Abstract
Background and aims
This study characterized chasing behaviour as the time to return to an online gambling website after a losing or a winning visit.
Methods
We analyzed a naturalistic dataset from an eCasino (PlayNow.com, the provincial platform for British Columbia, Canada), comprising 1,909,681 sessions from 15,544 individuals. Analyses distinguished sessions on slot machines, blackjack, roulette, video poker, probability games, or mixed-category sessions.
Results
Overall, gamblers on most games returned more slowly as a function of the prior loss, and more quickly as a function of the prior win. Loss chasing intensities in blackjack, probability, video poker, and mixed sessions did not differ significantly from slot machines, but roulette was associated with shorter intervals to return (b = −0.13, p < 0.001). Similarly, win chasing did not vary across slot machines, blackjack, probability games, and video poker, but roulette (b = −0.08, p < 0.001) and mixed (b = −0.02, p = 0.009) sessions were associated with shorter intervals.
Discussion and conclusions
The average behavioural patterns provide limited evidence for loss chasing but clearly indicate win chasing. Although slot machines are commonly considered a high-risk product, roulette in our analyses was associated with the greatest chasing intensities.
Introduction
Chasing is widely considered one of the hallmarks of disordered gambling (Lesieur, 1979; Zhang & Clark, 2020). Broadly speaking, it refers to the persistence or escalation of betting in an effort to recover the gambler's debts. From this perspective, chasing is conventionally seen as a response to losing. However, it has also been noted that wins can drive chasing, and exacerbate gambling problems (Delfabbro, King, & Griffiths, 2014; O'Connor & Dickerson, 2003; Young, Wohl, Matheson, Baumann, & Anisman, 2008). Given the house edge (negative expectancy) of commercial gambling products, chasing after wins is likely to increase cumulative losses, as well as providing further opportunity for habit formation (Ferrari, Limbrick-Oldfield, & Clark, 2022). In past work, win chasing was associated with impaired control over gambling (O'Connor & Dickerson, 2003), and high-risk gamblers reported greater gambling desire after wins compared to losses (Young et al., 2008). Hence, examining chasing after wins is also important for understanding the development of gambling problems. In the DSM-5, the chasing item specifically refers to a gambler who “often returns another day to get even” (section 312.31; American Psychiatric Association et al., 2013). Indeed, this ‘between-session chasing’ is the most frequently endorsed item among the DSM items for Gambling Disorder (Sleczka & Romild, 2021; Toce-Gerstein, Gerstein, & Volberg, 2003).
This study primarily aimed to characterize between-session chasing in a large ‘behavioural tracking’ dataset of online gamblers, examining the time interval between consecutive sessions, as a function of both the amounts lost (loss chasing) and amounts won (win chasing). In light of the close links between chasing and gambling problems, the study also compared chasing intensities across gambling product types, to derive insight into the potential risks associated with different gambling forms.
Most research on between-session chasing to date used retrospective surveys (Gainsbury, Suhonen, & Saastamoinen, 2014; Lesieur, 1979; O'Connor & Dickerson, 2003; Sleczka & Romild, 2021; Temcheff, Paskus, Potenza, & Derevensky, 2016). Temcheff et al. (2016) surveyed 8,674 college women athletes, only the chasing item could distinguish those with and without gambling problems. A Swedish longitudinal survey highlighted that the endorsement of chasing is a stable predictor of the risk of transitioning to more severe gambling problems over 5 years (Sleczka & Romild, 2021). However, self-report measures are subject to a number of biases (Braverman, Tom, & Shaffer, 2014). A small number of studies have examined chasing using field data from (land-based) casinos (Flepp, Meier, & Franck, 2021; Forrest & McHale, 2016; Kainulainen, 2020; Narayanan & Manchanda, 2012; Wardle, Excell, Ireland, Ilic, & Sharman, 2014). Measuring the time interval between visits as an index of chasing, these studies found that gamblers took longer to return after a losing session, and shorter intervals to return after winning (Forrest & McHale, 2016; Kainulainen, 2020; Narayanan & Manchanda, 2012). Chasing might also be expressed as increasing the overall bet amount over successive visits. Tracking slot machine engagement in a Swiss casino, a small loss (i.e., below USD 188) did not change the subsequent bet amount, whereas a larger loss reduced the subsequent bet amount (Flepp et al., 2021). Thus, the aggregate profile in land-based gambling indicates win chasing, but not loss chasing.
This study sought to capture between-session chasing in the online environment, which presents a direct means of linking gamblers' accounts with their behaviour, compared to tracking gamblers in land-based gambling venues (Deng, Lesch, & Clark, 2019). We used a dataset from the PlayNow.com eCasino, the provincial gambling platform in British Columbia (BC), Canada. Across many jurisdictions, online gamblers and land-based gamblers often have different demographic backgrounds (Gainsbury, Wood, Russell, Hing, & Blaszczynski, 2012, 2015). Online gamblers also have a greater risk of disordered gambling than land-based gamblers (e.g., Papineau et al., 2018). In a 2020 prevalence survey in BC, Canada, 24% of online gamblers were classified as high-risk for gambling problems on the Problem Gambling Severity Index (Ipsos & Strategic Science, 2020). Prior studies examining European online gambling datasets have considered the operationalization of chasing behaviour (Auer & Griffiths, 2022; Challet-Bouju et al., 2020; Perrot, Hardouin, Grall-Bronnec, & Challet-Bouju, 2018). Challet-Bouju et al. (2020) and Perrot et al. (2018) inferred chasing from successive money deposits within a short period of time or a deposit within 1 h after placing a bet. Auer and Griffiths (2022) defined five alternative metrics of loss chasing, including an ‘across days’ chasing measure based on the correlation between the amount lost and the subsequent amount bet, restricted to pairs of consecutive gambling days. In their study, another metric reflecting the percentage of sessions with more than one financial deposit appeared most sensitive to chasing (Auer & Griffiths, 2022). We note these studies did not quantify chasing intensities in relation to winning outcomes.
Our further objective was to examine chasing differences between gambling product categories. Past behavioural tracking studies rarely distinguished games (Forrest & McHale, 2016; Wardle et al., 2014). Some studies have focussed on single gambling products (Flepp et al., 2021; Kainulainen, 2020; Narayanan & Manchanda, 2012), where it is unclear how observed patterns would generalise to other games. Different games vary in their structural characteristics (Griffiths, 1993), in ways that are likely to influence chasing and the potential for disordered gambling. For example, slot machines have a fast and continuous speed of play, and intense audiovisual stimulation, which foster psychological states of immersion (Dixon et al., 2014, 2018; Dowling et al., 2017; Murch & Clark, 2021). In a survey, slot machine gamblers were more likely to “keep playing to try to win back their losses”, compared to roulette and blackjack gamblers (Gainsbury et al., 2014). Slot machines were the most popular game type in our dataset, representing 57% of sessions, and they were used as the reference category.
