Abstract
Background
Gambling content on streaming platforms has gained popularity. Given their intense, cue-laden nature, watching gambling streams may trigger cravings among viewers. At the same time, people who gamble may be motivated to watch gambling streams in an attempt to regulate their cravings.
Methods
We tested these ideas across two preregistered online studies, recruiting i) people who gamble to compare a subgroup of gambling stream viewers with non-viewers (Study 1; nviewers = 221, nnon-viewers = 642), and ii) a group of gambling stream viewers (Study 2; nviewers = 271).
Results
Gambling stream viewers were younger, tended to identify as men, and displayed higher levels of problem gambling and gambling cravings compared to non-viewers. Problem gambling severity was correlated positively with both the motivation to use gambling streams to regulate cravings and with cravings elicited by watching gambling streams.
Discussion
Our findings indicate that while viewers with higher levels of problem gambling may use gambling streams to regulate their cravings, doing so might evoke cravings.
The proliferation of digital streaming platforms has made watching gameplay a highly accessible form of entertainment. On platforms such as Twitch, which was originally dominated by video game streams, gambling has emerged as one of the most-watched categories, surpassing even highly popular video games (Gwilliam, 2022). Although Twitch has implemented policy curbs on gambling broadcasts, they remain ubiquitous (Gibbs, 2022), and newer streaming platforms such as Kick have successfully attracted a gambling audience by eschewing such restrictions (Taylor-Hill, 2023). Regardless of platform, it is clear that people are drawn to watching gambling. The emergence of gambling streams has fueled controversy on the risks and harms that come with spectatorship. Thus far, much of the debate has focused on the possible risks of exposing young people to gambling content below the legal age for gambling (Koncz, Demetrovics, Griffiths, & Király, 2023). However, gambling streams also attract adult viewers who may already gamble. In the present research, we focus on the potential risks associated with watching gambling streams among people who may already gamble.
On streaming platforms, streamers broadcast their engagement in an activity to an audience (Johnson, 2024). In contrast to traditional forms of media, streams are highly interactive, including elements from social networking sites. Viewers can follow their favorite streamers and use the chat function to communicate with other viewers and with the streamer. These social elements play an important role in motivating people to watch streams (Diwanji et al., 2020; Hilvert-Bruce, Neill, Sjöblom, & Hamari, 2018; Hu, Zhang, & Wang, 2017; Kowert & Daniel, 2021; Nah, 2022; Sjoblom et al., 2017), which, in part, explains the growing popularity of streams, expanding from broadcasts of video gaming to a plethora of other content such gambling.
Streamers try to make their streams as entertaining as possible to grow their audience (Woodcock & Johnson, 2019). In particular, gambling streams contain various events that make for an exciting viewing experience. Streamers broadcast themselves engaging in intense forms of gambling, such as placing high-stake bets, receiving jackpot wins, and suffering heavy losses. By analyzing 21 prerecorded gambling streams on YouTube, Hoebanx and French (2023) identified that a prominent feature of gambling streams (e.g., emphasized in titles and thumbnails) is the highlighting of jackpots. Still, watching gambling streams is distinct from gambling. Despite its highly stimulating nature and the fact that viewers can support streamers through paid subscriptions and donations, gambling streams are free to watch and viewers do not benefit financially even if the streamers win.
Nonetheless, gambling streams contain numerous cues due to their dynamic and multisensory nature (Yalachkov, Kaiser, & Naumer, 2012). According to behavioral theories, gambling-related stimuli elicit cravings through their repeated pairing with the act of gambling, forming Pavlovian associations and conditioned responses. Cravings play an important role in theoretical accounts of the maintenance of addiction. Among people who gamble, cravings are associated with higher levels of problem gambling (Mallorquí-Bagué, Mestre-Bach, & Testa, 2023; Young & Wohl, 2009). Brief exposure to gambling-related cues, such as pictures or short video clips, can elicit self-reported cravings among people with problem gambling (Limbrick-Oldfield et al., 2017; Wulfert, Maxson, & Jardin, 2009). Similarly, in other domains, people with internet gaming disorder report stronger video game cravings after watching footage of their favorite video game compared to a non-favorite game (Ha, Park, Park, Im, & Kim, 2021). These findings suggest that watching gambling streams may elicit gambling cravings, especially for people with higher levels of problem gambling.
At the same time, people may be motivated to watch gambling streams in an attempt to regulate their gambling craving. Qualitative reports show that viewers may watch other people eat tasty foods to satisfy their own food cravings (Anjani, Mok, Tang, Oehlberg, & Goh, 2020; Choe, 2019). Such motivation has also been shown for social casino games which are simulated gambling games that do not involve monetary wagers or prizes (Hollingshead, Kim, Wohl, & Derevensky, 2016, 2020). Among people with problem gambling, the motivation to use social casino games to reduce cravings was correlated with a reduction in reported gambling activity over 6 months (Hollingshead et al., 2016, 2020). We wonder if watching gambling streams may produce vicarious habituation (i.e. repeated exposure reduces a learned response). People tend to eat less food after watching other people eat the same food (Li & Lee, 2023) and are less motivated to complete a laboratory task after watching another person complete the same task (McCulloch, Fitzsimons, Chua, & Albarracín, 2011). Another possibility is that watching gambling streams may discourage gambling because viewers may observe the negative consequences of gambling; By this account, seeing a streamer experience heavy gambling losses might thwart desires to gamble.
To date, limited work has examined gambling streams. In a survey of gambling-like activities in a nationally-representative sample in the United Kingdom (n = 1,081), 4% of participants reported watching gambling streams in the past year, and this was associated with problem gambling severity (Zendle, 2020). Other work has highlighted gambling-like mechanics used in the chat on Twitch (e.g., raffles that do not involve monetary wagers or prizes; Abarbanel & Johnson, 2020). This mechanic is available across both gambling and non-gambling streams and speaks to engagement features on livestreaming platforms more broadly rather than the effects of exposure to gambling streams. Lastly, Hoebanx and French (2023) qualitatively examined themes of prerecorded gambling streams on YouTube, showing the intense and stimulating nature of these streams, but the psychological correlates of these elements are unknown.
The present study is among the first to focus on the psychological effects of watching gambling streams, so we began by recruiting a large sample of people who gamble to establish group-level characteristics between gambling stream viewers and non-viewers. We predicted that gambling stream viewers will score higher on problem gambling severity, as well as two psychological correlates of problem gambling, gambling cravings and boredom proneness, than non-viewers (https://aspredicted.org/XKD_VY3). We included boredom proneness because it is a risk factor for problem gambling (e.g., Mercer & Eastwood, 2010) and is related to engagement in livestreams (Zhang & Li, 2022). It has been conceptualized as an aspect of impulsivity and novelty-seeking, which are central to wider accounts of problem gambling (Blaszczynski & Nower, 2002). In Study 2, we recruited a more selective sample of gambling stream viewers to examine whether problem gambling severity is related to cravings elicited by watching gambling streams (i.e., evoked craving) and the motivation to use gambling streams to regulate their cravings (i.e., regulation motives). Viewers with higher levels of problem gambling are highly sensitive to gambling cues. Given that gambling streams contain rich cues (and thus likely evokes cravings), we suspected that cue reactivity from watching gambling streams would outweigh motivations to use them to regulate cravings, and thus in Study 2 we predicted that problem gambling severity would be more strongly correlated with evoked craving than regulation motives (https://aspredicted.org/PV8_C7G).1 Lastly, we explored whether cravings from watching gambling streams could be associated with affective responses to particular gambling stream events, such as high-stake bets, jackpot wins, and heavy losses. We have made our data, materials, and code publicly available: https://osf.io/gkt34/?view_only=09b572e99f424f46a561b7b1014f4e53.
