<.001	5.915	2	0.052
Demographics	Please indicate your sex	13.568	1	<.001	0.569	1	0.451	0.798	1	0.372
	How old are you?	22.414	6	0.001	27.816	6	<.001	19.585	6	0.003
	What is your highest level of education?	3.257	9	0.953	7.633	9	0.571	6.902	9	0.647
	What is your best estimate of your household debt?	9.532	8	0.299	12.461	8	0.132	8.058	8	0.428
	What is your current marital status?	6.581	4	0.160	3.689	4	0.450	1.665	4	0.797
	What is your best estimate of your household income?	8.033	7	0.330	21.249	7	0.003	24.616	7	<.001
	Were you born in Canada?	0.206	1	0.650	1.886	1	0.170	0.699	1	0.403
EGM Play	EGM spend per month	1.535	1	0.215	0.010	1	0.921	0.113	1	0.737
	EGM hours played per month	3.384	1	0.066	2.542	1	0.111	0.174	1	0.677
	EGM Frequency	13.392	5	0.020	2.072	5	0.839	6.676	5	0.246
	Net Monthly EGM Win(-) Loss (+) per Annual Household Income	0.170	1	0.680	0.024	1	0.876	0.007	1	0.932
Limit Setting	How often did you limit the amount of time playing?	8.449	4	0.076	7.303	4	0.121	16.351	4	0.003
	How often did you limit how often you play to control your gambling?	16.336	4	0.003	2.846	4	0.584	3.685	4	0.450
	How often did you set a predetermined spending limit?	9.151	4	0.057	5.284	4	0.259	3.936	4	0.415
Gambling Motivations	What is your primary reason that you gamble?	24.832	9	0.003	28.678	9	<.001	16.088	9	0.065
	How important to you is gambling as a leisure or recreational activity?	17.305	3	<.001	2.018	3	0.569	1.058	3	0.787
	How important is money to you?	2.725	3	0.436	11.443	3	0.010	10.010	3	0.018
Gambling Fallacies	Gambling Fallacies	0.043	1	0.836	4.794	1	0.029	7.644	1	0.006
2-way Interactions
Problem Gambling Severity	PPGM * Preset spending Limit	16.399	8	0.037	13.170	8	0.106	8.024	8	0.431
	PPGM * Limit time playing	2.836	8	0.944	6.162	8	0.629	7.199	8	0.515
	PPGM * Limit Frequency	12.119	8	0.146	16.183	8	0.040	3.848	8	0.871
	PPGM * EGM Frequency	15.912	10	0.102	16.392	10	0.089	11.394	10	0.328
	PPGM * Net Monthly EGM Win(-) Loss (+) per Annual Household Income	0.579	2	0.749	0.265	2	0.876	5.178	2	0.075
	PPGM * EGM Spend per month	6.522	2	0.038	0.546	2	0.761	2.002	2	0.368
	PPGM * EGM Hours	12.164	2	0.002	3.551	2	0.169	3.861	2	0.145
Preset Spending limit	Preset Spending limit * Limit Time playing	39.796	16	<.001	15.137	16	0.515	19.064	16	0.265
	Preset Spending limit * Limit Frequency	36.054	16	0.003	26.013	16	0.054	21.228	16	0.170
	Preset Spending limit * EGM Frequency	35.358	20	0.018	31.566	20	0.048	30.635	20	0.060
	Preset Spending limit * Net Monthly EGM Win(-) Loss (+) per Annual Household Income	7.601	4	0.107	3.080	4	0.544	2.709	4	0.608
	Preset Spending limit * EGM Spend per month	11.145	4	0.025	5.315	4	0.257	3.147	4	0.534
	Preset Spending limit * EGM Hours	5.442	4	0.245	9.026	4	0.060	3.724	4	0.445
Limit Time playing	Limit Time playing * Limit Frequency	8.954	16	0.915	28.737	16	0.026	14.767	16	0.542
	Limit Time playing * EGM Frequency	15.147	20	0.768	20.374	20	0.435	37.712	20	0.010
	Limit Time playing * Net Monthly EGM Win(-) Loss (+) per Annual Household Income	11.379	4	0.023	3.709	4	0.447	2.276	4	0.685
	Limit Time playing * EGM Spend per month	5.909	4	0.206	9.030	4	0.060	5.076	4	0.280
	Limit Time playing * EGM Hours	3.804	4	0.433	14.911	4	0.005	9.724	4	0.045
Limit Frequency	Limit Frequency * EGM Frequency	13.572	20	0.852	34.206	20	0.025	37.634	20	0.010
	Limit Frequency * Net Monthly EGM Win(-) Loss (+) per Annual Household Income	7.803	4	0.099	17.722	4	0.001	6.698	4	0.153
	Limit Frequency * EGM Spend per month	7.820	4	0.098	25.793	4	<.001	20.547	4	<.001
	Limit Frequency * EGM Hours	1.483	4	0.830	16.694	4	0.002	8.841	4	0.065
EGM Frequency	EGM Frequency * Net Monthly EGM Win(-) Loss (+) per Annual Household Income	2.046	5	0.843	18.371	5	0.003	17.118	5	0.004
	EGM Frequency * EGM Spend per month	2.529	5	0.772	19.019	5	0.002	12.459	5	0.029
	EGM Frequency * EGM Hours	3.301	5	0.654	15.209	5	0.010	11.172	5	0.048
Win Income	Net Monthly EGM Win(-) Loss (+) per Annual Household Income * EGM Spend	2.427	1	0.119	1.923	1	0.166	3.093	1	0.079
	Net Monthly EGM Win(-) Loss (+) per Annual Household Income * EGM Hours	1.445	1	0.229	0.639	1	0.424	0.000	1	0.982
EGM Spend	EGM Spend * EGM Hours	0.861	1	0.353	0.002	1	0.964	0.270	1	0.603

