Lukasz Walasek

Introduction: People place larger bets on risky events when they can cash-out before the bet concludes. This study experimentally investigates how cash-out availability impacts wider betting behaviour in a simulated betting task and whether risk of gambling problems moderates observed effects.Method: An incentivised, between-participants online experiment (n=377; 70.8% males, 28.6% females; Mage = 39.9), where participants could place bets on horse races via a simulated bookmaker’s website. Participants were randomly assigned to either have the option of cashing-out their bets or not. We examined choices to engage with the gambling task, stake sizes and implied win probabilities of placed bets, and measured risk of gambling problems by administering the PGSI.Results: The cash-out group was more likely to choose the gambling task over a distractor game (χ²(1, N=377)=4.59, p=0.032) and engaged with the gambling task for longer than the no-cash-out group (β=0.61, SE=0.23, p=0.0077). The cash-out group also made larger bets (W=19761.5, p=0.021), though this finding was more nuanced. There was no effect of cash-out availability on the average odds of placed bets (W=10635.5, p=0.67). No significant interactions between cash-out availability and PGSI were observed for either the implied win probability of placed bets (β<0.01, SE<0.01, p=0.44) or stake size (β=-0.02, SE=0.02, p=0.15), though a subgroup analysis revealed groupwise differences in stake size within a low-PGSI (W=6909.5, p=0.0076, n=260) subgroup. No such difference was found in the high-PGSI subgroup (W=1699.5, p=0.89, n=117).Conclusion: The provision of cash-out options in betting products results in significant changes in betting behaviour, which could potentially lead to increased risk of gambling harms.

We attempted replications of 274 claims of positive results from 164 quantitative papers published from 2009 to 2018 in 54 journals in the social and behavioral sciences. Replications were high-powered on average to detect the original effect size (Median = 99.6%), used original materials when relevant and available, and were peer-reviewed in advance through a standardized internal protocol. Replications showed statistically significant results in the original pattern for 151 of 274 claims (55.1% [95% CI 49.2 - 60.9%]) and for 80.8 of 164 papers (49.3% [95% CI 43.8 - 54.7%]) weighed for replicating multiple claims per paper. We observed modest variation in replication rates across disciplines (42.5% to 63.1%) though some estimates had high uncertainty. For claims with Pearson’s r effect sizes, the median was 0.25 [95% CI 0.21 - 0.27] for original studies and 0.10 [95% CI 0.09 - 0.13] for replication studies, a 82.4% [95% CI 67.8 - 88.2%] reduction in shared variance. Thirteen methods for evaluating replication success provided estimates ranging from 28.6% to 74.8% (median = 49.3%). Some decline in effect size and significance is expected based on power to detect original effects and regression to the mean due to replicating only positive results. The conditions that promote or inhibit replicability are worthy of additional investigation.

A common feature of contemporary sports-betting apps is ‘instant cash-out’, which allows users to settle a bet early in exchange for a discounted immediate payout. Despite high prevalence and links with gambling-related harm, relatively little is known about how personality traits associated with gambling, such as impulsivity, predict instant cash-out usage. To address this question, we recruited 145 general-population adult participants (69 men, 66 women, 10 non-binary or undisclosed; Mage = 36.3, SD = 10.7; participants resided in Australia, Canada, Ireland, New Zealand, the UK, or the USA) to complete five self-report questionnaires related to impulsivity, as well as the Problem Gambling Severity Index (PGSI), and a validated cognitive task measuring individual differences in cash-out frequency. We then assessed how cash-out frequency in the behavioral task was associated with both self-reported impulsivity and PGSI. We found that cash-out frequency was negatively correlated both with PGSI scores and with a number of impulsivity-related traits including Dysfunctional Impulsivity, Lack of Premeditation, Positive Urgency, Sensation Seeking, and Fun Seeking. An exploratory factor analysis revealed that higher scores on a latent ‘Dysfunctional Impulsivity’ factor were negatively associated with cash-out frequency overall, whereas higher scores on an ‘Inhibition and Inflexibility’ factor predicted higher cash-out frequency specifically for bets with a low win probability. Taken together, results suggest that instant cash-out may primarily appeal to less impulsive people and those with lower PGSI scores. This raises the possibility that instant cash-out may specifically facilitate increased gambling behaviors among people with less prior experience of gambling.