Study 1
Methods
Participants
Participants were recruited from the United States, United Kingdom, Ireland, Australia, Canada, and New Zealand using the online recruitment platform Prolific Academic. Participants first completed a prescreen that assessed their eligibility to participate in our study. Those that met the criteria (gambled at least once within the past three months), were invited to the main survey. Of the 2,000 respondents to the prescreen, 1,387 reported gambling within the past three months and were invited to the main survey.
As preregistered, in the main study we aimed to have a final sample of 800 people who gamble, comprising 300 gambling stream viewers and 500 non-viewers, informed by a pilot study. This would allow us to detect a minimal effect size of d = 0.24 at 90% power and an alpha level of 0.05 (Faul, Erdfelder, Lang, & Buchner, 2007) for mean differences in problem gambling, craving, and boredom proneness scores between viewers and non-viewers. We oversampled, anticipating some data loss from cleaning, recruiting 1,000 participants for our main survey (https://aspredicted.org/XKD_VY3). Following our preregistration, we excluded participants who failed at least one of two attention checks (n = 11), completed the survey in under 4.8 min (n = 17), or reported that they had not gambled in the past three months (n = 13, i.e., inconsistent with their prescreen response on the eligibility item). Our final sample consisted of 965 participants. Participants demographics are shown on Table 1.
Demographics of participants in Studies 1 and 2
| Study 1 | Study 2 | ||
| Viewers n (%) or M (SD) | Non-Viewers n (%) or M (SD) | Viewers n (%) or M (SD) | |
| Gender | |||
| Man | 100 (45.2%) | 190 (29.6%) | 214 (79.0%) |
| Woman | 119 (53.8%) | 447 (69.6%) | 54 (21.0%) |
| Non-binary | 2 (1.0%) | 5 (0.8%) | 3 (0.0%) |
| Age | 36.8 (12.5) | 42.1 (11.0) | 36.0 (11.4) |
| Marital Status | |||
| Divorced or separated | 10 (4.5%) | 49 (7.6%) | 20 (7.4%) |
| Married or domestic | 126 (57%) | 396 (61.7%) | 115 (42.4%) |
| Single, never married | 85 (38.5%) | 197 (30.7%) | 136 (50.2%) |
| Education | |||
| High school or below | 29 (13.1%) | 102 (15.9%) | 39 (14.4%) |
| Some college | 76 (34.4%) | 162 (25.2%) | 75 (27.7%) |
| Bachelor's | 86 (38.9%) | 264 (41.1%) | 117 (43.2%) |
| Above Bachelor's | 30 (13.6%) | 114 (17.8%) | 40 (14.8%) |
| Income (USD) | 53,914.0 (37859.8) | 52,492.2 (34664.9) | 51,254.6 (29035.3) |
| Employment | |||
| Employed full-time | 145 (65.6%) | 320 (49.9%) | 172 (63.5%) |
| Employed part-time | 28 (12.6%) | 131 (20.4%) | 19 (7.0%) |
| Self-employed | 25 (11.3%) | 127 (19.8%) | 29 (10.7%) |
| Not working | 19 (8.6%) | 52 (8.1%) | 51 (18.8%) |
Note. Study 2 did not select participants based on gambling status, but most participants (92%) reported gambling in the past 3 months. Income (USD) refers to household income in the previous year.
Measures
Gambling stream viewership. Participants reported how frequently they watched gambling on streaming platforms within the past year. Responses were made on a scale consisting of 0 (I have never done this), 1 (Not at all in the past 12 month, but I have done this before then), 2 (Less than 10 times in total), 3 (Once a month), 4 (2–3 times a month), 5 (Once a week), 6 (2–3 times a week), and 7 (4 or more times a week). Following our preregistered plan, participants who reported having watched gambling streams within the past year (i.e., scores of 2–7) were classified as gambling stream viewers, and participants who reported having never watched gambling streams were classified as non-viewers (scores of 0). A further 102 (11%) participants reported having watched gambling streams before but not within the past year, and those subjects were excluded from the group-level comparisons (as per our preregistration).2
Problem gambling severity. To measure problem gambling severity, we used the Problem Gambling Severity Index (PGSI; Ferris & Wynne, 2001). It consists of 9 items on a scale ranging from 0 (Never) to 3 (Almost Always). Example items included “Have you bet more than you could really afford to lose?”, “Have you borrowed money or sold anything to gamble?”, and “Has your gambling caused you any health problems, including stress and anxiety?” Scores were summed. Scores could range from 0 to 27. Higher scores corresponded to higher problem gambling. Reliability was good (Cronbach's alpha = 0.90; Taber, 2018).
Gambling cravings. Gambling cravings were assessed using the gambling urge scale (Raylu & Oei, 2004), which consists of 6 items ranging from 1 (Strongly Disagree) to 7 (Strongly Agree). Example items include “All I want to do now is to gamble”, “It would be difficult to turn down a gamble right now”. Scores were averaged. Scores could range from 1 to 7. Higher scores corresponded to stronger gambling cravings. Reliability was good (Cronbach's alpha = 0.95; Taber, 2018).
Boredom proneness. Participants completed the Boredom Proneness Scale (Struk, Carriere, Cheyne, & Danckert, 2017), which consists of 8 items ranging from 1 (Strongly Disagree) to 7 (Strongly Agree). Example items include “I find it hard to entertain myself”, “It takes more stimulation to get me going than most people”. Scores were averaged. Scores could range from 1 to 7. Higher scores corresponded to higher propensity to feel bored. Reliability was good (Cronbach's alpha = 0.91; Taber, 2018).
Statistical analyses
To test the differences in problem gambling, gambling cravings, and boredom proneness between gambling stream viewers and non-viewers, we conducted our preregistered paired t-tests (https://aspredicted.org/XKD_VY3). We also ran a number of exploratory analyses. We tested demographic differences using chi-squared tests on categorical variables, and t-tests on continuous variables. In the survey, participants reported their age and income in categorical brackets (e.g., “25–29 years old”), but for the group tests and logistic regression, we recoded them as continuous using their midpoints (e.g., 27). To test which variables were most strongly associated with being a gambling stream viewer compared to non-viewer, we conducted a logistic regression model predicting viewership status from the above variables. Analyses were conducted in R Studio. The base R package “stats” was used for t-tests, chi-square tests, and regression models, the package “effectsize” was used to calculate effect sizes for the t-tests and chi-square tests (Ben-Shachar, Lüdecke, & Makowski, 2020), and the package cocor was used to compare correlation coefficients (Diedenhofen & Musch, 2015).
Ethics
The study was approved by The University of British Columbia's Behavioural Ethics Review Board. Participants were reimbursed £0.14 for completing a pre-screen survey and £1.20 for completing the main survey. Data was collected in June 2022. Participants provided informed consent at the beginning of the study, and they received a debriefing form including problem gambling resources at the end of the study.
Results
Descriptive statistics
Among 965 people who regularly gamble, 221 (23%) reported watching gambling streams within the past year and 642 (66%) reported never having watched gambling streams. Among gambling stream viewers, 74% reported gambling with real money before watching gambling streams online. Among non-viewers, 78.5% reported watching other content online (i.e., streams unrelated to gambling and unrelated to online television), indicating that most participants consumed some form of streaming media. Figure 1 shows density plots of PGSI, gambling cravings, and boredom proneness.
