Although gambling is a pastime for many, some individuals experience harmful consequences because of their gambling (Badji et al., 2023). While gambling disorder (American Psychiatric Association, 2022) is associated with adverse outcomes, less severe but nonetheless problematic levels of gambling are also associated with a range of negative consequences (Marionneau et al., 2023; Tseng et al., 2023). For instance, problem gambling is associated with increased risk of experiencing a range of psychological issues and socio-economic problems (Muggleton et al., 2021), addictive disorders (Dash et al., 2019; Grant & Chamberlain, 2020; Jauregui et al., 2020), and other mental health disorders (Churchill & Farrell, 2018; Vaughan & Flack, 2022). A range of factors are implicated in the onset and maintenance of problem gambling, with gambling as a coping mechanism or means of escape from negative emotions often referred to in theoretical approaches and featuring in contemporary research. Despite this consistent research focus, it is not yet well understood whether these associations hold across different approaches to the conceptualization and measurement of escape as it pertains to problem gambling.

Background

Escapism is often purported to play a central role in addiction and addictive behaviours, including gaming disorder (Melodia et al., 2022), internet use disorders (Kardefeldt-Winther, 2014) and problem gambling (Jouhki & Oksanen, 2022). Historically, addiction theories that incorporate escape acknowledge physiological and developmental origins (e.g., Jacobs’ general theory of addictions; Jacobs, 1986). Although escapism as a concept includes many and varied definitions (Hagström & Kaldo, 2014), theories applied to gambling commonly implicate gambling as a means to distract from or otherwise avoid or manage negative emotional states or cope with distress (e.g., Jacobs, 1986; Thomas et al., 2009; Wood & Griffiths, 2007; Young & Wohl, 2009). Indeed, the pathways model of problem gambling by Blaszczynski and Nower, (2002) outlines an emotionally vulnerable pathway, characterised by higher rates of comorbid disorders such as depression and anxiety, as well as developmental and physiological elements such as stress-coping motivation (Nower et al., 2022), consistent with Jacobs’ general theory. Similar findings are reflected in various correlational studies (Flack & Buckby, 2020; Jauregui et al., 2017; Lelonek-Kuleta & Bartczuk, 2021; Richardson et al., 2023). In essence, despite the differences in the theories, research designs, and assessment approaches, the prevailing theme of “gambling to escape” is notably salient across the literature in relation to perpetuating problem gambling.

Concomitantly, a growing body of research has found evidence that difficulties managing or tolerating unwanted emotional states is associated with addictive behaviours, as well as problem gambling. For instance, difficulties accepting, adjusting, and tolerating negative emotions has been shown to be positively associated with problem gambling (Jauregui et al., 2017; Marchica et al., 2020; Navas et al., 2019) and meta-analytic research has revealed general emotion regulation difficulties are positively correlated with problem gambling severity (Velotti et al., 2021). Similarly, avoidant emotion-focused coping styles, reflecting a tendency to engage in behaviours to escape or manage negative emotions, may direct individuals to adopt gambling as a more proximal coping strategy to manage stressors and associated distress, which may lead to or exacerbate problem gambling (Caudwell et al., 2024; Solberg et al., 2022). A recent systematic review revealed that a range of coping styles and strategies that are reflective of general emotional avoidance are positively associated with problem gambling (Neophytou et al., 2023). Taken together, emotional regulation and avoidant emotion coping models of problem gambling converge on the point that individual differences in the ability to manage negative emotions may explain why gambling is continued despite the accrual of longer-term negative consequences.