We hypothesized that among online gamblers, the time interval between sessions would decrease as a function of the amount lost in the prior session (H1: loss chasing), and would decrease as a function of the amount won in the prior session (H2: win chasing). Furthermore, our analyses distinguished game types of slot machines, blackjack, roulette, video poker, and probability games (e.g., pachinko, reactors), and we hypothesized that slot machine sessions would be associated with the shortest between-session intervals, as a function of the amount lost (H3: loss chasing by game type) and amount won (H4: win chasing by game type) in the prior session.
Methods
Data overview
The study used a behavioural tracking dataset obtained from the eCasino section of PlayNow.com, the provincial gambling platform for BC, Canada, which is restricted to BC residents. The website requires customers to create a user account, which allows the website to track individual gamblers' bet-by-bet behaviour with timestamps. The dataset was de-identified by the British Columbia Lottery Corporation (BCLC) Data Analytics team by randomly assigning each gambler a unique ID. The dataset spans from 2014-10-01 to 2015-08-31, comprising 527,015,222 individual bets placed by 29,964 gamblers. During the time period under scrutiny, the eCasino contained five game categories, which further comprised 240 specific products: slot machines (n = 195 individual products), roulette (n = 8), blackjack (n = 9), video poker (n = 13), and probability games (n = 15).
Our analysis code is on https://osf.io/dcv65/. We began by aggregating the bet-by-bet data into sessions. We define a session as a period of activity that begins with the first bet and ends with the last bet before the gambler is inactive for 30 min, in which case they are logged off automatically. This time interval reflects the gambler's log-on and log-off periods on the website. During a session, gamblers can play one game type (79% of sessions), or place bets across multiple game categories, termed as “mixed” sessions (21% of sessions). Thus, the aggregated session data comprised six types: slot machines, roulette, blackjack, video poker, probability games, and mixed sessions. Slot machines were the most popular product type in the eCasino (57% of sessions, see Table 1).
Descriptive statistics of session data by game type
| Game | Session count (%) | Gambler count (%) | |
| Slot machines | 1,081,499 (56.63%) | 12,285 (79.03%) | |
| Mixed sessions | 412,858 (21.62%) | 12,673 (81.53%) | |
| Probability games | 184,933 (9.68%) | 5,870 (37.36%) | |
| Blackjack | 141,796 (7.43%) | 4,420 (28.44%) | |
| Video poker | 48,691 (2.55%) | 1,264 (8.13%) | |
| Roulette | 39,904 (2.09%) | 2,402 (15.45%) | |
| Outcome (Canadian $) | |||
| Game | Median | Mean | SD |
| Mixed sessions | −35.04 | −116.58 | 874.02 |
| Slot machines | −26.95 | −97.58 | 633.69 |
| Blackjack | −19.96 | −104.39 | 399.74 |
| Video poker | −17.92 | −99.68 | 642.14 |
| Probability games | −12.10 | −43.70 | 784.11 |
| Roulette | −10.00 | −79.58 | 355.21 |
| Time to return (hours) | |||
| Game | Median | Mean | SD |
| Roulette | 144.86 | 1,111.26 | 4,193.87 |
| Blackjack | 139.93 | 913.43 | 3,414.25 |
| Slot machines | 127.05 | 529.91 | 2,081.96 |
| Mixed sessions | 117.51 | 524.84 | 2,230.79 |
| Video poker | 100.26 | 435.61 | 2,057.99 |
| Probability games | 86.29 | 407.46 | 1,784.55 |
Note: Gambler count is non-exclusive in game played. For example, the same person can have both slot machine sessions and probability game sessions.
The analysis included gamblers who visited the eCasino more than five times. This represents an arbitrary threshold (although see Finkenwirth, MacDonald, Deng, Lesch, & Clark, 2021; Percy, França, Dragičević, & d'Avila Garcez, 2016), but between-session chasing inherently requires more than one session, and it is challenging to use data from gamblers with low levels of activity in modelling behavioural markers of high-risk gambling. We excluded a further gambler who was an outlier in terms of high bet frequency. The analytical sample included 1,909,681 sessions from 15,544 gamblers.
Statistical analysis
Variables
The analysis aimed to measure how prior session outcomes impact the following time to return to the eCasino. The time interval (Time to Return, expressed in hours) between the end of the prior session and the start of the next session is the dependent variable. A faster time to return reflects a greater chasing tendency. We applied a natural log transform on Time to Return because of its positive skewed nature in distribution, with most gamblers returned in modest time intervals, but some gamblers returned to the eCasino after extremely long intervals. The key predictors were:
- •Game: game types included slot machines, probability games, blackjack, video poker, roulette, and mixed sessions.
- •Outcome Dummy: a dummy variable indicated the outcome valence - win or loss, which might lead to differential chasing patterns. A negative net balance refers to a net loss. A positive net balance, which also includes zeros as ‘break-even’ sessions, refers to a net win. We re-parametrized the model in two ways: we used loss as Outcome Dummy reference (loss = 0, win = 1) to interpret loss chasing, and win as Outcome Dummy reference (win = 0, loss = 1) to interpret win chasing.
- •Outcome: the absolute values of net outcome, which is the total paid amount minus the winning amount in a session (see also Leino et al., 2016). Outcome values were standardized with respect to individual gamblers' mean of loss or win amounts; depending on Outcome Dummy, a value of zero indicates an average loss (or win) amount for that gambler over the 11-month data. This standardized within person method is in line with our research interest in individual gamblers' responses to the prior outcome when winning or losing more than their personal average.
Additionally, we covaried for other time-related nuisance variables that could impact the timing of a new session:
- •Session Order: the order of the current session on any given day, included in analyses as a covariate. We tested models that had linear or quadratic effects of raw or (natural) log-transformed session order, and ultimately used a linear effect of log-transformed Session Order as it yielded the best fit via BIC.
- •Start hour: we anchored time of the day by the session's start time and classify it into the following periods: morning (12:00 AM–8:59 AM), early daytime (9:00 AM–2:59 PM), late daytime (3:00 PM–6:59 PM), night (7:00 PM–11:59 PM).