Density plots of problem gambling severity, gambling cravings, and boredom proneness between gambling stream viewers and non-viewers
Note. Density plots of PGSI (viewers M = 4.45, SD = 5.02; non-viewers M = 1.64, SD = 3.00), gambling cravings (viewers M = 1.85, SD = 0.99; non-viewers M = 1.25, SD = 0.54), and boredom proneness (viewers M = 3.54, SD = 1.31; non-viewers, M = 3.01, SD = 1.26).
Citation: Journal of Behavioral Addictions 13, 4; 10.1556/2006.2024.00062
Differences between gambling stream viewers and non-viewers
Supporting our preregistered hypotheses (https://aspredicted.org/XKD_VY3), gambling stream viewers reported higher PGSI scores, stronger gambling craving, and higher boredom proneness, compared to non-viewers. Regarding the demographic variables, chi-square tests showed that gambling stream viewership differed between gender, age, employment status, and marital status but not income or education. Specifically, gambling stream viewers were younger than non-viewers; On inspection of the standardized residuals for the categorical outcomes, gambling stream viewership was overrepresented among people who identified as men, in full-time employment, and single. Results are shown on Table 2.
Group differences between gambling stream viewers and non-viewers
| Variables | Logistic regression model | Bivariate tests | ||||
| OR | 95% CI | p | Effect size | 95% CI | p | |
| Problem Gambling | 1.07 | 1.00, 1.14 | 0.026 | 0.78 | 0.58, 0.97 | <0.001 |
| Gambling Craving | 2.12 | 1.47, 2.92 | <0.001 | 0.87 | 0.69, 1.07 | <0.001 |
| Boredom Proneness | 0.98 | 0.84, 1.17 | 0.895 | 0.42 | 0.26, 0.58 | <0.001 |
| Gender | 2.76 | 1.81, 3.99 | <0.001 | 0.22 | 0.15, 0.29 | <0.001 |
| Age | 0.96 | 0.94, 0.98 | <0.001 | −0.44 | −0.28, −0.59 | <0.001 |
| Income | 1.00 | 1.00, 1.00 | 0.897 | −0.04 | −0.19, 0.11 | 0.623 |
| Employ: Full | 1.40 | 0.77, 2.53 | 0.245 | 0.15 | 0.08, 0.21 | <0.001 |
| Employ: Part time | 0.89 | 0.41, 1.87 | 0.741 | |||
| Employ: Self | 1.34 | 0.58, 2.91 | 0.445 | |||
| Marital: Married | 1.07 | 0.34, 2.64 | 0.750 | 0.08 | 0.02, 0.14 | 0.049 |
| Marital: Divorced | 1.05 | 0.70, 1.65 | 0.903 | |||
| Edu: Some college | 1.34 | 0.70, 2.71 | 0.328 | 0.10 | 0.03, 0.15 | 0.050 |
| Edu: Bachelor's | 0.97 | 0.52, 2.00 | 0.933 | |||
| Edu: Above | 0.89 | 0.39, 1.71 | 0.457 | |||
Note. Results from logistic regression model and bivariate tests comparing viewers and non-viewers. Effect Size for PGSI, gambling craving, and boredom proneness represent Cohen's d. The reference group was men for gender, single for marital status, high school for education, and unemployed for employment status; Their Effect Size represents Cramer's V. Effect Size for age and income represent Cohen's d.
We conducted a logistic regression with these variables predicting gambling stream viewership.3 The reference group was men for gender, single for marital status, high school for education, and unemployed for employment status. The odds of being a gambling stream viewer compared to non-viewer was 7% greater for each additional point on the PGSI, 112% greater for every additional score on the gambling craving scale, 176% greater for men, and 4% less for every 5-year increase in age. Boredom proneness, marital status, employment statuses, and income were not significantly associated with gambling stream viewership in this model. Results are shown on Table 2.
Brief discussion
In Study 1, we examined characteristics that differentiate people who gamble and watch gambling streams from people who gamble and do not watch gambling streams. Gambling stream viewers tended be younger, identify as men, and reported higher problem gambling severity and stronger gambling cravings compared to non-viewers. Some other variables, including boredom proneness as a pre-registered variable, did not differ significantly in viewers in the logistic regression model. These findings indicate that among people who gamble, gambling stream viewers tend to have increased levels of gambling problems as well as other risk factors associated with emerging gambling-like activities (Closer et al., 2022; Macey, Abarbanel, & Hamari, 2021).
Study 2
In Study 2, we recruited a sample of gambling stream viewers to examine a potentially paradoxical scenario: Viewers may be motivated to watch gambling streams to regulate their gambling cravings; Yet, watching gambling streams may also elicit cravings to gamble via cue reactivity. We examined these variables as a function of PGSI and tested whether cravings from watching gambling could be attributed to particular gambling stream events, such as watching jackpot wins or heavy losses.
Methods
Participants
Participants were recruited from the United States, United Kingdom, Ireland, Australia, Canada, and New Zealand using the online recruitment platform Prolific Academic. We first used a prescreen to identify participants who watched gambling streams at least once within the past year. Those that met this criterion were invited to the main survey. Of 3,000 prescreen respondents, 888 reported watching gambling streams within the past year and were invited to complete the main survey.
We aimed to have a final sample of 240 participants, which would allow us to detect a small effect size of at least r = 0.21 at 90% power and at an alpha level of 0.05 (Faul et al., 2007), which is considered clinically meaningful (Ferguson, 2016), for the correlation between i) PGSI and evoked craving and ii) PGSI and regulation motives. Anticipating some exclusions from cleaning, we oversampled and recruited 300 participants for the main survey (https://aspredicted.org/PV8_C7G). As preregistered, we excluded participants who failed one of two attention check (n = 2), completed the survey in under 6.2 min (n = 9), or reported in the main survey that they have not watched gambling streams within the past year (n = 21, i.e., inconsistent with their prescreen response). Our final sample consisted of 271 participants (see Table 1).
Measures
Evoked craving and regulation motive. Evoked craving (i.e., experiencing an increase in cravings from watching gambling streams) and regulation motive (i.e., the motivation to watch gambling streams as self-regulation to reduce cravings) were measured using four items that were adapted from a similar scale designed for social casino games (Hollingshead et al., 2016, 2020). Regulation motive was assessed using two items: “I purposely watch gambling online to reduce my desire to gamble” and “I intentionally watch gambling streams to help myself cope with my cravings to gamble.” Responses were made on a scale ranging from 1 (Strongly Disagree) to 5 (Strongly Agree). These two items were strongly correlated, ρ = 0.73, p < 0.001. Evoked craving was assessed using two items: “When I watch gambling streams it seems to increase my urge to gamble for real money” and “Watching gambling online seems to trigger my desire to gamble.” These items were strongly correlated, ρ = 0.81, p < 0.001. The two items for each measure were averaged. Higher scores corresponded to stronger regulation motive and evoked craving.
Other measures. Problem gambling severity was assessed using the PGSI (Ferris & Wynne, 2001). Reliability was good (Cronbach's alpha = 0.91; Taber, 2018). Participants also reported how they felt when watching different gambling stream events, using an affect grid (Russel et al., 1989). They were asked, “Please use the cursor to select how watching [Gambling Stream Event] makes you feel.” The seven events were “high-stake bets,” “low-stake bets,” “jackpot wins,” “heavy losses,” “interactions in the chat,” “interactions with the streamer,” and “chat raffles/bets.” The 500 × 500 affect grid consisted of a horizontal valence axis labelled from unpleasant to pleasant, and a vertical arousal axis labelled from sleepy to high arousal. Participants reported their affect rating by using their cursor to click on the grid, thus capturing valence and arousal ratings simultaneously. Before these ratings, participants were asked to indicate a neutral rating (i.e., the center of the grid). Participants who failed to report their neutral rating within the 50 × 50 border from the center were excluded from analyses of the gambling stream events for careless responding (n = 10).