Existing reviews (e.g., Neophytou et al., 2023; Velotti et al., 2021) support the premise that gambling as an escape plays a central role in the maintenance of problem gambling, although the measures used in these studies consider dispositional characteristics such as emotion regulation and avoidant coping style. Although such characteristics can offer insights into why people may persist at gambling in the face of stress or emotion, they do not directly consider the reasons of individuals who engage in problem gambling in the explicit sense. Focusing on the reasons why people gamble can allow for more direct assessment of gambling as escape in relation to problem gambling. For instance, the Gambling Motives Questionnaire (GMQ; Stewart & Zack, 2008) and modified GMQ-F (Dechant, 2014); the Gambling Functional Assessment–Revised (GFA-R; Weatherly et al., 2011), the Reasons for Gambling Questionnaire (RGQ; Wardle et al., 2011), and the Gambling Motivation Scale (GMS; Chantal et al., 1995) all comprise a construct that reflects gambling as a means to escape to some extent (e.g., coping motive). Similarly, gambling outcome expectancies scales are designed to assess the consequences individuals associate with engaging in gambling behaviour (e.g., The Gambling Expectancy Questionnaire [GEQ; Gillespie et al., 2007] and the Gambling Outcome Expectancies Scale [GOES; Flack & Morris, 2015]). Such scales are often used in prevalence studies that aim to capture the influential reasons associated with gambling participation within a given population. For instance, the British Gambling Prevalence Survey uses the RGQ (UK Gambling Commission, 2024), whereas the Northern Territory Gambling Prevalence and Wellbeing Survey (Australia; Stevens et al., 2017) uses the GOES. Commonly identified reasons for gambling in such surveys incorporate excitement, financial, and social reasons for gambling. Escape, however, has been found to differentiate low and at-risk gamblers, in that at-risk gamblers score significantly higher on escape expectancies when other gambling expectancies are controlled for (Flack & Stevens, 2019).

Related to motives and beliefs is perhaps the most notable characteristic of problem gambling that separates it from other addictive behaviours—the allure of potential monetary gain (i.e., winning). Indeed, pecuniary gain has long been considered to be the primary reason for gambling (Hagfors et al., 2022; Ladouceur & Walker, 1996; Lee et al., 2007). There is indirect support for this contention, with research noting that gamblers’ erroneous beliefs and perceptions of skill and luck are positively associated with problem gambling severity (Buen & Flack, 2022; Devos et al., 2020; Flack & Morris, 2017; MacLaren et al., 2015). Indeed, a recent meta-analytic study shows monetary motives and expectancies share a small but significant relationship with problem gambling severity after controlling for other reasons for gambling (Tabri et al., 2022). However, it remains unclear whether measures of gambling as escape are consistently associated with problem gambling severity across the literature, and, whether the relationship will may hold after adjusting for motives, reasons, and expectancies related to financial gain.

The Current Study

Though reviews have been conducted to further establish the state of the emotion regulation and coping style research literature, to the authors knowledge no studies to date have systematically reviewed the literature to ascertain the consistency and strength of the relationship between the motivational type measures of gambling as an escape and problem gambling severity. Such research is necessary to inform gambling research priorities, that can lead to the development of targeted interventions and better means of community support. 2022). Despite this, gambling is distinct from other behavioural addictions that are characterised by escape motives, beliefs, or expectancies, as it incorporates the potential for financial gain (Tabri et al., 2022). Accordingly, many approaches that aim to reduce problem gambling focus on articulating the high probability for financial loss with continued gambling, and/or attempting to correct erroneous beliefs about the likelihood of monetary gain (Newall et al., 2023). Such approaches may not be as effective for all problem gamblers, with recent research indicating it is the notion of escape that largely determines problem gambling behaviour and continued gambling engagement, as opposed to financial loss—which is not considered ‘in the moment’ (Oakes et al., 2020). Therefore, it is of interest to test the extent to which escape is associated with problem gambling when financial reasons are taken into account. As such, the present study had two main areas of focus: (1) to clarify whether specific measures of gambling motives, reasons, or expectancies, related to escape (i.e., gambling for escape) share a positive relationship with problem gambling severity; (2) whether the relationship holds after adjusting for gambling for financial gain. These aims informed the development of a systematic review and meta-analysis of the relationship between validated measure of gambling to escape, and problem gambling severity, controlling for the effect of financial motives or monetary expectancies.