Bold indicates p<.0025.

Main Effects

The GLM awareness total score analysis found significant main effects (non-intercept) differences for gender (man vs. woman; Mdif = .5215, p < .001), age (35-44 yrs vs. 65-74 yrs, Mdif = 1.0462, p < .001), and how important is gambling as a leisure activity to you (not at all vs. somewhat important, Mdif = −.6207, p < .001). The GLM gambling expenditure total score analysis found significant main effects differences for gambling category (at-risk vs. problem/pathological, Mdif = 1.8909, p < .001), age (18-24 yrs vs. 65-74 yrs, Mdif = 1.8281; 18-24 yrs vs. 75+yrs, Mdif = 1.9676; 35-44 yrs vs. 75+yrs, Mdif = .9705, p < .001), and primary reason to gamble (to develop my skills vs. I don’t have a primary reason, β=-2.172, p < .001). The GLM enjoyment total score analysis found significant main effects differences for household income ($60,000–$79,000; β = .659, $100,000-$119,000; β = .810, $120,000-$139,000; β = .828 all vs. >$140,000, p < .001).

2-way Interactions

The GLM awareness total score analysis found significant 2-way interaction effects for gambling category and EGM hours (recreational vs. problem/pathological, β = .029, p=.002), and preset spending limit and limit time playing (never presetting limits and always limiting time vs. always presetting limits and usually limiting time, Mdif = −4.9436, p < .001). The GLM gambling expenditure total score analysis found significant 2-way interaction effects for limit frequency and win/loss per annual income (p=.001), limit frequency and EGM spend per month (limit frequency sometimes vs. always, β = .002, p=<.001), limit frequency and EGM hours (p=.002), and EGM frequency and EGM spend (two to three times a month vs. 4 or more times a week, β=-.001, p=.002). The GLM enjoyment total score analysis found significant 2-way interaction effects for limit frequency and EGM spend (never, β=-.002, and sometimes, β=-.002, vs. always, p < .001).

Discussion

The individual chi-square analyzes of the RG/HM measures mirror previous research that shows moderate to high awareness of RG/HM policies (Hodgins et al., 2022), and minimal impact on gambling expenditure and enjoyment (Jackson et al., 2016). The primary results for awareness of the individual RG/HM measures were that the recreational category had statistically significant lower scores compared to the problem/pathological category for most of the RG/HM measures. The primary results for gambling expenditure were that the recreational category had statistically significant higher scores than the problem/pathological category for the ability to pre-set limits and clocks. The only statistically significant result for enjoyment was that the recreational category had a lower score than the problem/pathological category for responsible gambling messaging.

In contrast, the multivariate analyzes are the first to show that multiple demographic, gambling play, and gambling motivation variables had statistically significant effects on total RG/HM awareness, gambling expenditure, and enjoyment. Specifically, the main effects results (i.e. the influence of single variables) showed that awareness of RG/HM measures were higher for males, and higher for middle aged vs older gamblers, and lower if gambling was not an important leisure activity vs somewhat important. Gambling expenditure was higher for at-risk than problem/pathological gamblers, higher for younger gamblers, and lower for those who wanted to develop their skills vs those for don’t have a primary reason to gamble. Gambling enjoyment was higher for lower incomes compared to those earning the highest incomes.

The multivariate main effects results suggest two opposing aspects; 1) those often at-risk for problem gambling or those more vulnerable to harms (men, younger gamblers, those with lower incomes; Statistics Canada, 2022b), and 2) those who have ambiguous attitudes toward gambling (who consider gambling a somewhat important leisure activity, who don’t have a primary reason to gamble, and are possibly experiencing sub-clinical gambling harms), appear more sensitive to RG/HM measures. These results suggest two types of messages for educational campaigns; 1) targeting those who are vulnerable for problem gambling (but not experiencing harms) and 2) targeting those who are experiencing subclinical gambling harms (and probably not wanting more).