- •Weekends: a dummy variable indicating whether the session occurred on a weekday (No = 0) or a weekend (Yes = 1).
- •BC statutory holidays: a dummy variable indicating whether the session occurred on a holiday (Yes = 1) or not (No = 0).
Analysis
Analyses used R (Version 4.1.1; R Core Team, 2021) and Python (Version 3.6). The data structure is in two-level hierarchical: level 1 contains the variables we introduced earlier, which are clustered by gambler ID at level 2. Since there could be random differences in sessional variables between gamblers, we used multilevel linear modelling. An alternative method to analyze time-to-event data is survival analysis, which estimates the probability of a target event occurring. We considered this approach but our dataset comprises multiple events per participant, and our hypothesis refers to the time interval between events, as a continuous dependent variable, which is better suited to the multilevel linear model.
We used the lme4 package in R (Bates, Mächler, Bolker, & Walker, 2015) for modelling. We use p = 0.01 as the cut-off alpha level to determine statistical significance given our large data size. Log-transforming the dependent variable makes the model estimates not intuitive for interpretation, thus we transformed estimated coefficients to the percentage of changes using
Ethics
The study is an analysis on the de-identified secondary gambling data, which were given to the Centre by the BCLC under a Non-Disclosure Agreement that does not allow data sharing or any reporting of information from individual users. The University of British Columbia's Behavioural Research Ethics Board gave ethical approval to store and analyse the secondary dataset.
Results
Descriptive results
We report medians due to the heavy skewness of the data (Table 1). On average, gamblers lost the most (Median = $-35.04) in mixed sessions. Slot machines – as the most played game category - ranked second (Median = $-26.95). Gamblers lost the least on roulette (Median = $-10.00). In terms of the time to return, gamblers returned the slowest overall after roulette sessions (Median = 144.86 h) and returned the fastest after probability game sessions (Median = 86.29 h). The time to return was intermediate for slot machine sessions (Median = 127.05 h).
Time to return as a function of prior session outcome
Figure 1 shows the time to return as a function of the prior loss and win. The x-axis depicts the standardized loss (win) amount based on the distribution of outcomes for that individual gambler over the data window. Zero constitutes their personal average loss (win) amount, and a positive outcome indicates the gambler lost (won) more than their personal average. A downward slope indicates that gamblers return faster as a function of larger prior loss (win), i.e. chasing, whereas an upward slope indicates that gamblers return slower after larger losses (wins).
Between-session chasing by game type
Note: The shaded area is standard errors.
Citation: Journal of Behavioral Addictions 13, 2; 10.1556/2006.2024.00022
After losing the prior session, gamblers returned more slowly across all game types, with the exception of roulette (Table 2). For a standard deviation increase in the prior session net loss, average gamblers playing slot machine sessions were estimated to took 8.59% longer to return to the platform (Table 3a). Compared to slot machines, blackjack (p = 0.740), probability games (p = 0.157), video poker (p = 0.103) on average did not differ significantly in the loss slopes. Mixed sessions on average were marginally faster in comparison to slot machine sessions (p = 0.015), but this was not significant at the conservative threshold. For roulette, average gamblers were estimated to return faster overall as a function of the amount lost: a standard deviation increase in the prior loss was estimated to reduce the time to return by approximately 4.78%.1 The downward slope for roulette was significantly less steep than the upward slope for slot machines (p < 0.001), indicating that gamblers in roulette sessions were less sensitive to the prior loss than gamblers in slot machine sessions. The overall model yielded an intraclass correlation coefficient (ICC) of 0.36, a conditional R2 of 0.39, and a marginal R2 of 0.05, indicating that fixed and random effects together explained 39% of the variance in time to return, and fixed effect alone explained 5% of the variance.
Summary of between-session across game types
| Time to return (hours) | ||
| Loss chasing | Win chasing | |
| Slot machines | ✗ | ✓ |
| Mixed sessions | ✗ | ✓* |
| Roulette | ✓* | ✓* |
| Blackjack | ✗ | ✓ |
| Probability games | ✗ | ✓ |
| Video poker | ✗ | ✓ |
Note: ‘✓’ indicates a numerical effect in the direction of chasing; ‘✗’ indicates a directional absence of chasing under the measurement. ‘*’ indicates a significant difference (p < 0.01) in chasing relative to slot machines as the reference category.
Between-session chasing regression results
| a: the reference of loss | ||||||||
| Term | estimate | std.error | statistic | df | p value | conf.low | conf.high | transformed estimate |
| (Intercept) | 5.78 | 0.01 | 568.55 | 18,612.91 | <0.001 | 5.76 | 5.80 | 32,420.96 |
| Outcome | 0.08 | 0.00 | 24.61 | 10,311.00 | <0.001 | 0.08 | 0.09 | 8.59 |
| Outcome Dummy (Win = 1) | −0.70 | 0.00 | −198.74 | 1,881,406.20 | <0.001 | −0.70 | −0.69 | −50.23 |