An additional point of focus was to determine whether the magnitude of the relationship between gambling to escape and problem gambling varied as a function of the measure used (i.e., whether the relationships are stronger or more consistent depending on the use of motive versus outcome expectancy-type measures). An advantage of the motive and outcome expectancies measures of gambling motivation is they provide a more direct assessment of the reasons for gambling, compared to that inferred by the correlational work investigating measures of emotion regulation and coping style. Though motive and outcome expectancies measures are associated with problem gambling, a fundamental difference between them is that gambling motives measures (e.g., GMS, GMQ) use self-report questions that assess the frequency of behavioural engagement for certain reasons (e.g., “how often do you gamble to relax?”) whereas outcome expectancy instruments (e.g., GOES) use belief-based statements that reflect individuals’ anticipated outcomes of gambling engagement (e.g., “gambling is the best way to forget about everyday problems”).

Method

The current systematic review and meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and checklist (Moher et al., 2009). Prior to conducting the searches in the databases, a protocol of the study was developed and registered (PROSPERO registration number: CRD42023391522).

Search Strategies

The systematic review was carried out in May 2024 for articles published from 1994 using the following databases: EBSCOhost (Medline with full text, APA PsycArticles, Psychology and Behavioural Sciences Collection, APA PsycInfo, AMED-The Allied and Complementary Medicine Database), Science Direct and those available using the institution’s library search. The search strategy was developed with identified keywords was developed to search the databases for title, abstract and content of records in each database.

The keywords were used with appropriate truncations in the selected databases to allow for retrieval of the different variants of the keywords: (coping, escapism, negative escapism, negative reinforcement, problem gambling, problematic gambling, gambling motives, gambling, gamble, gambling disorder, pathological gambling, motive, expectance, reason). For example, in PsycINFO, the search strategy used was “(“Problem Gambling" OR "Gambling Problem" OR "Problematic Gambling" OR " Gambling Disorder” OR "Patholog* gambl*" OR Gambl*) AND ((Cop* OR Escap* OR "Negative Escapism" OR Avoid* OR " Negative Reinforcement”) AND (Expect* OR Reason* OR Motiv*))”. Different keywords for same constructs were combined using the OR operator, while the phrases of the constructs were combined with the AND operator.

Eligibility Criteria

Studies were included if they (1) were peer reviewed manuscripts published after 1993 (the year the first gambling motive scale was developed); (2) were published in English language; (3) were a quantitative study; (4) included study participants aged 18 years old or older; (5) included a psychometrically validated problem gambling scale; (6) used a validated, multiple item measure of gambling motives, reasons, or expectancies, related to escape; and (7) included relevant statistical data (i.e., zero-order correlations). Studies were excluded if they (1) were a qualitative study; (2) measured escape with a single item or an open-ended question; (3) did not report on escape (coping); or (4) shared the same data with a study that was otherwise included in our review. A fidelity checklist was developed based on the inclusion and exclusion criteria, to assist with consistent screening of the articles.

Selection and Quality Assessment Process

The first author conducted the searches in all the selected databases and exported all findings to EndNote. Both manual and automated processes (e.g., deduplication) were performed to remove all duplicates. A preliminary screening was manually carried out to test the utility of the fidelity checklist, with adjustments made as required. All titles and abstracts were screened by the first author against the checklist and studies that did not meet criteria at these levels were removed. Criteria checking at the title and abstract phase was carried out by the third author, using a randomly selected 10% of the deduplicated records, using the fidelity checklist. At the full text phase, 20% of the studies were randomly selected for screening by the third author, again using the same process. Disparities, which were mainly due to reasons for exclusions, were discussed and reviewed, until a consensus was reached.

The quality of each study included in this meta-analysis was assessed using the Joanna Briggs Institute (JBI) critical appraisal checklist for analytical cross-sectional studies (Moola et al., 2015; Supplementary S1). This analysis assesses the methodological quality of a study to determine the extent to which the study addresses the possibility of bias in its design, and analysis. The overall result of the quality assessment of the included studies indicated that the studies met the criteria of a systematic review due to the conservative selection criteria.