Statistical analyses
To test the relationships between problem gambling, evoked craving, and regulation motives, we first examined their bivariate correlations using Spearman's rank coefficients, as preregistered (https://aspredicted.org/PV8_C7G). To test if the strength of the correlation for problem gambling was larger for evoked craving than regulation motives, we compared the z-transformed correlation coefficients (Diedenhofen & Musch, 2015). As exploratory analyses, we examined if the relationship between problem gambling and evoked craving was moderated by regulation motives using a linear regression model with PGSI, regulation motive, and their interaction term predicting evoked craving. These variables were standardized. We additionally ran two models to examine the effect of PGSI on evoked craving at low (−1 SD) and high (+1 SD) levels of regulation motive. Most participants in Study 1 reported that they gambled before watching gambling streams (74%), so we tested PGSI as the predictor, rather than the outcome variable, though this was somewhat arbitrary given our cross-sectional design. Lastly, we explored the correlations between these variables and the affect ratings associated with the different gambling stream events, adjusting for multiple comparisons using False Discovery Rate corrections (Benjamini & Hochberg, 1995; Colquhoun, 2014).
Ethics
The study was approved by The University of British Columbia's Behavioural Research Ethics Board. Participants were reimbursed £0.15 for completing the prescreen and £2.25 for the main study. Data was collected in December 2022. Participants gave their informed consent at the beginning of the study and received a debriefing form with problem gambling resources at the end.
Results
Descriptive statistics
Among gambling stream viewers, most reported gambling within the past three months (92%). The average PGSI was 4.63 (SD = 5.15), indicating that our sample of viewers had moderate levels of problem gambling on average. Similar to Study 1, most participants (75%) reported gambling with real money before watching gambling streams. On average, participants reported watching gambling streams 3.20 (SD = 3.69) hours per week and watched 3.62 (SD = 2.87) different streamers per week. They reported following 3.66 (SD = 6.08) and subscribing to 2.43 (SD = 4.79) gambling streamers on average. On average, participants reported low levels of endorsement (i.e., disagreement, on average) of using gambling streams to regulate their gambling cravings (M = 2.31, SD = 1.12). On average, they were approximately neutral to gambling streams evoking cravings (M = 2.98, SD = 1.27).
Relationships between evoked craving, regulation motives, and problem gambling
We found that PGSI was significantly correlated with evoked craving, ρ = 0.42, p < 0.001 and regulation motives, ρ = 0.40, p < 0.001. In contrast to our prediction (https://aspredicted.org/PV8_C7G), these two correlation coefficients were not significantly different from each other, z = 0.45, p = 0.650. Next, we examined the correlation between regulation motive and evoked craving. If people use gambling streams to successfully thwart their gambling craving, regulation motives should be negatively correlated with evoked cravings. However, regulation motive was positively, albeit modestly, correlated with evoked craving, ρ = 0.18, p = 0.003, suggesting that some viewers who attempt to use gambling streams as a form of self-regulation may actually experience more craving as a result of watching the streams.
To better understand these relationships, we examined whether the relationship between PGSI and evoked craving (i.e., cue reactivity) was moderated by regulation motives (Fig. 2). In this model, PGSI was significantly associated with evoked craving, β = 0.43, 95% CI [0.26, 0.57], p < 0.001, as expected. However, regulation motives was no longer significantly associated with craving, β = 0.03, 95% CI [-0.11, 0.17], p = 0.617. We found a significant interaction between PGSI and regulation motives in predicting evoked craving, β = −0.24, 95% CI [-0.37, −0.08], p < 0.001. The relationship between PGSI and evoked craving was stronger at low levels, β = 0.67, 95% CI [0.36, 0.91], p < 0.001, than high levels of regulation motives, β = 0.19, 95% CI [0.06, 0.32], p = 0.005. Among people with high problem gambling severity, the predicted mean of evoked craving was 3.91 for those with low regulation motive and 3.38 for those with high regulation motive, which corresponds to d = 0.46, indicating a medium effect size.
Relationship between problem gambling and cravings as a function of regulation motive
Note. Left: Scores for evoked craving and regulation motive with box plots and histograms. Right: Relationship between PGSI and evoked craving at low (−1 SD), mean, and high (+1 SD) levels of regulation motive.
Citation: Journal of Behavioral Addictions 13, 4; 10.1556/2006.2024.00062
Craving and affect from gambling stream events
We examined whether evoked craving from gambling streams could be linked to specific gambling stream events. Table 3 shows the bivariate correlations between the valence (pleasantness) and arousal (stimulation) ratings of the seven gambling stream events, with evoked craving, regulation motive, and PGSI. After correcting for multiple comparisons, evoked craving was correlated with affect ratings from three gambling stream events: Stimulation from watching high-stake bets, ρ = 0.22, p < 0.001; stimulation from watching jackpot wins, ρ = 0.27, p < 0.001; and lower pleasantness from watching heavy losses, ρ = −0.19, p = 0.002. PGSI was correlated with stimulation from watching jackpot wins, ρ = 0.23, p < 0.00, and lower pleasantness from watching heavy losses, ρ = −0.17, p = 0.026. Lastly, regulation motives were correlated with stimulation, ρ = 0.21, p = 0.006, and pleasantness, ρ = 0.16, p = 0.010, from jackpot wins, as well as stimulation, ρ = 0.24, p < 0.001, and pleasantness, ρ = 0.18, p = 0.004, from chat raffles.
Correlations between affect ratings, regulation motive, evoked craving, and PGSI
| Stream event | Evoked craving | Regulation motives | PGSI | ||||||
| Valence | ρ | p | Cor p | ρ | p | Cor p | ρ | p | Cor p |
| High-Stake Bets | 0.08 | 0.198 | 0.309 | 0.12 | 0.048 | 0.136 | 0.10 | 0.102 | 0.200 |
| Low-Stake Bets | −0.04 | 0.556 | 0.707 | −0.06 | 0.370 | 0.520 | 0.01 | 858 | 0.925 |
| Jackpot Wins | 0.15 | 0.019 | 0.081 | 0.16 | 0.010 | 0.048 | 0.13 | 0.038 | 0.122 |
| Heavy Losses | −0.19 | 0.002 | 0.017 | −0.02 | 0.722 | 0.843 | −0.17 | 0.005 | 0.026 |
| Viewer Chat | −0.03 | 0.574 | 0.709 | 0.08 | 0.178 | 0.286 | −0.10 | 0.107 | 0.200 |
| Streamer Chat | −0.01 | 0.973 | 0.973 | 0.11 | 0.090 | 0.200 | −0.06 | 0.372 | 0.520 |
| Raffle Chat | 0.14 | 0.022 | 0.084 | 0.24 | <0.001 | 0.001 | 0.14 | 0.029 | 0.102 |
| Arousal | ρ | p | Cor p | ρ | p | Cor p | ρ | p | Cor p |
| High-Stake Bets | 0.22 | <0.001 | 0.003 | 0.10 | 0.093 | 0.200 | 0.09 | 0.130 | 0.219 |
| Low-Stake Bets | −0.03 | 0.666 | 0.799 | 0.00 | 0.954 | 0.973 | −0.04 | 0.538 | 0.706 |
| Jackpot Wins | 0.27 | <0.001 | <0.001 | 0.21 | <0.001 | 0.006 | 0.23 | <0.001 | 0.002 |
| Heavy Losses | 0.10 | 0.110 | 0.200 | −0.01 | 0.855 | 0.925 | −0.11 | 0.058 | 0.153 |
| Viewer Chat | −0.01 | 0.900 | 0.945 | 0.07 | 0.288 | 0.432 | −0.12 | 0.049 | 0.136 |
| Streamer Chat | 0.04 | 0.485 | 0.657 | 0.10 | 0.100 | 0.200 | −0.01 | 0.833 | 0.925 |
| Raffle Chat | 0.09 | 0.128 | 0.219 | 0.18 | 0.004 | 0.026 | 0.11 | 0.087 | 0.200 |
Note. Bivariate correlations between affect ratings and evoked craving, regulation motive, and PGSI. Cor p represents the p-value after applying false discovery rate corrections. Bolded effects are statistically significant (alpha = 0.05) after adjusting for fales discovery rates.