| Blackjack | 0.14 | 0.01 | 13.15 | 1,204,495.17 | <0.001 | 0.12 | 0.16 | 14.52 |
| Probability | −0.14 | 0.01 | −21.90 | 1,799,009.01 | <0.001 | −0.15 | −0.13 | −13.13 |
| Mixed | −0.01 | 0.00 | −2.48 | 1,894,476.10 | 0.013 | −0.02 | 0.00 | −1.03 |
| Roulette | 0.31 | 0.02 | 19.09 | 1,212,709.45 | 0.000 | 0.28 | 0.34 | 35.96 |
| Video poker | 0.01 | 0.01 | 0.54 | 1,583,210.18 | 0.586 | −0.02 | 0.04 | 0.82 |
| Outcome * Outcome Dummy (Win = 1) | −0.15 | 0.00 | −30.51 | 7,102.44 | <0.001 | −0.16 | −0.14 | −14.06 |
| Outcome * Game (Loss chasing by Game) | ||||||||
| Blackjack | 0.00 | 0.01 | −0.33 | 36,217.27 | 0.740 | −0.02 | 0.01 | −0.30 |
| Probability | 0.01 | 0.01 | 1.42 | 166,567.99 | 0.157 | 0.00 | 0.02 | 1.02 |
| Mixed | 0.01 | 0.00 | 2.44 | 190,163.82 | 0.015 | 0.00 | 0.02 | 0.97 |
| Roulette | −0.13 | 0.02 | −8.41 | 67,375.14 | <0.001 | −0.16 | −0.10 | −12.31 |
| Video poker | 0.02 | 0.02 | 1.63 | 51,664.05 | 0.103 | −0.01 | 0.05 | 2.51 |
| Outcome Dummy (Win = 1) * Game | ||||||||
| Blackjack | −0.15 | 0.01 | −16.18 | 1,896,136.80 | <0.001 | −0.17 | −0.14 | −14.36 |
| Probability | 0.10 | 0.01 | 10.58 | 1,545,180.88 | <0.001 | 0.08 | 0.12 | 10.72 |
| Mixed | 0.06 | 0.01 | 8.59 | 1,885,785.52 | <0.001 | 0.04 | 0.07 | 5.88 |
| Roulette | −0.29 | 0.02 | −16.55 | 1,837,680.26 | <0.001 | −0.33 | −0.26 | −25.50 |
| Video poker | 0.04 | 0.02 | 2.69 | 1,845,311.57 | 0.007 | 0.01 | 0.07 | 4.42 |
| Outcome * Outcome Dummy (Win = 1) * Game | ||||||||
| Blackjack | −0.01 | 0.01 | −0.71 | 14,101.80 | 0.477 | −0.03 | 0.02 | −0.90 |
| Probability | −0.02 | 0.01 | −1.60 | 19,349.36 | 0.109 | −0.05 | 0.01 | −2.24 |
| Mixed | −0.03 | 0.01 | −3.60 | 30,746.84 | <0.001 | −0.04 | −0.01 | −2.60 |
| Roulette | 0.05 | 0.02 | 2.53 | 27,129.20 | 0.011 | 0.01 | 0.10 | 5.59 |
| Video poker | 0.03 | 0.02 | 1.12 | 9,090.02 | 0.261 | −0.02 | 0.07 | 2.74 |
| Time nuisance | ||||||||
| BC holidays (Yes = 1) | 0.04690 | 0.01 | 7.27 | 1,889,227.59 | <0.001 | 0.03 | 0.06 | 4.80 |
| Early daytime 9am–2pm | −0.08159 | 0.00 | −22.45 | 1,895,871.95 | <0.001 | −0.09 | −0.07 | −7.84 |
| Late daytime 3pm–6pm | 0.02936 | 0.00 | 7.73 | 1,897,520.24 | <0.001 | 0.02 | 0.04 | 2.98 |
| Night | 0.52839 | 0.00 | 140.19 | 1,900,153.44 | <0.001 | 0.52 | 0.54 | 69.62 |
| Weekend (Yes = 1) | 0.04741 | 0.00 | 18.78 | 1,892,680.48 | <0.001 | 0.04 | 0.05 | 4.86 |
| Log(Session count) | −0.22919 | 0.00 | −93.33 | 1,897,962.69 | <0.001 | −0.23 | −0.22 | −20.48 |
| b: the reference of win | ||||||||
| Term | estimate | std.error | statistic | df | p value | conf.low | conf.high | transformed estimate |
| (Intercept) | 5.09 | 0.01 | 485.57 | 20,924.79 | <0.001 | 5.07 | 5.11 | 16,084.84 |
| Outcome | −0.07 | 0.00 | −16.07 | 6,344.69 | <0.001 | −0.08 | −0.06 | −6.68 |
| Outcome Dummy (Loss = 1) | 0.70 | 0.00 | 198.74 | 1,881,406.18 | <0.001 | 0.69 | 0.70 | 100.93 |
| Blackjack | −0.02 | 0.01 | −1.79 | 1,262,439.64 | 0.073 | −0.04 | 0.00 | −1.92 |
| Probability | −0.04 | 0.01 | −4.11 | 1,396,047.74 | <0.001 | −0.06 | −0.02 | −3.82 |
| Mixed | 0.05 | 0.01 | 7.45 | 1,886,143.90 | <0.001 | 0.03 | 0.06 | 4.79 |
| Roulette | 0.01 | 0.02 | 0.75 | 1,305,816.86 | 0.452 | −0.02 | 0.05 | 1.29 |
| Video poker | 0.05 | 0.02 | 2.86 | 1,691,066.32 | <0.001 | 0.02 | 0.09 | 5.27 |
| Outcome * Outcome Dummy (Loss = 1) | 0.15 | 0.00 | 30.51 | 7,102.47 | <0.001 | 0.14 | 0.16 | 16.36 |
| Outcome * Game (Win chasing by Game) | ||||||||
| Blackjack | −0.01 | 0.01 | −1.22 | 9,578.00 | 0.222 | −0.03 | 0.01 | −1.20 |
| Probability | −0.01 | 0.01 | −0.99 | 15,655.75 | 0.324 | −0.04 | 0.01 | −1.24 |
| Mixed | −0.02 | 0.01 | −2.63 | 23,893.85 | 0.009 | −0.03 | 0.00 | −1.66 |
| Roulette | −0.08 | 0.02 | −4.75 | 14,347.93 | <0.001 | −0.11 | −0.05 | −7.40 |
| Video poker | 0.05 | 0.02 | 2.51 | 7,325.55 | 0.012 | 0.01 | 0.09 | 5.32 |
| Outcome Dummy (Loss = 1) * Game | ||||||||
| Blackjack | 0.15 | 0.01 | 16.18 | 1,896,136.83 | <0.001 | 0.14 | 0.17 | 16.76 |
| Probability | −0.10 | 0.01 | −10.58 | 1,545,183.59 | <0.001 | −0.12 | −0.08 | −9.68 |
| Mixed | −0.06 | 0.01 | −8.59 | 1,885,785.57 | <0.001 | −0.07 | −0.04 | −5.55 |
| Roulette | 0.29 | 0.02 | 16.55 | 1,837,680.76 | <0.001 | 0.26 | 0.33 | 34.23 |
| Video poker | −0.04 | 0.02 | −2.69 | 1,845,312.00 | 0.007 | −0.07 | −0.01 | −4.23 |
| Outcome * Outcome Dummy (Loss = 1) * Game | ||||||||
| Blackjack | 0.01 | 0.01 | 0.71 | 14,101.90 | 0.477 | −0.02 | 0.03 | 0.91 |
| Probability | 0.02 | 0.01 | 1.60 | 19,349.58 | 0.109 | −0.01 | 0.05 | 2.29 |
| Mixed | 0.03 | 0.01 | 3.60 | 30,747.18 | <0.001 | 0.01 | 0.04 | 2.67 |
| Roulette | −0.05 | 0.02 | −2.53 | 27,129.38 | 0.011 | −0.10 | −0.01 | −5.30 |
| Video poker | −0.03 | 0.02 | −1.12 | 9,090.11 | 0.261 | −0.07 | 0.02 | −2.66 |
| Time nuisance | ||||||||
| BC holidays (Yes = 1) | 0.05 | 0.01 | 7.27 | 1,889,227.57 | <0.001 | 0.03 | 0.06 | 4.80 |
| Early daytime 9am–2pm | −0.08 | 0.00 | −22.45 | 1,895,871.93 | <0.001 | −0.09 | −0.07 | −7.84 |
| Late daytime 3pm–6pm | 0.03 | 0.00 | 7.73 | 1,897,520.23 | <0.001 | 0.02 | 0.04 | 2.98 |
| Night | 0.53 | 0.00 | 140.19 | 1,900,153.43 | <0.001 | 0.52 | 0.54 | 69.62 |
| Weekend (Yes = 1) | 0.05 | 0.00 | 18.78 | 1,892,680.47 | <0.001 | 0.04 | 0.05 | 4.86 |
| Log(Session count) | −0.23 | 0.00 | −93.33 | 1,897,962.67 | <0.001 | −0.23 | −0.22 | −20.48 |
| Conditional R-squared | 0.39 | |||||||
| Marginal R-squared | 0.05 | |||||||
Note: panel (a) as losses as the reference level, and panel (b) has wins as the reference level.