Statistical Analysis

The literature search identified 3477 studies, and 474 duplicates were removed. A total of 3003 records were screened for title and abstract, and 2888 did not meet the inclusion criteria, leaving 115 articles for full text review. In total, 88 studies were excluded (see Fig. 1). For articles where data were available but not reported, a total of two emails were sent to the respective authors to request the relevant information. In total, 27 studies were able to be included in the review and meta-analysis.

Fig. 1
figure 1

PRISMA flow chart (Moher et al., 2009) demonstrating the study selection and screening process

Full size image

Effect Size

The following details were extracted systematically from all included studies: participant demographics, type of problem gambling and motive scales, and the Pearson correlation coefficient effect size (r) between the escape, monetary reasons for gambling and problem gambling. These correlations were transformed to Fisher’s Z scores, and with their corresponding sample sizes were used to calculate (1) the zero order associations and variances between escape and problem gambling, and; (2) the partial correlations between escape and problem gambling, when the influence of financial motive or expectancies were adjusted for.

Meta-Analysis

Random effects model using restricted maximum likelihood was fitted, and forest plots of effect sizes with 95% CI were plotted for effects of gambling to escape on problem gambling for both zero-order and partial correlations. A similar method was adopted for the subgroup analyses, which examined the effect size based on the motive or expectancy scale used.

Heterogeneity between studies was assessed using the I-squared statistic (I2; Higgins et al., 2003), with Egger’s regression-based test (Egger et al., 1997) and the nonparametric trim and-fill analysis (Duval & Tweedie, 2000) used to ascertain the existence and influence of publication bias on the effect size observed. Specifically, Egger's test identifies publication bias via asymmetry within the funnel plot, whereas the trim and fill analysis estimates the number of studies likely missing due to publication bias, adjusting effect sizes accordingly.

Results

Descriptive Statistics

The studies reviewed span publications from 1994 to 2024. A total of 27 studies met the inclusion criteria. The most frequent used motive scale was GMQ/F at 52% (n = 14) and the majority of studies were from Canada (n = 10; 37%). Supplementary S2 shows descriptive data for studies included in the meta-analysis (country, sample size, population sampled, type of gambling, reasons for gambling scale, and the problem gambling measure used).

Zero-Order Associations Meta-Analysis

Escape was positively associated with problem gambling with an overall effect size of 0.59 (95% CI = 0.49; 0.69). The effect sizes displayed a high degree of variability (I2 = 97.8) indicating that a random effects model was suitable (Dettori et al., 2022). Although the heterogeneity was high, the effect sizes of all the studies were significant and positive. Egger’s test (t = 4.01, p < 0.001) revealed the potential presence of publication bias, however, a trim-and-fill analysis (Duval & Tweedie, 2000) indicated the results were unlikely to have overestimated the effect size, as indicated by the small change in corrected effect size: 0.66 (95% CI = 0.56; 0.75). Figure 2 displays forest plot of the Fisher Zr effect sizes from the zero-order association analysis.

Fig. 2
figure 2

Zero-order associations between escape and problem gambling

Full size image

Subgroup Analysis

Subgroup analyses were carried out for the associations between escape and problem gambling for the different scale categories, grouped into motives, expectancies, and others. The motives group included 14 studies that used either the GMQ or GMQ-F (e.g., Devos et al., 2017; Grubbs & Rosansky, 2020; Hagen et al., 2023; Hollingshead et al., 2016; Jauregui et al., 2018; Kim et al., 2019; MacLaren et al., 2012, 2015; Marchetti et al., 2019; Myrseth & Notelaers, 2017; Rapinda et al., 2023; Schellenberg et al., 2016; Stewart & Zack, 2008; Tabri et al., 2015). The expectancies group included five studies; all used the GOES, the only outcome expectancies measure identified that incorporates gambling for escape. These studies included Caudwell et al. (2024), Flack and Buckby (2020), Flack and Morris (2015, 2016), Flack and Stevens (2019). Given no more than two studies were found that utilised each of the remaining scales, they were collapsed into the other category for the purposes of subgroup analyses, comprising the GMS (Rodriguez et al., 2015;), the RGQ (Francis et al., 2015; Keough et al., 2018;), the, GFA-R (Dighton et al., 2023; Weatherly & Cookman, 2014) and three author developed or modified measures (Jouhki et al., 2022; Thomas et al., 2009; Wang et al., 2020).