Gambling-related harms can have a significant negative impact on disordered gamblers, lower risk gamblers, and affected others. Yet, most disordered and lower risk gamblers will never seek formal treatment, often due to the stigma and shame surrounding gambling. Online self-help forums are a popular alternative way for gamblers to anonymously seek help from others. Analysis of these interactions can provide a deeper understanding of gambling than more commonly used research methodologies. In the present study, we leverage recent developments in natural language processing to analyze posts on a U.K.-based online self-help gambling forum. Using correlated topic modeling, we canvass the various types of discussions among forum members. We also combine this approach with semantic similarity analysis based on sentence embeddings, to map first the posts, and then the 10 topics, onto six previously established gambling-related harm domains. The topic modeling revealed a cluster of discussions of many negative emotions, a topic regarding the positive emotions underlying the potential for change, a distinct topic regarding gambling's relationship harms, and numerous environmental factors that contributed to harm. Emotional/psychological and health harms were most strongly associated with users' posts, illustrating the multidimensionality of severe gambling-related harm. Our results reveal the co-occurrence of different harms, such as the frequent mentions of financial harms and concomitant emotional/psychological harms. The analysis of the lived experiences of gambling-related harm in natural language represents a useful tool for gambling research and can provide a different perspective to inform policy. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Collaboration with a partner can lead to more cheating behaviour than would occur individually, but much is unknown about the social factors contributing to this tendency. In two preregistered experiments we investigate the effects of commitment to one’s partner and reward structure on cheating in a sequential dyadic die-rolling paradigm. In Experiment 1, we manipulated two features of the social context: perceived effort investment by one’s partner and the duration of partnership. We found that participants cheated more when they perceived that their partner exerted high effort, which was significant in the short partnership condition. Moreover, participants cheated more individually post-collaboration than they had pre-collaboration, showing that collaborative cheating may carry over into subsequent behaviour. In Experiment 2, we manipulated who received each round’s rewards. Participants cheated more for joint rewards, but less when they were the reward recipients and perceived that their partner exerted high effort.

We present a dataset of over 100 K textual descriptions of real-life choice dilemmas, obtained from social media posts and large-scale survey data. Using large language models (LLMs), we extract hundreds of choice attributes at play in these dilemmas and map them onto a common representational space. This representation allows us to quantify the broader themes and specific trade-offs inherent in life choices and analyze how they vary across different contexts. We also present our dilemmas to human participants and find that our LLM pipeline, when combined with established decision models, accurately predicts people’s choices, outperforming models based on unstructured textual content, demographics, and personality. In this way, our research provides insights into the attributes, outcomes, and goals that underpin life choices, and shows how large-scale LLM-based structure extraction can be used in combination with existing scientific theory to study complex real-world human behavior.

Reputation is multidimensional, with some traits being more relevant than others in particular contexts. Can people selectively respond to reputational cues relevant to the task at hand? Across three studies, we examined how people weigh cues about helpfulness and competence when forming expectations about strangers’ behavior. Using adapted investment games, we varied whether a stranger’s helpfulness or competence predicted participants’ future payoffs. We found that when helpfulness is task-relevant (Experiments 1 and 2), participants correctly use this cue in investment decisions. When competence matters most (Experiment 3), participants use it as the primary cue. Overall, a high reputation for outcome-irrelevant characteristics did not compensate for a low reputation for the outcome-relevant reputational cue. However, we also find an asymmetric spillover: Decision-makers prefer cooperating with others who are highly competent and highly helpful, regardless of task demands. We discuss our results within the theoretical framework of person perception and theories of reputation.

Importance: Cognitive Behavioural Therapy for insomnia (CBT-I) is the recommended first-line treatment for insomnia. Dissemination of this proven effective treatment at scale in places where it is most needed remains a challenge and requires sensitive adaptation and evaluation.Objective: To determine whether a hybrid digital CBT-I and emotion regulation (ER) delivered through workplaces improves insomnia, depression, anxiety, and productivity.Design: Randomised waitlist-controlled trial, with assessments at baseline and at eight weeks post randomisation.Setting: Online treatment, accessed in confidence through workplaces or self-referral.Participants: 159 non-shift workers with at least mild symptoms of insomnia, depression or anxiety, randomised to either digital CBT-I+ER or a waitlist control group.Intervention: dCBT-I+ER was delivered through a web-based platform and four therapy sessions via video-conferencing. Participants were asked to track their sleep using actigraphy and a sleep diary that was used to pace the intervention delivered.Main outcomes and measures: Primary outcomes were scores on standardised measures for insomnia, depression and anxiety. Secondary outcomes comprised psychological wellbeing, quality of life, work productivity and satisfaction, as well as self-reported and objective sleep parameters collected through sleep diaries and actigraphy, respectively.Results: Among 159 participants (mean age 43.6±9.4 years, 76.7% female, 80.5% white), 80 were assigned to dCBT-I+ER and 79 to waitlist control. Using general linear models, case-complete analyses showed that dCBT-I+ER had significant improvements on insomnia [F1, 134=71.46, p<.0001], depression [F1, 134=35.67, p<.0001], and anxiety [F1, 134=17.63, p<.0001]. Intention to treat analyses showed equivalent results for insomnia [F1, 1429=513.79, p<.001], depression [F1, 1429=243.53, p<.001] and anxiety [F1, 1429=136.20, p<.001]. Bonferroni corrected significance at p=0.01667 was reached except for psychological wellbeing [F1, 132.13=10.64, p<0.001], with a large effect size of d=1.05 in dCBT-I+ER. Actigraphy data showed no significant effects, while sleep diary data demonstrated significant improvements in all parameters except total sleep time, favouring dCBT-I+ER with large effect sizes (0.7<d<1.5). Improvements in quality of life, work productivity and satisfaction outcomes were not significant.Conclusions and relevance: Given its efficacy in workplaces, delivering dCBT-I+ER could be an effective scalable intervention with longer-term efficacy and economic benefits to be further evaluated.Trial registration: ISRCTN13596153, Registered on 08/06/2021.