For the analyses of win chasing, average gamblers were estimated to return faster as a function of the prior amount won across all game types, thus indicating a propensity for win chasing (see Fig. 1, Table 3b). For a standard deviation increase in the prior win, average slot machine gamblers were estimated to take 6.68% less time to return. The rates of win chasing differed by game types: for mixed sessions and roulette, the slope for win chasing was on average significantly steeper than for slot machines (mixed, p = 0.009; roulette, p < 0.001). For mixed sessions, a one standard deviation increase was associated with an average 8.23% shorter time to return, and for roulette, an average 13.59% shorter time to return. Blackjack (p = 0.222), probability (p = 0.324), and video poker (p = 0.012) sessions on average did not differ significantly from slot machine sessions.
Discussion
The DSM-5 operationalizes chasing as often returning to the casino another day (American Psychiatric Association et al., 2013), termed between-session chasing. As a diagnostic criterion, it is arguably the only behaviourally-observable item used in the identification of Gambling Disorder, and it is considered a defining hallmark of problematic gambling (Gainsbury et al., 2014). By analyzing timestamped online gambling data, our analyses refine the understanding of between-session chasing by measuring the time taken for gamblers to return to the platform after winning or losing sessions, across a number of different gambling products. Overall, the fixed factors alone, which included outcome and game type, as well as the time-related nuisance variables, only explained a small portion (5%) of the variance in the time to return. In combination with the level 2 random factors that allowed the degree of win and loss chasing to differ randomly across subjects, 39% of the variance in the time to return was explained. This moderate effect highlights the high degree of individual variation in chasing behaviours. On an aggregate level, we saw that gamblers on most game types returned more slowly after losing sessions, and returned more quickly after winning sessions. These findings are consistent with previous research that measured the interval between sessions for land-based casinos, betting shops, and online horse-racing games (Forrest & McHale, 2016; Kainulainen, 2020; Narayanan & Manchanda, 2012). In our data, roulette was a notable exception: this game had the longest intervals between sessions overall, but was the only game to show loss chasing as a function of greater amounts lost, as well as significantly steeper slopes (in comparison to slot machines) for both loss and win chasing.
In sharp contrast to loss chasing, win chasing – defined as a faster time to return as a function of the previous amount won – was a consistent pattern here across all game types. In a previous field study from Swiss land-based casinos, windfall wins carried over to increase the amount bet on the subsequent casino visit (Rüdisser, Flepp, & Franck, 2017). One possible explanation for these win chasing patterns is the so-called house money effect (Thaler & Johnson, 1990), which describes increased risk-taking because windfalls are not yet internalized as the gamblers' own funds; any loss of such windfalls would not hurt as much as the loss of one's own funds (Peng, Miao, & Xiao, 2013). As a cognitive explanation, one might expect such effects to be short-lived, whereas we observe between-session chasing of wins over the order of days, by which time the win should have been internalized as one's own. In our view, this may point to other explanations, such as a wealth effect, by which an increase in personal wealth enables larger spending in the future (Mehra, 2001). In the context of gambling sessions, prior wins increase financial resources to allow a faster time to return.
In contrast, we did not observe faster return times after losing sessions for most game types. With a greater loss, gamblers took longer to return to the website, indicating an absence of loss chasing in the aggregate data. In the break-even effect, individuals typically increase risk-taking when losing in an effort to recover their losses, but this should only occur when the losses are recoverable (Thaler & Johnson, 1990). In our data, the time intervals to return did not get faster with increasing amounts lost (except for roulette, discussed below), and this might be because further risk-taking did not offer the potential to break-even. We note that this explanation requires significant cognitive processing, to calculate the current difference from one's reference point (e.g., their starting account balance (Imas, 2016)), and the possible impact of a large win to offset that difference. Alternatively, longer return times after losing could be driven by financial constraints: significant losses may simply deplete financial resources to continue gambling (i.e. the converse of the ‘wealth effect’ described above for win chasing). Times to return may also be exogenously influenced by financial factors such as the next payday (Dahan, 2019). In future research, these explanations could be disambiguated by merging gambling data with banking information (Muggleton et al., 2021): for example, if the longer return times are due to insufficient funds, gamblers may be more likely deposit funds into their gambling account following pay days, which would be visible in the banking data.