Overall, the effect sizes obtained indicated positive associations with between escape and problem gambling, irrespective of the instrument used. However, there was significant difference in the magnitude of the positive association between the groups, Q = 18.72, p < 0.001. For instance, compared to the GOES group which displayed a small to moderate effect size of 0.38 (95% CI = 0.35; 0.42) the GMQ and GMQ-F and other group revealed a moderate effect size, 0.61 (95% CI = 0.49; 0.73) and 0.68 (95% CI = 0.45; 0.90), respectively. In terms of heterogeneity, the GOES group displayed a low level of heterogeneity (I2 = 26.1) whereas the motive and other groups displayed a high variability (both I2 > 96.0).

Partial Associations Meta-Analysis

The meta-analysis was repeated with studies that included measures reflecting both escape and financial reasons for gambling (n = 14), allowing for consideration of the intercorrelations between these constructs to derive the partial associations between escape reasons and problem gambling. Therefore, all GMQ studies (n = 7), and five of the other category studies (Hagen et al., 2023; Jouhki et al., 2022; Keough et al., 2018; Thomas et al., 2009; Wang et al., 2020; Weatherly & Cookman, 2014) were excluded on this basis. Correlations between escape and financial reasons for Rodriguez et al., 2015 were not provided, and as such this study was also excluded. The partial associations meta-analysis revealed that after controlling for financial reasons, the association between escape and problem gambling remained moderate in size: 0.51 (95% CI = 0.38; 0.65). Similar to the previous zero-order analysis, there was a high level of heterogeneity (I2 = 98.1). The results from the partial associations meta-analysis are displayed in the Fig. 3 Forest plot.

Fig. 3
figure 3

Partial associations between escape and problem gambling, controlling for financial reasons for gambling

Full size image

Subgroup Analyses

Subgroup analyses were similarly conducted for the partial associations between escape and problem gambling, revealing that the effect sizes remained positive for the GMQ-F and GOES subgroups, although the other group comprised of the RGQ and GFA (Dighton et al., 2023; Francis et al., 2015) displayed a high degree of heterogeneity (I2 = 99.9) and failed to reach significance: 0.76 (95% CI = -0.03; 1.55). The subgroup analysis was therefore rerun to compare the GMQ-F and GOES subgroups only. There was significant difference between the GMQ-F and GOES, Q = 30.75, p < 0.001. Comparatively, the GOES group displayed a smaller effect size of 0.30 (95% CI = 0.27; 0.33) than the GMQ-F with an effect size of 0.60 (95% CI = 0.50; 0.70). However, the GOES group displayed a lower level of heterogeneity (I2 = 0.0) whereas the GMQ-F displayed a high variability (I2 = 86.5).

Discussion

The review identified 27 studies that assessed the association between escape and problem gambling. High heterogeneity was observed across studies exploring associations between reasons for gambling and problem gambling, although the heterogeneity varied somewhat between the escape measures used. Overall, the meta-analysis revealed positive associations of between small to moderate effect sizes between escape reasons for gambling and problem gambling, both in terms of the zero order and partial associations (i.e., controlling for financial reasons for gambling). This suggests that the association between escape and problem gambling is not due to conceptual overlap with financial reasons for gambling, highlighting the importance of escape to problem gambling specifically.