Brief discussion
In Study 2, we examined the relationship between problem gambling severity, evoked craving, and regulation motives. Problem gambling severity was positively correlated with evoked craving, in line with cue reactivity to gambling streams among people with more severe gambling problems (Limbrick-Oldfield et al., 2017). At the same time, problem gambling severity was correlated with regulation motives, adding to evidence that people who gamble use related activities to control their cravings (e.g., Hollingshead et al., 2020). In some respects, it is ironic that people with higher levels of problem gambling try to use streams to regulate cravings but also experience more cravings from watching streams. However, the relationship between problem gambling and evoked craving was negatively moderated by regulation motives, indicating that although people with gambling problems experience cravings from watching, this effect is attenuated by regulation motives.
General discussion
The goal of the present research was two-fold. First, we set out to establish the basic characteristics of people who gamble and watch gambling streams, finding that gambling stream viewers tended to be younger, tended to identify as men, and had higher levels of problem gambling and gambling cravings, compared non-viewers. Second, we investigated whether problem gambling was associated with cravings to gamble elicited by watching gambling streams or motivations to watch gambling streams as a form of self-regulation. Recognizing the antagonism between these effects, we predicted a stronger relationship between problem gambling severity and evoked cravings than regulation motives. We found that problem gambling severity was interestingly associated with both stronger evoked cravings and regulation motives, of similar strengths. To make sense of this, in a moderation analysis, we found that the relationship between PGSI and evoked cravings was actually attenuated by higher levels of regulation motives.
Participants with higher problem gambling severity scores reported stronger cravings from watching gambling streams. Shi, Renwick, Turner, and Kirsh (2019) proposed that various forms of social media, such as online forums, team-chat applications, and video streaming websites, can either “push” or “pull” people from the content they are engaging with. For watching video game streams, these media consistently act as a pull mechanism, making viewers want to play video games for longer (Cabeza-Ramírez, Sánchez-Cañizares, Fuentes-García, & Santos-Roldán, 2022; Macey & Hamari, 2018). Our findings suggest that this may be, in part, due to cravings evoked from watching the streams. Our findings are in line with behavioral theories that suggest gambling-related stimuli may elicit cravings via Pavlovian conditioning and conditioned responses. Elaborated Intrusion Theory further proposes that learned associations (which include classical conditioning of cues but also intrusive thoughts) can lead to the construction of mental imagery which have strong affective associations central to the development of cravings (Kavanagh, Andrade, & May, 2005). Given that gambling streams contain rich, multisensory cues that are more salient than traditional, simpler stimuli (Yalachkov et al., 2012), it is no surprise they may trigger cravings, akin to gambling-related images and other addiction-related stimuli in many laboratory studies (Limbrick-Oldfield et al., 2017; Tiffany & Wray, 2012).
Viewers reported watching gambling streams in an attempt to reduce their cravings to gamble. This highlights two competing possibilities – “the good and the bad” – as previously described for social casino games as another free-to-play, gambling-like format (Hollingshead et al., 2016, 2020; Wohl, Salmon, Hollingshead, & Kim, 2017). Watching gambling streams does not directly incur or exacerbate financial losses which is a key vector to harm in problem gambling (Langham, Thorne, & Rockloff, 2016). This may create a perception among people who gamble that gambling streams could be strategically and adaptively used as an alternative to real-money gambling. But if watching gambling streams also evokes cravings, this creates a delicate balancing act in which regulation motives may backfire, driving an ‘ironic’ desire to gamble. This may ultimately increase the subsequent likelihood of real-money gambling, especially among those with existing gambling problems, and future studies could test this possibility by tracking behavioural patterns between gambling stream viewing and subsequent gambling (e.g. through online applications). Our observation that the link between problem gambling severity and evoked craving was buffered by regulation motive indicates that that the triggering effects of gambling streams may be malleable and could be targeted by psychological interventions (Mena-Moreno et al., 2021).
Among people who gamble, we found that viewers tended to be young men, who are likely impressionable (Pitt, Thomas, Bestman, Daube, & Derevensky, 2017). From a social learning perspective, watching gambling streams may increase positive attitudes towards gambling. Future studies could seek to disentangle the factors underlying migration from watching to gambling. Nonetheless, this profile resonates with the risk factors seen for engaging in emerging forms of gambling in other research, including betting on esports (Macey et al., 2021) and purchasing video game loot boxes (Close, Spicer, Nicklin, Lloyd, & Lloyd, 2022). Across Studies 1 and 2, three quarters of participants reported engaging in real-money gambling before they began watching gambling streams, indicating that – conversely – some people appear to migrate from spectating gambling within the context of streaming platforms to real-money gambling. Similar migration effects have been characterized using longitudinal designs in social casino gamers (Gainsbury, Russell, King, Delfabbro, & Hing, 2016), and video gamers with higher levels of engagement with loot boxes (Brooks & Clark, 2023; González-Cabrera et al., 2023). This resonates with concerns surrounding the negative consequences of gambling streams as form of exposure among youth (Koncz et al., 2023).
Our study design did not characterize reactivity to gambling streams in situ. The cue-reactivity paradigm presents participants with stimuli depicting learned associations (e.g. images of drug paraphernalia or drug use, e.g., Carter & Tiffany, 1999). This has also been shown in the context of gambling (Limbrick-Oldfield et al., 2017). In our study, we did not directly present stimuli or streams to participants, and participants instead reported their cravings from watching gambling streams based on their memory of these experiences. According to Elaborated Intrusion Theory (Kavanagh et al., 2005), both exposure (e.g., cues) and recall (e.g., intrusive thoughts) can lead to the construction of gambling imagery and the experience of cravings. Nonetheless, exposure, given the saliency, may evoke stronger cravings than recall, and future research is needed to determine the joint role of exposure and recall on cravings and gambling.
Evoked craving from watching gambling streams was correlated with specific events that are widespread in gambling streams. Using an affect grid as a standard procedure in affective science (Russell, Weiss, & Mendelsohn, 1989), we found that participants who reported higher stimulation from watching high-stake bets and jackpot wins reported stronger cravings and higher problem gambling severity. These data forge a link between the intense forms of gambling that are accessed via gambling streams (Hoebanx & French, 2023) and the subjective experience of cravings. This is in line with work finding that high-stake bets and jackpot wins are structural characteristics linked to the high addictive potential of digital slot machine gambling (Edson et al., 2023). An alternative possibility is that the aversive qualities of watching heavy losses (specifically) might discourage viewers from gambling. If so, stronger negative affective responses to watching heavy losses should weaken cravings, but we did not find evidence for this effect. Unpleasantness from watching heavy losses was correlated with stronger cravings, and regulation motives were not closely correlated with unpleasantness from watching heavy losses. Thus, the satiation of cravings from watching does not seem to be attributable to the unpleasantness evoked from watching negative outcomes.