Slot machines are considered to have among the highest risk potential for any form of gambling (Meyer, Fiebig, Häfeli, & Mörsen, 2011). Accordingly, we expected slot machine sessions to be associated with the shortest times to return. Overall, we found little support for this hypothesis: times to return did not differ significantly between blackjack, probability, video poker, and slot machines. Notably, roulette displayed shorter times to return than slot machines, as a function of both losses and wins, and roulette was the only game type where time to return decreases as a function of the prior amount lost. This apparent uniqueness of roulette should be considered in the context of roulette having the longest overall times to return of any product type, as well as a smaller net loss than other games (Table 1). These differences may render roulette gamblers more sensitive to the moderation by prior outcome, such that each unit of loss would result in larger change in time to return. In terms of structural characteristics, we note that these effects pertain to online roulette, and this is a faster game than land-based (i.e. casino) roulette, and is also accompanied by more intense audiovisual feedback. There is limited research considering the specific product risk associated with online roulette, or indeed for any individual types of eCasino products. It is possible that online gamblers who prefer roulette may display distinct psychological characteristics; for example, Bonnaire, Bungener, and Varescon (2009) reported that gamblers who preferred table games, including roulette, showed lower levels of depression and alexithymia compared to gamblers who preferred slot machines and racetrack gambling. Our findings for roulette arose from exploratory analyses, and as our predictions were of intensified chasing on slot machines, these findings for online roulette clearly warrant replication. Nevertheless, they do provide evidence for an overarching hypothesis that chasing tendencies differ by product category, which is relevant to developing algorithms for detecting high-risk gambling from behavioural data (Edson et al., 2023; Ghaharian, Abarbanel, Kraus, Singh, & Bernhard, 2023). We also recognize that gamblers at risk for gambling problems typically engage with a range of product types (LaPlante, Nelson, & Gray, 2014) such that future work in this area should give attention to both the expressions of chasing on specific gambling products, but also the characteristics of gamblers who chase.
In this study, we quantified the DSM-5 criterion for chasing – the tendency to return another day (“to get even”) – as the time interval between successive online gambling sessions. Previous research (Auer & Griffiths, 2022; Challet-Bouju et al., 2020; Perrot et al., 2018) has focused on increased spending or deposits over time as expressions of chasing. Our study operationalizes the diagnostic item for gambling disorder to derive a concrete behavioural indicator for high-risk gambling. This approach further allows for the characterization of chasing after wins and losses, showing how between-session chasing varies by session outcome.
A number of limitations should be noted. Our data back to 2014–2015 and the online gambling landscape has continued to evolve since that time. We lack demographic descriptives for this de-identified dataset. As with most analyses of behavioural tracking data, it is possible that customers within our dataset held accounts on other gambling platforms (The Behavioural Insights Team, 2021), and visits to other websites would not be represented in our time to return variable. Mitigating this limitation, PlayNow is the only licensed online gambling website in BC, compared to other jurisdictions with many licensed operators. Second, sessions were defined based on a cut-off of 30 min of inactivity on the website, which is somewhat arbitrary. For example, a gambler may sporadically play over the course of a day, with breaks of an hour or longer; should this be classified as one session or several? In principle, the same issue can arise in land-based casinos, e.g., when visiting the bar or ATM. Another limitation is that our analyses focus on time to return as a specific expression of between-session chasing, which may alternatively be measured in the bet volume over successive sessions (Auer & Griffiths, 2022; Flepp et al., 2021). Further behavioural markers exist to characterize within-session chasing tendencies (Chen, Doekemeijer, Noël, & Verbruggen, 2022; O’Connor & Dickerson, 2003). We are exploring these alternative expressions in ongoing research, which also incorporates self-exclusion data (Deng, Lesch, & Clark, 2021; Finkenwirth et al., 2021) as a marker of disordered gambling. Lastly, as noted by a reviewer, mixed sessions represent an interesting instance of a multiple (or chained) reinforcement schedule (Saini, Miller, & Fisher, 2016) controlled by the participant. Even single-category sessions could still involve the gambler switching between specific games, such as different slot machines. The product categories, and within-category products, will vary in structural characteristics, some of which are highly relevant to chasing; for example higher volatility games are associated with longer losing streaks, which may prompt switching products (Delfabbro, King, & Parke, 2023). Future research may drill down further on these session characteristics, viewed from the perspective of operant behaviour.
Funding sources
The Centre for Gambling Research at UBC is supported by the Province of British Columbia government and the British Columbia Lottery Corporation (BCLC; a Canadian Crown Corporation). LC holds a Discovery Award from the Natural Sciences and Engineering Research Council (Canada). KZ held the Graduate Fellowship in Gambling Research, a fellowship supported by the BCLC and adjudicated by the UBC Faculty of Arts.
Author's contributions
KZ conceptualized ideas, analysed data, and drafted the manuscript. JR guided data analysis and manuscript review & editing. XD contributed to data management and server maintenance. TL contributed to conceptualization, data analysis, and manuscript review. LC supervised the project and contributed to manuscript review & editing. All authors except for JR did not have access to data. This study used a secondary dataset provided by BCLC to the Centre for Gambling Research, and JR's guidance on the analysis was based on a simulated dataset. All authors have approved the final manuscript.
Conflict of interests
KZ held the Graduate Fellowship in Gambling Research (2021–2022), a fellowship supported by the BCLC and adjudicated by the UBC Faculty of Arts.
LC is the Director of the Centre for Gambling Research at UBC, which is supported by funding from the Province of British Columbia and the BCLC. The Province of BC government and the BCLC had no role in the preparation of this manuscript, and imposed no constraints on publishing. LC has received a speaker/travel honorarium from the International Center for Responsible Gaming (US) and Scientific Affairs (Germany), and has received fees for academic services from the International Center for Responsible Gaming (US), GambleAware (UK), Gambling Research Australia, and Gambling Research Exchange Ontario (Canada). He has been remunerated for legal consultancy by the BCLC. LC receives an honorarium for his role as Co-Editor-in-Chief for International Gambling Studies from Taylor & Francis, and he has received royalties from Cambridge Cognition Ltd. relating to neurocognitive testing. KZ and LC have not received any further direct or indirect payments from the gambling industry or groups substantially funded by gambling. The remaining authors declare no conflict of interest.
Acknowledgements
The dataset for this study was provided to the researchers by the BCLC, under a Non-Disclosure Agreement that prohibits further data sharing. We would like to thank the Social Responsibility and Data Analytics teams at BCLC for their willingness to share the data, and technical support to de-identify and transfer the data.
References
American Psychiatric Association, D. S. M. T. F., & American Psychiatric Association, D. S. (2013). Diagnostic and statistical manual of mental disorders: DSM-5 (5th ed., Vol. 5). Washington, DC: American psychiatric association.
Auer, M., & Griffiths, M. D. (2022). An empirical attempt to operationalize chasing losses in gambling utilizing account-based player tracking data. Journal of Gambling Studies, 1–15. https://doi.org/10.1007/s10899-022-10144-4.
Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67(1), 1–48. https://doi.org/10.18637/jss.v067.i01.
