With this document you can:

This box is not visible in the printed version.

Measuring the adverse consequences from gambling

Read how we have developed new questions about adverse consequence from gambling which are included in the GSGB survey.

Published: 25 July 2024

Last updated: 25 July 2024

This version was printed or saved on: 10 September 2024

Online version: https://www.gamblingcommission.gov.uk/report/measuring-the-adverse-consequences-from-gambling

Introduction

This technical report describes the process of developing a new set of questions to assess adverse consequences from gambling, which have been included in the Gambling Survey for Great Britain (GSGB). The report outlines the methodologies used to test and validate the questionnaire items and summarises the key findings from each of the testing phases.

Background

Introduction

Gambling-related harm is defined as the negative impacts of gambling on individuals' resources, relationships, and health, affecting both people who gamble and those close to them (Wardle et al., 2018). One of the Gambling Commission's key licensing objectives is to ensure that children and vulnerable people are protected from gambling-related harms, and in our Evidence Gaps and Priorities we have highlighted the need for further research to establish the ways in which people may experience gambling harms and who is most at risk. To help address this evidence gap, we collaborated with National Centre for Social Research (NatCen) and researchers at the University of Glasgow to develop and validate a set of survey questions that explore people's experiences of gambling and its associated consequences. These questions have been included in the new Gambling Survey for Great Britain (GSGB), which uses a push-to-web methodology to provide estimates of gambling participation on an annual and quarterly basis.

Consultations with stakeholders, including those with lived experience of gambling-related harm, have highlighted the variability in people’s experiences of gambling and its consequences. To reflect this variability and align with terminology used within established frameworks of gambling-related harm, our future publications will refer to these questions as assessing the ‘adverse consequences from gambling’.

The importance of understanding gambling-related harms

The Gambling Commission has previously relied on the Problem Gambling Severity Index (PGSI) (Ferris & Wynne, 2001), and a screening tool adapted from the Diagnostic and Statistical Manual of the American Psychiatric Association (DSM-IV) as proxy measures of gambling-related harm (Fisher, 2000). The DSM-IV screening questionnaire consists of 9 items, and individuals who score 3 or more are thought to be experiencing problem gambling. The PGSI is a 9-item validated scale that measures gambling behaviours and negative experiences of gambling, and categorises individuals into 4 groups based on their scores:

The PGSI and DSM-IV measure have been included within the health surveys of England, Scotland, and Wales. These surveys gathered data on gambling behaviour from approximately 4,000 adults per nation each year through interviewer-led surveys. The Commission also regularly tracked trends in the prevalence of people scoring 8 or more on the PGSI via its quarterly telephone survey, although this was conducted with a relatively small sample and used a short-form version of the PGSI (that is, the PGSI mini screen; Volberg & Williams, 2012 (opens in new tab) (PDF)), and was therefore not used for official statistics.

In recent years, rates of participation in face-to-face and interviewer-led surveys, such as the quarterly telephone survey and health surveys, have declined (Ipsos MORI 2012 (opens in new tab) (PDF), Sturgis, 2024 (opens in new tab)). This trend has prompted the need to develop new cost-effective methods to obtain reliable and representative data from the general population. We have therefore developed a new survey, the Gambling Survey for Great Britain (GSGB), to gather data on gambling participation and its potential impacts. The GSGB uses a push-to-web methodology which eliminates the need for an interviewer to be present and helps to reduce the likelihood of socially desirable responses. The survey content has also been updated to include more detailed categories of online gambling activities, reflecting the modern gambling landscape.

In a recent review, (opens in new tab) (PDF), Professor Patrick Sturgis endorsed the GSGB methodology, and noted that the push-to-web approach yields high-quality estimates of gambling prevalence in Great Britain. The GSGB will serve as the primary source for estimating the prevalence of people scoring 8 or more on the PGSI, and will provide regular and consistent reporting of gambling participation across England, Scotland, and Wales. Unlike the health surveys that used both the PGSI and the DSM-IV screening tool, the GSGB will only use the PGSI to estimate experiences of problem gambling. This aligns with stakeholder suggestions that including both measures would be unnecessary given the development of additional questions relating to gambling-related harms. Importantly, due to changes in survey methodology, GSGB estimates of the prevalence of people who score 8 or more on the PGSI are not comparable with findings from previous gambling or health surveys. Nonetheless, the GSGB provides a new baseline against which future trends can be compared. For a discussion of the methodological differences that may account for variations in survey estimates of gambling participation and PGSI scores, see Sturgis (2024) (opens in new tab).