The multivariate interaction analyzes show that limit setting frequencies was a contributing factor in most of the significant results. Specifically, regular limit setting (preset spending limits and limit time playing) predicted greater awareness of RG/HM measures, and infrequent limit setting and EGM spend predicted less gambling expenditure, and regular limit setting and EGM spend predicted greater enjoyment. Further, limit setting frequency and win/loss per annual income, and limit frequency and EGM hours were also predictive of gambling expenditure (see footnote 1). Again, these results suggest two foci for RG/HM interventions and messaging: 1) limit setting frequencies (Delfabbro & King, 2021), and 2) EGM play (el-Guebaly et al., 2015).

These results indicated that a) contrary to our hypothesis, greater awareness was not related to greater perceived impact of RG/HM measures, and b) supporting our hypothesis, the at-risk category was related to greater perceived influence of RG/HM measures on gambling expenditure scores. Further, our exploratory interaction analyzes found that limit setting frequencies in concert with EGM play were related to greater perceived impact on gambling expenditure and enjoyment.

These results imply that RG/HM policies that target multiple characteristics of gamblers and gambling play (e.g. demographics, EGM play, limit setting, gambling category, and gambling motivations) are effective for influencing gambling perceptions (Christensen, 2019). These results also appear to fall into gambler experiences that are consistent with each other: those that are at some risk to gambling harms but not experiencing them (and possibly engaging in more limit setting behaviors), and those who are experiencing some harms (likely from more frequent gambling episodes) who are less likely to set limits (i.e. for this group, who don’t set as many limits, any limit setting is impactful because of the frequency of their gambling). These results suggest that traditional RG/HM policies focusing on recognizing the signs of problem gambling might be missing two important groups, those that have some risk characteristics for problem gambling but are not experiencing problems (and wish this to continue), and those who are beginning to experience problems (but are not problem gamblers). Targeting those with an interest in avoiding possible or worsening problems seems an obvious approach, and likely more successful for reducing individual and population level gambling harms than focusing on those uninterested in the harms of gambling. Also, these results imply those experiencing problem gambling probably need more direct interventions (e.g. self-exclusion, counseling) or mandatory mechanisms (e.g. loss limits), to change their behavior.

Although these conclusions are based on statistically significant results, the average responses were highest for awareness of RG/HM measures, and markedly lower for gambling expenditure and enjoyment. Consequently, the significant effects in the multivariate analyzes might indicate small changes in behavior (although the effects sizes of the multivariate analyzes varied between RG/HM measures indicating some of the results are likely to be more robust than others). Also, replications of these analyzes are necessary to clarify these results, especially the discrepancies between factors and individual level analyzes. Further research will need to examine whether these significant interactions exist in other samples. In addition, as these results are based on self-reported perceptions of the impact of selected RG/HM measures, future research needs to look at the impact of RG/HM on actual gambling behavior. However, similar research indicates that a significant minority of respondents will enact their limit setting intentions (Currie et al., 2020). Also, there were no meaningful gender or ethnicity analyzes conducted. Finally, the data collection period was only two months, so changes to the gambling experience or seismic events (e.g. COVID-19) occurring outside of this period (Shaw et al., 2022b, 2022b) were not reflected in the data or analyzes.

In conclusion, we found greater awareness of RG/HM measures by those experiencing more gambling harms, and those at-risk for problem gambling thought that the RG/HM policies would decrease their gambling expenditure. However, there was limited evidence linking awareness of RG/HM measures to responsible gambling behavior (Christensen et al., 2022b), rather a range of variables seem to influence the perception of RG/HM effectiveness. Namely, demographics, gambling play, gambling category, and limit setting frequency in concert with gambling play. Consequently, these results have implications for public health messaging, including specific messages for gamblers who wish to avoid harms, and a focus on the frequency of limit setting in the context of gambling play. Nevertheless, further research is needed to empirically test these conclusions.

Disclosure statement

DC, AR, RW, YB, CS, NE, DH, DM, GS, & RS: No conflicts of interest. YA: YA has received consulting fees from the responsible gambling division of a provincial (Crown Corporation) gambling operator in Canada, for work unrelated to the research presented here. FN: Dr Fiona Nicoll held a research chair with the Alberta Gambling Research Institute (AGRI) from 2016 to 2021. She has also held a large grant from AGRI which was independently peer-reviewed. She does not accept funds for providing peer-review of grant applications nor does she accept honoraria for sharing her research at events related to gambling that are sponsored by governments or gambling businesses or from organizations promoting ‘responsible gambling’ or ‘harm minimisation’ or treatment services for problem gambling.