Our study suffers a number of limitations. First, our samples were crowdsourced from Prolific, so our data should not be treated as representative of population rates (Pickering & Blaszczynski, 2021; c.f. Russell, Browne, Hing, Rockloff, & Newall, 2022). For example, the data from Study 1 may not reflect prevalence of gambling stream viewership among the wider population of people who gamble. We recruited participants based on an eligibility screening in Prolific, and future studies could apply representative sampling (see Zendle, 2020), to establish population-level prevalences and generalizability. Second, our study was cross-sectional, meaning that we could not test the direction of the effects. Future studies using longitudinal designs (e.g., ecological momentary assessment) or experimental manipulations (e.g., exposing participants to gambling streams) could provide more insight into directional relationships and causal effects, as shown for loot boxes (Brooks & Clark, 2023; D’Amico, et al., 2022). Third, we focused on self-reported cravings, which is a widely used and validated measure in addiction science (e.g., Ha et al., 2021; Limbrick-Oldfield et al., 2017), but craving is a complex construct in terms of measurement, its relevance to psychopathology, and its underlying motivational substrates (Sayette et al., 2000). Future studies are required to test the degree to which watching gambling streams impact the various mechanisms that give rise to the phenomenology of cravings. Fourth, not all participants in Study 2 reported gambling within the past 3 months, and we could not distinguish whether PGSI scores of 0 indicated non-problematic gambling or non-gambling (in 8% of the sample). Lastly, our conclusions were drawn from participants who were English-speaking. Future cross-cultural studies are required to determine the degree to which our findings in English-speaking samples generalize to other cultures that use other streaming services.
Conclusion
The popularity of gambling streams has grown in recent years, raising concerns on the possible harms of watching gambling. Here, we report two preregistered studies that provide an initial investigation on a) the characteristics of gambling stream viewers, and b) the seemingly contrasting roles of gambling streams in evoking versus regulating cravings. We found that gambling stream viewers have characteristics that may render them more vulnerable to problem gambling. In Study 2, viewers with higher level of problem gambling report stronger cravings from watching gambling streams, but this relationship was weakened among viewers with stronger regulation motives, which may indicate some malleability in the evoked responses to gambling streams.
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 receives a Discovery Award from the Natural Sciences and Engineering Research Council of Canada (NSERC; RGPIN-2017-04069, RGPIN-2023-03528). RW's graduate training is funded by NSERC.
Authors' contribution
RW and LC were involved in the conceptualization and design. RW collected the data and performed the analyses with guidance from LC which were discussed with BA. LC obtained funding. RW wrote the original draft with guidance from LC which was presented to BA for further edits and feedback. All authors approved the final manuscript.
Conflict of interest
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 British Columbia Lottery Corporation (BCLC), a Canadian Crown Corporation. The Province of BC government and the BCLC had no role in the design, analysis, or interpretation of the study, and impose no constraints on publishing. LC has received travel expenses from Scientific Affairs (Germany) and the International Center for Responsible Gaming (US), and the Institute fur Glucksspiel und Gesellschaft (Germany). He has received fees for academic services and consultancy from Scientific Affairs (Germany), 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 BCLC. He has not received any further direct or indirect payments from the gambling industry or groups substantially funded by gambling. 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.
During the past five years, BA has received funding for research and/or consulting services from the Connecticut Council on Problem Gambling, Canadian Partnership for Responsible Gambling, Sports Betting Alliance, GLG Consulting, MGM Resorts International, Eilers & Krejcik Gaming, ProPress Germany, Scientific Affairs, McGill University, University of North Carolina School of Social Work, Marina Bay Sands, Aristocrat Gaming, Life Works, and Jones Ward. Dr. Abarbanel has received reimbursement for travel from Association Cluster Sport International, Kansspelautoriteit, Gamification Group (Finland), Scientific Affairs, British Columbia Lottery Corporation, International Association of Gaming Advisors, Las Vegas Convention and Visitors Authority, Canadian Partnership for Responsible Gambling, University of Salford, and National Collegiate Athletic Association (USA). During the time period, Dr. Abarbanel was a member of the Singapore National Council on Problem Gambling International Advisory Panel, for which she was reimbursed for her time. None of these entities played role in the design, analysis, or interpretation of this study, and impose no constraints on publishing. The other authors have no conflicts to disclose. During the past five years, International Gaming Institute at University of Nevada, Las Vegas, has received research and program funding from DraftKings, American Gaming Association, ESPN, MGM Resorts International, Wynn Resorts Ltd, Las Vegas Sands Corporation, Entain Foundation, Aristocrat Gaming, San Manuel Band of Mission Indians, Axes.ai, Sports Betting Alliance, Playtech, Sightline Payments, Global Payments, the State of Nevada Knowledge Fund, and State of Nevada Department of Health and Human Services. IGI runs the triennial research-focused International Conference on Gambling and Risk Taking, whose sponsors include industry, academic, and legal/regulatory stakeholders in gambling. A full list of sponsors for the most recent conference can be found at https://www.unlv.edu/igi/conference/18th/sponsors. IGI maintains a strict research policy (https://www.unlv.edu/igi/research-policy), as well as partnership and transparency framework (https://www.unlv.edu/igi/policies/partnership) to ensure appropriate firewalls exist between funding entities – no matter the entity’s classification – and IGI’s research and programs.
Supplementary material
Data, code, materials, and supplemental materials: https://osf.io/gkt34/
Preregistrations: https://aspredicted.org/XKD_VY3; https://aspredicted.org/PV8_C7G.
References
Abarbanel, B., & Johnson, M. R. (2020). Gambling engagement mechanisms in Twitch live streaming. International Gambling Studies, 20(3), 393–413. https://doi.org/10.1080/14459795.2020.1766097.
Anjani, L., Mok, T., Tang, A., Oehlberg, L., & Goh, W. B. (2020). Why do people watch others eat food? An empirical study on the motivations and practices of mukbang viewers. In Proceedings of the 2020 CHI Conference on human Factors in computing systems (pp. 1–13). https://doi.org/10.1145/3313831.3376567.
Benjamini, Y., & Hochberg, Y. (1995). Controlling the falase discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society. Series B (Methodological). http://www.jstor.org/stable/2346101.
Ben-Shachar, M. S., Lüdecke, D., & Makowski, D. (2020). effectsize: Estimation of effect size indices and standardized parameters. Journal of Open Source Software, 5(56), 2815. https://doi.org/10.21105/joss.02815.
Blaszczynski, A., & Nower, L. (2002). A pathways model of problem and pathological gambling. Addiction, 97(5). https://doi.org/10.1046/j.1360-0443.2002.00015.x.
Brooks, G. A., & Clark, L. (2023). The people who gamble of the future? Migration from loot boxes to gambling in a longitudinal study of young adults. Computers in Human Behavior, 141, 107605. https://doi.org/10.1016/j.chb.2022.107605.