The findings from this study are consistent with a common theme throughout emergent systematic reviews and meta-analyses of gambling research, noting the significance of emotion and coping in relation to gambling frequency, and problem gambling. For example, the meta-analysis by Velotti et al., (2021) found a consistent significant, positive association between the need to avoid or suppress negative emotions and increased problem gambling. Their study found that avoidance of aversive feelings and negative emotions led to increased risks of using strategies (such as gambling) as a temporary escape, providing short-term relief. Similarly, findings from the recent systematic review of Neophytou et al., (2023) indicates that problem gamblers likely rely heavily on avoidant coping strategies. These reasons for gambling will likely negatively reinforce gambling behaviour, leading to increased gambling frequency and problem gambling risk in the longer term. While the importance of financial motives to gambling, as identified by Tabri et al., (2022), may lead to considering things like digital payment technologies as a mechanism of harm minimisation, recent longitudinal research has found that both money and escape reasons predict problem gambling over time (Hagfors et al., 2023). Similarly, lived experience research among those using addictive substances and experiencing behavioural addictions commonly cite coping as influencing problematic use or engagement (Coelho et al., 2024). Accordingly, interventions that consider the development of adaptive coping skills, or which challenge escape or coping-related expectancies, would appear relevant to gambling (e.g., Caudwell et al., 2024), but also to behavioural addictions more generally.

A key finding of the subgroup analysis was the consistent positive significant associations between escape and problem gambling, irrespective of the scale used (although the magnitude of these associations did vary). However, it is important to further consider the finding that the pooled effect size for the studies that used the GOES was smaller than that for studies that used the GMQ, GMQ-F, and other scales. One tentative explanation that may, in part, explain stronger effects on problem gambling from the GMQ and GMQ-F relates to their measurement of how often respondents gamble for certain reasons; whereas the GOES measures agreement that gambling brings about certain outcomes. This could contribute to differences between studies sampling more frequent or higher problem gambling risk participants, which would lead to greater correspondence between motive items measuring frequency, and gambling-related measures (e.g., the behavioural dependence items within the PGSI; Tseng et al., 2023). Overall, the subgroup analysis emphasises the importance of considering approaches to measuring reasons for gambling, gambling engagement, and problem gambling, in relation to different sampling methods employed throughout this research area (Pickering & Blaszczynski, 2021).

Strengths, Limitations, and Future Research Directions

The present systematic review and meta-analysis exhibits strengths and limitations that need to be considered when interpreting its findings and provide avenues for continuing work in this area of inquiry. That the systematic review and meta-analysis included a fidelity checklist for screening ensured a well-defined scope (Aromataris & Pearson, 2014), and allowed for a methodologically thorough analysis of heterogeneity and publication bias that increased the robustness of estimates of the associations between reasons for gambling and problem gambling. Conversely, as a consequence of applying stringent inclusion criteria (i.e., including studies using a validated measure of reasons for gambling, where authors reported or made available correlations between reasons), fewer studies were included. This limited the subgroup analyses, which, while yielding important insights, carries implications for the generalisability of these results to general and clinical populations. Future studies may need to consider adopting validated multidimensional measures of reasons for gambling; considering individuals may gamble for different or multiple reasons, some of which are more related to problem gambling than others. Similarly, research in this area would be improved by the routine reporting of correlations between differing reasons (i.e., motives, expectancies). For instance, while escape reasons are highly influential and offer avenues for intervention, it would be interesting to see if the associations between the constructs would still hold if the shared variance with other affect-based reasons for gambling (e.g., excitement, enhancement) were similarly controlled. This is especially important, as gambling for reasons associated with positive reinforcement has been found to predict gambling problems depending on gambling modality (Richardson et al., 2023). Similarly, given some reasons for gambling are conceptualised as positive, affect-based reinforcers (i.e., Puiras et al., 2020; Schellenberg et al., 2016), it is essential to clarify their relative contribtions to problem gambling across different cohorts of gamblers. For instance, the systematic review by Merkouris et al., (2016) includes reference to studies that have found gender differences in reasons for gambling (e.g., in relation to affect or mood), and preferred gambling activity (e.g., EGM use), although, these are operationalised differently throughout the available literature. Future meta-analytic work in this area would be facilitated by consensus and consistency in the approach to measuring gambling types and modalities, and reasons for gambling (Richardson et al., 2023).

Conclusion

The findings of this review and meta-analysis align with the convergence of current thinking that gambling to escape carries a significant motivational influence in the precipitation and maintenance of problem gambling. Additionally, the findings show that the influence of escape on problem gambling is present regardless of the scale used. Future work in this area should continue to investigate the relative strength of associations between escape and problem gambling in different population groups in an effort to inform effective public health and clinical interventions.