Bonnaire, C., Bungener, C., & Varescon, I. (2009). Subtypes of French pathological gamblers: Comparison of sensation seeking, alexithymia and depression scores. Journal of Gambling Studies, 25(4), 455–471. https://doi.org/10.1007/s10899-009-9142-z.
Braverman, J., Tom, M. A., & Shaffer, H. J. (2014). Accuracy of self-reported versus actual online gambling wins and losses. Psychological Assessment, 26(3), 865–877. https://doi.org/10.1037/a0036428.
Challet-Bouju, G., Hardouin, J. -B., Thiabaud, E., Saillard, A., Donnio, Y., Grall-Bronnec, M., & Perrot, B. (2020). Modeling early gambling behavior using indicators from online Lottery gambling tracking data: Longitudinal analysis. Journal of Medical Internet Research, 22(8), e17675. https://doi.org/10.2196/17675.
Chen, Z., Doekemeijer, R. A., Noël, X., & Verbruggen, F. (2022). Winning and losing in online gambling: Effects on within-session chasing. Plos One, 17(8), e0273359. https://doi.org/10.1371/journal.pone.0273359.
Dahan, M. (2019). Poverty and economic behavior: Gambling at social security paydays. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3467946.
Delfabbro, P., King, D., & Griffiths, M. D. (2014). From adolescent to adult gambling: An analysis of longitudinal gambling patterns in south Australia. Journal of Gambling Studies, 30(3), 547–563. https://doi.org/10.1007/s10899-013-9384-7.
Delfabbro, P., King, D., & Parke, J. (2023). The complex nature of human operant gambling behaviour involving slot games: Structural characteristics, verbal rules and motivation. Addictive Behaviors, 137, 107540. https://doi.org/10.1016/j.addbeh.2022.107540.
Deng, X., Lesch, T., & Clark, L. (2019). Applying data science to behavioral analysis of online gambling. Current Addiction Reports, 6(3), 159–164. https://doi.org/10.1007/s40429-019-00269-9.
Deng, X., Lesch, T., & Clark, L. (2021). Pareto distributions in online casino gambling: Sensitivity to timeframe and associations with self exclusion. Addictive Behaviors, 120, 106968. https://doi.org/10.1016/j.addbeh.2021.106968.
Dixon, M. J., Graydon, C., Harrigan, K. A., Wojtowicz, L., Siu, V., & Fugelsang, J. A. (2014). The allure of multi-line games in modern slot machines. Addiction, 109, 1920–1928. https://doi.org/10.1111/add.12675.
Dixon, M. J., Stange, M., Larche, C. J., Graydon, C., Fugelsang, J. A., & Harrigan, K. A. (2018). Dark flow, depression and multiline slot machine play. Journal of Gambling Studies, 34(1), 73–84. https://doi.org/10.1007/s10899-017-9695-1.
Dowling, N. A., Merkouris, S. S., Greenwood, C. J., Oldenhof, E., Toumbourou, J. W., & Youssef, G. J. (2017). Early risk and protective factors for problem gambling: A systematic review and meta-analysis of longitudinal studies. Clinical Psychology Review, 51, 109–124. https://doi.org/10.1016/j.cpr.2016.10.008.
Edson, T. C., Louderback, E. R., Tom, M. A., Philander, K. S., Slabczynski, J. M., Lee, T. G., & LaPlante, D. A. (2023). A large-scale prospective study of big wins and their relationship with future involvement and risk among actual online sports bettors. Computers in Human Behavior, 142, 107657. https://doi.org/10.1016/j.chb.2023.107657.
Ferrari, M. A., Limbrick-Oldfield, E. H., & Clark, L. (2022). Behavioral analysis of habit formation in modern slot machine gambling. International Gambling Studies, 22(2), 317–336. https://doi.org/10.1080/14459795.2022.2088822.
Finkenwirth, S., MacDonald, K., Deng, X., Lesch, T., & Clark, L. (2021). Using machine learning to predict self-exclusion status in online gamblers on the PlayNow.com platform in British Columbia. International Gambling Studies, 21(2), 220–237. https://doi.org/10.1080/14459795.2020.1832132.
Flepp, R., Meier, P., & Franck, E. (2021). The effect of paper outcomes versus realized outcomes on subsequent risk-taking: Field evidence from casino gambling. Organizational Behavior and Human Decision Processes, 165, 45–55. https://doi.org/10.1016/j.obhdp.2021.04.003.
Forrest, D., & McHale, I. G. (2016). Tracked play on B1 gaming machines in British casinos, 66. https://www.begambleaware.org/sites/default/files/2020-12/tracked-play-revision-14-12-16.pdf.
Gainsbury, S., Russell, A., Hing, N., Wood, R., Lubman, D., & Blaszczynski, A. (2015). How the internet is changing gambling: Findings from an Australian prevalence survey. Journal of Gambling Studies, 31(1), 1–15. https://doi.org/10.1007/s10899-013-9404-7.
Gainsbury, S. M., Suhonen, N., & Saastamoinen, J. (2014). Chasing losses in online poker and casino games: Characteristics and game play of Internet gamblers at risk of disordered gambling. Psychiatry Research, 217(3), 220–225. https://doi.org/10.1016/j.psychres.2014.03.033.
Gainsbury, S., Wood, R., Russell, A., Hing, N., & Blaszczynski, A. (2012). A digital revolution: Comparison of demographic profiles, attitudes and gambling behavior of Internet and non-Internet gamblers. Computers in Human Behavior, 28(4), 1388–1398. https://doi.org/10.1016/j.chb.2012.02.024.
Ghaharian, K., Abarbanel, B., Kraus, S. W., Singh, A., & Bernhard, B. (2023). Players Gonna Pay: Characterizing gamblers and gambling-related harm with payments transaction data. Computers in Human Behavior, 143, 107717. https://doi.org/10.1016/j.chb.2023.107717.
Griffiths, M. (1993). Fruit machine gambling: The importance of structural characteristics. Journal of Gambling Studies, 9(2), 101–120. https://doi.org/10.1007/BF01014863.
Imas, A. (2016). The realization effect: Risk-taking after realized versus paper losses. American Economic Review, 106(8), 2086–2109. https://doi.org/10.1257/aer.20140386.
Ipsos, & Strategic Science, & British Columbia Gaming Policy and Enforcement Branch; British Columbia Lottery Corporation. (2020). British Columbia online problem gambling prevalence study. British Columbia government. https://www2.gov.bc.ca/assets/gov/sports-recreation-arts-and-culture/gambling/gambling-in-bc/reports/bc_online_problem_gambling_prevalence_report_2020.pdf.