Cabeza-Ramírez, L. J., Sánchez-Cañizares, S. M., Fuentes-García, F. J., & Santos-Roldán, L. M. (2022). Exploring the connection between playing video games and watching video game streaming: Relationships with potential problematic uses. Computers in Human Behavior, 128, 107130. https://doi.org/10.1016/j.chb.2021.107130.
Carter, B. L., & Tiffany, S. T. (1999). Meta-analysis of cue-reactivity in addiction research. Addiction, 94(3), 327–340. https://doi.org/10.1046/j.1360-0443.1999.9433273.x.
Choe, H. (2019). Eating together multimodally: Collaborative eating in mukbang, a Korean livestream of eating. Language in Society, 48(2), 171–208. https://doi.org/10.1017/S0047404518001355.
Close, J., Spicer, S. G., Nicklin, L. L., Lloyd, J., & Lloyd, H. (2022). Loot box engagement: Relationships with educational attainment, employment status and earnings in a cohort of 16 000 United Kingdom gamers. Addiction, 117(8), 2338–2345. https://doi.org/10.1111/add.15837.
Colquhoun, D. (2014). An investigation of the false discovery rate and the misinterpretation of p-values. Royal Society Open Science, 1(3), 140216. https://doi.org/10.1098/rsos.140216.
D’Amico, N. J., Drummond, A., de Salas, K., Lewis, I., Waugh, C., Bannister, B., & Sauer, J. D. (2022). No effect of short term exposure to gambling like reward systems on post game risk taking. Scientific Reports, 12(1), 16751. https://doi.org/10.1038/s41598-022-21222-3.
Diedenhofen, B., & Musch, J. (2015). Cocor: A comprehensive solution for the statistical comparison of correlations. Plos One, 10(4), e0121945. https://doi.org/10.1371/journal.pone.0121945.
Diwanji, V., Reed, A., Ferchaud, A., Seibert, J., Weinbrecht, V., & Sellers, N. (2020). Don’t just watch, join in: Exploring information behavior and copresence on Twitch. Computers in Human Behavior, 105, 106221. https://doi.org/10.1016/j.chb.2019.106221.
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.
Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39(2), 175–191. https://doi.org/10.3758/BF03193146.
Ferguson, C. J. (2016). An effect size primer: A guide for clinicians and researchers (pp. 310). American Psychological Association. https://doi.org/10.1037/14805-020.
Ferris, J., & Wynne, H. (2001). The Canadian problem gambling Index: Final report. Canadian Centre on Substance Abuse (CCSA).
Gainsbury, S. M., Russell, A. M. T., King, D. L., Delfabbro, P., & Hing, N. (2016). Migration from social casino games to gambling: Motivations and characteristics of gamers who gamble. Computers in Human Behavior, 63, 59–67. https://doi.org/10.1016/j.chb.2016.05.021.
Gibbs, E. (2022, November 4). Twitch’s recent gambling content ban leads to 20 percent viewership loss. Casino. https://www.casino.org/news/recent-gambling-content-ban-on-twitch-leads-to-20-loss-in-viewership/.
González-Cabrera, J., Basterra-González, A., Ortega-Barón, J., Caba-Machado, V., Díaz-López, A., Pontes, H. M., & Machimbarrena, J. M. (2023). Loot box purchases and their relationship with internet gaming disorder and online gambling disorder in adolescents: A prospective study. Computers in Human Behavior, 143, 107685. https://doi.org/10.1016/j.chb.2023.107685.
Gwilliam, M. (2022). Gambling enters top 10 most-watched Twitch categories as slots overtake Warzone. Dexerto. https://www.dexerto.com/entertainment/gambling-enters-top-10-most-watched-twitch-categories-as-slots-overtake-warzone-1855212/.
Ha, J., Park, W., Park, S. I., Im, C.-H., & Kim, L. (2021). EEG response to game-craving according to personal preference for games. Social Cognitive and Affective Neuroscience, 16(9), 995–1005. https://doi.org/10.1093/scan/nsaa131.
Hilvert-Bruce, Z., Neill, J. T., Sjöblom, M., & Hamari, J. (2018). Social motivations of live-streaming viewer engagement on Twitch. Computers in Human Behavior, 84, 58–67. https://doi.org/10.1016/j.chb.2018.02.013.
Hoebanx, P., & French, M. (2023). Interpassive gambling: The case of slot machine vlogs on YouTube. Critical Gambling Studies, 4(1), 66–76. https://doi.org/10.29173/cgs150.
Hollingshead, S., Kim, H. S. (Andrew), Rockloff, M., McGrath, D., Hodgins, D., & Wohl, M. (2020). Motives for playing social casino games and the transition from gaming to gambling (or vice versa): Social casino game play as harm reduction? Journal of Gambling Issues, in press. https://doi.org/10.4309/jgi.2021.46.4.
Hollingshead, S. J., Kim, H. S., Wohl, M. J. A., & Derevensky, J. L. (2016). The social casino gaming-gambling link: Motivation for playing social casino games determines whether self-reported gambling increases or decreases among disordered people who gamble. Journal of Gambling Issues, 33, 52–67. https://doi.org/10.4309/jgi.2016.33.4.
Hu, M., Zhang, M., & Wang, Y. (2017). Why do audiences choose to keep watching on live video streaming platforms? An explanation of dual identification framework. Computers in Human Behavior, 75, 594–606. https://doi.org/10.1016/j.chb.2017.06.006.
Johnson, M. R. (2024). Twitch. Cambridge, UK: Polity Press.
Kavanagh, D. J., Andrade, J., & May, J. (2005). Imaginary relish and exquisite torture: The elaborated intrusion theory of desire. Psychological Review, 112(2), 446–467. https://doi.org/10.1037/0033-295X.112.2.446.
Koncz, P., Demetrovics, Z., Griffiths, M. D., & Király, O. (2023). The potential harm of gambling streams to minors. Journal of the American Academy of Child and Adolescent Psychiatry, S0890-8567(23)00082-5. https://doi.org/10.1016/j.jaac.2023.01.020.
Kowert, R., & Daniel, E. (2021). The one-and-a-half sided parasocial relationship: The curious case of live streaming. Computers in Human Behavior Reports, 4, 100150. https://doi.org/10.1016/j.chbr.2021.100150.
Langham, E., Thorne, H., & Rockloff, M. (2016). Understanding gambling related harm: A proposed definition, conceptual framework, and taxonomy of harms. BMC Public Health. https://doi.org/10.1186/s12889-016-2747-0.
Li, B., & Lee, H. M. (2023). Exploring the effects of habituation and scent in first-person 360-degree videos on consumption behavior. Scientific Reports, 13(8353). https://doi.org/10.1038/s41598-023-35669-5.
Limbrick-Oldfield, E. H., Mick, I., Cocks, R. E., McGonigle, J., Sharman, S. P., Goldstone, A. P., … Clark, L. (2017). Neural substrates of cue reactivity and craving in gambling disorder. Translational Psychiatry, 7(1), Article 1. https://doi.org/10.1038/tp.2016.256.
Macey, J., Abarbanel, B., & Hamari, J. (2021). What predicts esports betting? A study on consumption of video games, esports, gambling and demographic factors. New Media & Society, 23(6), 1481–1505. https://doi.org/10.1177/1461444820908510.
Macey, J., & Hamari, J. (2018). Investigating relationships between video gaming, spectating esports, and gambling. Computers in Human Behavior, 80, 344–353. https://doi.org/10.1016/j.chb.2017.11.027.
Mallorquí-Bagué, N., Mestre-Bach, G., & Testa, G. (2023). Craving in gambling disorder: A systematic review. Journal of Behavioral Addictions, 12(1), 53–79. https://doi.org/10.1556/2006.2022.00080.