The GSGB has been developed and tested by the Commission in collaboration with National Centre for Social Research (NatCen) and the University of Glasgow. In the first stage of development, we conducted a consultation with a range of stakeholders to gather their ideas on the design and content of the new survey. Following the consultation, a pilot phase was conducted to test the feasibility and effectiveness of using a push-to-web methodology to measure gambling participation, gambling harms, and the prevalence of people scoring 8 or more on the PGSI. The experimental phase involved further testing and refinement of the survey methodology, which was conducted across 3 stages. Experiments 1 and 2 focused on refining survey questions and data collection methods, while step 3 aimed to finalise the survey design and prepare for full implementation. The GSGB has now transitioned to a continuous data collection phase, with the goal of collecting data from 20,000 participants each year (further information about the development of the GSGB is available).

While the GSGB will continue to gather data on the prevalence of people scoring 8 or more on the PGSI, we recognise the need to develop a more comprehensive understanding of gambling-related harm. This has been emphasised by previous research which discusses the limitations of relying on the PGSI as a proxy measure of the negative impacts of gambling (Langham et al., 2016 (opens in new tab) ; Browne et al., 2017)(opens in new tab) (PDF). A recent Ipsos report commissioned by GambleAware (2023) (opens in new tab)(PDF) argued that by combining questions about gambling behaviours and their consequences into a single measure, the PGSI does not differentiate between gambling behaviours and the harms they cause. The authors also point out that the PGSI primarily focuses on the experiences of the individual gambler and does not adequately capture the wider impact of gambling on friends, family members, and the broader community. Addressing these limitations is crucial because the potential consequences of gambling can extend beyond the individual, leading to financial hardship, emotional distress, and relationship breakdown for those close to someone who gambles. Given that the PGSI was designed to measure symptoms of gambling disorders, it may not optimally assess the nature and severity of gambling-related harm experienced by individuals and others within their social circles.

Based on the need to develop a comprehensive understanding of gambling-related harms, we have developed a series of questions to be included in the GSGB. During the initial consultation phase of the GSGB development, all stakeholders unanimously recognised the importance of monitoring gambling-related harm. However, there was some disagreement about how gambling-related harm should be assessed. To address this, we set out the following objectives.

  1. To develop and validate a new suite of survey questions about the negative impacts of gambling on people who gamble, and those who know someone who gambles, using both quantitative and qualitative methods.
  2. To incorporate recommendations from a wide range of stakeholders in the development of the new questions.
  3. To conduct rigorous analysis to guide decisions about how gambling-related harms should be reported in official statistics publications.

By achieving these objectives, we aim to develop an optimal method for monitoring the negative impacts of gambling. This will provide valuable insight into how gambling can affect individuals and their friends and family and will help to identify risk factors that may be associated with gambling-related harm (such as gambling product, method of play, and so on).

Existing frameworks and measures for understanding gambling harms

Several frameworks have been developed to provide a theoretical understanding of gambling-related harms. For example, Langham et al.'s (2016) (opens in new tab) (PDF) framework categorises harms into domains such as financial, health, relationship and emotional and/or psychological impacts. The framework also emphasises the temporal nature of gambling harms, from immediate distress to long-term impacts on personal and social wellbeing (Figure 1).

Figure 1. The Langham Conceptual Framework (Langham et al., 2016)

Figure 1: The Langham Conceptual Framework of Gambling Related Harms (Langham et al., 2016)

The framework categorises gambling-related harms into seven dimensions:

  1. Financial harm.
  2. Relationship disruption, conflict or breakdown.
  3. Emotional or psychological distress.
  4. Decrements to health.
  5. Cultural harm.
  6. Reduced performance at work or study.
  7. Criminal activity.

The framework also contains a Temporal Category, classifying harms into ‘General harms’, which are ongoing and pervasive, ‘Crisis harms’, which are acute and occur in specific situations, and ‘Legacy harms’, which are long-term consequences that persist over time.

Similarly, Wardle et al.'s (2018) (opens in new tab) (PDF) 'framework for action' (Figure 2) groups harms into impacts on resources, relationships and health, and emphasises how gambling can affect the lives of those close to them.

Figure 2. ‘Framework for Action’ (Wardle et al., 2018)

Figure 2. ‘Framework for Action’ (Wardle et al., 2018)

Wardle's Framework for Action (2018), which categorises gambling-related harms into three domains:

  1. Resources, including work and employment, money and debt, and crime.
  2. Health, including physical health, psychological distress, and mental health.
  3. Relationships, including partners, family and friends, community.