Kainulainen, T. (2020). Does losing on a previous betting day predict how long it takes to return to the next session of online horse race betting? Journal of Gambling Studies. https://doi.org/10.1007/s10899-020-09974-x.
LaPlante, D. A., Nelson, S. E., & Gray, H. M. (2014). Breadth and depth involvement: Understanding Internet gambling involvement and its relationship to gambling problems. Psychology of Addictive Behaviors: Journal of the Society of Psychologists in Addictive Behaviors, 28(2), 396–403. https://doi.org/10.1037/a0033810.
Leino, T., Torsheim, T., Pallesen, S., Blaszczynski, A., Sagoe, D., & Molde, H. (2016). An empirical real-world study of losses disguised as wins in electronic gaming machines. International Gambling Studies, 16(3), 470–480. https://doi.org/10.1080/14459795.2016.1232433.
Lesieur, H. (1979). The compulsive gambler’s spiral of options and involvement. Psychiatry, 42, 79–87. https://doi.org/10.1080/00332747.1979.11024008.
Mehra, Y. P. (2001). The wealth effect in empirical life-cycle aggregate consumption equations. FRB Richmond Economic Quarterly, 87(2), 45–68. https://www.richmondfed.org/publications/research/economic_quarterly/2001/spring/mehra.
Meyer, G., Fiebig, M., Häfeli, J., & Mörsen, C. (2011). Development of an assessment tool to evaluate the risk potential of different gambling types. International Gambling Studies, 11(2), 221–236. https://doi.org/10.1080/14459795.2011.584890.
Muggleton, N., Parpart, P., Newall, P., Leake, D., Gathergood, J., & Stewart, N. (2021). The association between gambling and financial, social and health outcomes in big financial data. Nature Human Behaviour, 1–8. https://doi.org/10.1038/s41562-020-01045-w.
Murch, W. S., & Clark, L. (2021). Understanding the slot machine zone. Current Addiction Reports, 8(2), 214–224. https://doi.org/10.1007/s40429-021-00371-x.
Narayanan, S., & Manchanda, P. (2012). An empirical analysis of individual level casino gambling behavior. Quantitative Marketing and Economics, Dordrecht, 10(1), 27–62. https://doi.org/10.1007/s11129-011-9110-7.
O’Connor, J., & Dickerson, M. (2003). Definition and measurement of chasing in off-course betting and gaming machine play. Journal of Gambling Studies, New York, 19(4), 359–386 https://doi.org/10.1023/a:1026375809186.
Papineau, E., Lacroix, G., Sévigny, S., Biron, J.-F., Corneau-Tremblay, N., & Lemétayer, F. (2018). Assessing the differential impacts of online, mixed, and offline gambling. International Gambling Studies, 18(1), 69–91. https://doi.org/10.1080/14459795.2017.1378362.
Peng, J., Miao, D., & Xiao, W. (2013). Why are gainers more risk seeking. Judgment and Decision Making, 8(2), 150–160. https://doi.org/10.1017/S1930297500005076.
Percy, C., França, M., Dragičević, S., & d’Avila Garcez, A. (2016). Predicting online gambling self-exclusion: An analysis of the performance of supervised machine learning models. International Gambling Studies, 16(2), 193–210. https://doi.org/10.1080/14459795.2016.1151913.
Perrot, B., Hardouin, J. -B., Grall-Bronnec, M., & Challet-Bouju, G. (2018). Typology of online lotteries and scratch games gamblers’ behaviours: A multilevel latent class cluster analysis applied to player account-based gambling data. International Journal of Methods in Psychiatric Research, 27(4), e1746. https://doi.org/10.1002/mpr.1746.
R Core Team (2021). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/.
Rüdisser, M., Flepp, R., & Franck, E. (2017). When do reference points update? A field analysis of the effect of prior gains and losses on risk-taking over time. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3047800.
Saini, V., Miller, S. A., & Fisher, W. W. (2016). Multiple schedules in practical application: Research trends and implications for future investigation. Journal of Applied Behavior Analysis, 49(2), 421–444. https://doi.org/10.1002/jaba.300.
Sleczka, P., & Romild, U. (2021). On the stability and the progression of gambling problems: Longitudinal relations between different problems related to gambling. Addiction, 116(1), 116–125. https://doi.org/10.1111/add.15093.
Temcheff, C. E., Paskus, T. S., Potenza, M. N., & Derevensky, J. L. (2016). Which diagnostic criteria are most useful in discriminating between social gamblers and individuals with gambling problems? An examination of DSM-IV and DSM-5 criteria. Journal of Gambling Studies, 32(3), 957–968. https://doi.org/10.1007/s10899-015-9591-5.
Thaler, R., & Johnson, E. J. (1990). Gambling with the house money and trying to break even: The effects of prior outcomes on risky choice. Management Science, 36(6), 643–660. https://www.jstor.org/stable/2631898.
The Behavioural Insights Team (2021). Gambling behaviour: What can bank transaction data tell us? A feasilbility study. https://www.bi.team/wp-content/uploads/2021/07/Patterns-of-Play-BIT-HSBC-report-final-June-4th-2021.pdf.
Toce-Gerstein, M., Gerstein, D. R., & Volberg, R. A. (2003). A hierarchy of gambling disorders in the community. Addiction, 98(12), 1661–1672. https://doi.org/10.1111/j.1360-0443.2003.00545.x.
Wardle, H., Excell, D., Ireland, E., Ilic, N., & Sharman, S. (2014). Report 2: Identifying problem gambling – findings from a survey of loyalty card customers. London: National Centre for Social Research. https://www.begambleaware.org/media/1225/report-2-identifying-problem-gambling-findings-from-a-survey-of-loyalty-card-customers.pdf.
Young, M. M., Wohl, M. J. A., Matheson, K., Baumann, S., & Anisman, H. (2008). The desire to gamble: The influence of outcomes on the priming effects of a gambling episode. Journal of Gambling Studies, 24(3), 275–293. https://doi.org/10.1007/s10899-008-9093-9.
Zhang, K., & Clark, L. (2020). Loss-chasing in gambling behaviour: Neurocognitive and behavioural economic perspectives. Current Opinion in Behavioral Sciences, 31, 1–7. https://doi.org/10.1016/j.cobeha.2019.10.006.