McCulloch, K. C., Fitzsimons, G. M., Chua, S. N., & Albarracín, D. (2011). Vicarious goal satiation. Journal of Experimental Social Psychology, 47(3), 685–688. https://doi.org/10.1016/j.jesp.2010.12.019.
Mena-Moreno, T., Fernández-Aranda, F., Granero, R., Munguía, L., Steward, T., López-González, H., … Jiménez-Murcia, S. (2021). A serious game to improve emotion regulation in treatment-seeking individuals with gambling disorder: A usability study. Frontiers in Psychology, 12. https://doi.org/10.3389/fpsyg.2021.621953.
Mercer, K. B., & Eastwood, J. D. (2010). Is boredom associated with problem gambling behaviour? It depends on what you mean by ‘boredom’. International Gambling Studies, 10(1), 91–104. https://doi.org/10.1080/14459791003754414.
Nah, H. S. (2022). The appeal of “real” in parasocial interaction: The effect of self-disclosure on message acceptance via perceived authenticity and liking. Computers in Human Behavior, 134, 107330. https://doi.org/10.1016/j.chb.2022.107330.
Pickering, D., & Blaszczynski, A. (2021). Paid online convenience samples in gambling studies: Questionable data quality. International Gambling Studies, 21(3), 516–536. https://doi.org/10.1080/14459795.2021.1884735.
Pitt, H., Thomas, S. L., Bestman, A., Daube, M., & Derevensky, J. (2017). Factors that influence children’s gambling attitudes and consumption intentions: Lessons for gambling harm prevention research, policies and advocacy strategies. Harm Reduction Journal, 14(1), 11. https://doi.org/10.1186/s12954-017-0136-3.
Raylu, N., & Oei, T. P. S. (2004). The Gambling Urge Scale: Development, confirmatory factor validation, and psychometric properties. Psychology of Addictive Behaviors, 18, 100–105. https://doi.org/10.1037/0893-164X.18.2.100.
Russell, A. M. T., Browne, M., Hing, N., Rockloff, M., & Newall, P. (2022). Are any samples representative or unbiased? Reply to pickering and Blaszczynski. International Gambling Studies, 22(1), 102–113. https://doi.org/10.1080/14459795.2021.1973535.
Russell, J. A., Weiss, A., & Mendelsohn, G. A. (1989). Affect grid: A single-item scale of pleasure and arousal. Journal of Personality and Social Psychology, 57(3), 493–502. https://doi.org/10.1037/0022-3514.57.3.493.
Sayette, M. A., Shiffman, S., Tiffany, S. T., Niaura, R. S., Martin, C. S., & Schadel, W. G. (2000). The measurement of drug craving. Addiction, 95(8s2), 189–210. https://doi.org/10.1046/j.1360-0443.95.8s2.8.x.
Shi, J., Renwick, R., Turner, N. E., & Kirsh, B. (2019). Understanding the lives of problem gamers: The meaning, purpose, and influences of video gaming. Computers in Human Behavior, 97, 291–303. https://doi.org/10.1016/j.chb.2019.03.023.
Sjöblom, M., & Hamari, J. (2017). Why do people watch others play video games? An empirical study on the motivations of Twitch users. Computers in Human Behavior, 75, 985–996. https://doi.org/10.1016/j.chb.2016.10.019.
Struk, A. A., Carriere, J. S. A., Cheyne, J. A., & Danckert, J. (2017). A short boredom proneness scale: Development and psychometric properties. Assessment, 24(3), 346–359. https://doi.org/10.1177/1073191115609996.
Taber, K. S. (2018). The use of Cronbach’s alpha when developing and reporting research instruments in science education. Research in Science Education, 48(6), 1273–1296. https://doi.org/10.1007/s11165-016-9602-2.
Taylor-Hill, G. (2023, March 2). Is Kick the controversial New Home for gambling streams? EsportsBets. https://www.esportsbets.com/news/kick-streaming-site-new-home-for-gambling/.
Tiffany, S. T., & Wray, J. M. (2012). The clinical significance of drug craving. Annals of the New York Academy of Sciences, 1248(1), 1–17. https://doi.org/10.1111/j.1749-6632.2011.06298.x.
Wohl, M. J. A., Salmon, M. M., Hollingshead, S. J., & Kim, H. S. (2017). An examination of the relationship between social casino gaming and gambling: The bad, the ugly, and the good. Journal of Gambling Issues, 35, 1–23. https://doi.org/10.4309/jgi.2017.35.1.
Woodcock, J., & Johnson, M. R. (2019). The affective labor and performance of live streaming on Twitch.tv. Television & New Media, 20(8), 813–823. https://doi.org/10.1177/152747641985107.
Wulfert, E., Maxson, J., & Jardin, B. (2009). Cue-specific reactivity in experienced people who gamble. Psychology of Addictive Behaviors, 23, 731–735. https://doi.org/10.1037/a0017134.
Yalachkov, Y., Kaiser, J., & Naumer, M. J. (2012). Functional neuroimaging studies in addiction: Multisensory drug stimuli and neural cue reactivity. Neuroscience and Biobehavioral Reviews, 36(2), 825–835. https://doi.org/10.1016/j.neubiorev.2011.12.004.
Young, M. M., & Wohl, M. J. A. (2009). The gambling craving scale: Psychometric validation and behavioral outcomes. Psychology of Addictive Behaviors, 23, 512–522. https://doi.org/10.1037/a0015043.
Zendle, D. (2020). Beyond loot boxes: A variety of gambling-like practices in video games are linked to both problem gambling and disordered gaming. PeerJ, 8, e9466. https://doi.org/10.7717/peerj.9466.
Zhang, N., & Li, J. (2022). Effect and mechanisms of state boredom on consumers’ livestreaming addiction. Frontiers in Psychology, 13. https://doi.org/10.3389/fpsyg.2022.826121.
We also preregistered this hypothesis in Study 1 (see https://aspredicted.org/XKD_VY3). However, after collecting the data, we found low reliability for a regulation motive item that was reverse-coded. It was negatively correlated with the other regulation item (ρ = −0.39), so it was not sensible to average the scores. Since we only had two items, the removal of this item influenced the pattern of results. This item was fixed in Study 2. The preregistered results are reported in our Online Supplemental (see Table S1). In Study 1, we additionally preregistered testing correlations between these measures and problem gambling severity, boredom proneness, and general gambling cravings, as well as whether these correlations hold among men and women. Given the low reliability of our regulation item in Study 1, we report these in our Online Supplemental instead (see Table S2).
We conducted a secondary analysis comparing PGSI, cravings, and boredom proneness scores between viewers, non-viewers, and this subset of previous viewers (not within the past year). Generally, previous viewers scored between viewers and non-viewers. Results are reported in our Online Supplemental.
We preregistered that we would test group differences in two additional ways (https://aspredicted.org/XKD_VY3). First, we initially reported that we would run separate multiple regressions with viewership predicting problem gambling, boredom proneness, and gambling cravings while controlling for demographic variables. However, this would not allow us to control for all the psychological variables and all the demographic variables simultaneously, so instead we conducted a logistic regression with all the variables predicting viewership. Second, we preregistered that we would look to identify a third group who watched non-gambling streams. Supporting our predictions, the third group who watched non-gambling streams scored lower on problem gambling, boredom proneness, and gambling cravings compared to the group who watched gambling streams. They did not differ from the group who did not watch any streaming content. We report this in our Online Supplemental (Table S3).