Data availability statement

The data used in this study is available from Gambling Research Exchange Ontario. Please contact them for more details. See: https://www.greo.ca/en/index.aspx.

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

This work was supported by funding from the Alberta Gambling Research Institute.

Notes on contributors

Darren R. Christensen

Darren R. Christensen Associate Professor, Faculty of Health Sciences, University of Lethbridge, Canada. His research interests include developing behavioral treatments for problem gambling, counselling for problem gambling, evaluations of the effectiveness of harm minimization measures, investigations of regular opioid antagonist dosing on gambling urge and brain function, opioid agonist replacement therapies, and electroencephalogram studies.

Amanda Roberts

Amanda Roberts Professor, School of Psychology University of Lincoln, United Kingdom. Her research interests extend across topics that relate to gambling-related harm, gambling and mental health, gambling in vulnerable populations, gambling and interpersonal violence, NPS use, and homelessness.

Robert J. Williams

Robert J. Williams Professor, Faculty of Health Sciences, University of Lethbridge, Canada. Professor Williams is widely published and is one of the world’s leading authorities in the areas of: prevention of problem gambling, the etiology of problem gambling, online gambling, the socioeconomic impacts of gambling, the proportion of gambling revenue deriving from problem gamblers, the prevalence and nature of gambling in Indigenous communities, and best practices in the population assessment of problem gambling.

Youssef Allami

Youssef Allami Post Doctoral Fellow, Faculty of Health Sciences, University of Lethbridge, Canada. Dr. Allami’s research interests include CBT & ACT theoretical approaches, gambling & substance use disorders, motivational interviewing, among other topics.

Yale Belanger

Yale Belanger Professor, Political Sciences, Arts and Science, University of Lethbridge, Canada. Professor Belanger’s research interests include problem gambling, homelessness and homeless people, sustainable development, intercultural and ethnic relationships, public policies, philosophy and ideology, Canada, political culture, society and ideology, social organization and political systems.

Carrie Shaw

Carrie Shaw Manager, AGRI National Project, Alberta Gambling Research Institute, Canada. Dr. Shaw research interests include erroneous and anti-scientific beliefs, how these beliefs influence behavior (e.g., gambling behaviors, endorsement of alternative medicines, etc.), and how to change these beliefs.

Nady el-Guebaly

Nady el-Guebaly Professor Emeritus, Department of Psychiatry, University of Calgary; Psychiatrist, Alberta Health Services, Canada. Professor el-Guebaly was the Head, Division of Addiction, Department of Psychiatry at the University of Calgary; Research Director of the Alberta Gambling Research Institute, Chief Examiner of the International Society of Addiction Medicine, Editor-in-Chief of the Canadian Journal of Addiction.

David C. Hodgins

David C. Hodgins Professor, Department of Psychology, University of Calgary, Canada. Professor Hodgins research interests include addictive behaviors, including alcohol and gambling addictions, and comorbid psychiatric disorders. He is also interested in the process of recovery and relapse from problems including brief motivational interventions.

Daniel S. McGrath

Daniel S. McGrath Associate Professor, Department of Psychology, University of Calgary, Canada. His research interest includes applications from prospective graduate students in the psychology program (not clinical psychology). His research focus is primarily behavioral finance (e.g. cryptocurrency, financial speculation, trading, investing), personality (e.g. HEXACO) and behavioral addictions (e.g. disordered gambling, gaming).

Fiona Nicoll

Fiona Nicoll Professor, Political Sciences, Faculty of Arts, University of Alberta, Canada. Professor Nicoll’s research include the politics of art and everyday life, reconciliation, Indigenous sovereignty, racism, public art, nationalism, whiteness, gambling policy. Frameworks of analysis include critical gambling studies, critical race and whiteness studies, Indigenous knowledges and STS, theories of aesthetics and cultural history.

Garry Smith

Gary Smith Professor, Recreation and Leisure Studies, University of Alberta, Canada. Professor Smith’s research interests include social policy, sociology, and gambling.

Rhys M. G. Stevens

Rhys M. G. Stevens Librarian, Library, University of Lethbridge. Mr. Stevens research interests include Gambling Research Resources and Information Sources, Digitization and Description of Materials, Self-Archiving and Institutional Repositories.

Notes

1. This is likely the result of differences between a) the tests of model effects that examine all the responses within a factor, and b) parameter estimates that examine responses within specific levels of a factor (and so have different degrees of freedom).

2. The ‘other’ responses were consistent with the other 13 RG/HM measures we were investigating (e.g. ‘set a limit’, ‘no smoking’, ‘gamble responsible messages’, ‘no play when intoxicated’, ‘max payouts’, ‘casino hours’ etc.).