The diagram shows the interrelation between these domains, highlighting how harm in one area can affect and be affected by the others.

Wardle et al.’s (2018) framework for action was developed in collaboration with the Gambling Commission to guide the assessment and monitoring of gambling-related harms.

The framework provides a pragmatic approach by focusing on measurable aspects of gambling harm, and offers actionable insights for policy-making and interventions. While frameworks differ with regards to their organisation of harms, they share a similar understanding of the types of adverse consequences that are associated with gambling (Marionneau, Egerer, & Raisamo, 2023)(opens in new tab) (PDF).

Both the Langham et al. (2016) and Wardle et al. (2018) frameworks provide a theoretical basis for developing measures of gambling-related harm. In particular, a 72-item harm checklist was developed based on Langham's taxonomy and assesses gambling harms experienced within domains of finances, health, relationships, emotional wellbeing, work and study and cultural or social factors (Li et al., 2016) (opens in new tab).

The checklist provides a comprehensive measure of gambling-related harm, but the length of the scale limits its utility in large-scale population studies. The Short Gambling Harms Screen (SGHS) (Browne et al., 2018) (opens in new tab) provides a concise alternative, consisting of 10 harm items within the domains relationships, finances, and health. The SGHS has good internal reliability and validity, as demonstrated by a significant association with personal wellbeing. However, Price et al. (2021) (opens in new tab) (PDF) suggests that the SGHS may conflate potentially minor harms, or 'opportunity costs', with more severe consequences of gambling. Furthermore, while the SGHS provides a valid measure of gambling-related harms experienced by people who gamble, it does not assess the impact of gambling on people’s wider social networks. Another limitation of the SGHS is that the majority of items focus on financial harms, and so the scale may not provide a comprehensive assessment of other negative impacts of gambling (such as impacts on relationships and health). While a revised 18-item version of the scale (the SGHS-18) provides a more comprehensive measure of gambling harm, there is a need for more concise measures that can be easily incorporated within larger questionnaires and population surveys (Latvala, Browne, Rockloff, & Salonen, 2021) (opens in new tab).

While psychometric measures of gambling-related harm are useful for certain purposes, they provide limited insight into the complexities of harms and their broader impacts. We therefore did not aim to develop a new psychometric measure, but rather to develop a new set of survey questions that provide a more comprehensive and nuanced understanding of gambling-related harms. Guided by Wardle et al.'s (2018) ‘framework for action’, we included survey questions that assess impacts on people’s resources, relationships, and health, as well as the wider consequences for people’s friends and family.

Survey development, testing, and selection of items

Initial selection of items

In June 2020, the Gambling Commission began developing survey questions to assess the adverse consequences of gambling. An initial set of 27 items was selected from the 72-item checklist developed by Li et al. (2016). The chosen items aligned with Wardle et al.'s (2018) framework and covered 3 main categories of harm: relationships, resources, and health. These items were tested in multiple waves of the Commission's online tracker survey, with approximately 2,000 participants taking part in each wave between June 2020 and June 2021. Participants were asked whether they had experienced each of the adverse consequences in the last 12 months because of their own or someone else’s gambling (response options: ‘Yes - as a result of my own gambling’, ‘Yes – as a result of someone else’s gambling’, ‘No’).

Results from these questions within the online tracker were analysed to inform our initial selection of items and have therefore not been published externally. The findings showed that the most commonly reported negative effects of one’s own gambling, amongst those who reported at least one gambling-related harm, were feelings of shame, guilt and embarrassment. Other commonly reported adverse consequences included reduced savings, reduced spending on leisure activities, reduced spending on everyday essentials, spending less time with loved ones and experiencing more conflict with friends and family. People who had been affected by someone else's gambling reported similar adverse consequences and were also likely to report taking out more credit or borrowing money and experiencing dishonesty in relationships.

Questions relating to the adverse consequences of gambling included in the online tracker survey

Health

These are the response options included in the online tracker survey:

Financial

These are the response options included in the online tracker survey:

Relationships

These are the relationship response options included in the online tracker survey:

Refining items

To refine the measure for the pilot phase of the Gambling Survey for Great Britain (GSGB), the Gambling Commission conducted a principal components analysis (PCA) using data obtained from the online tracker. This analysis helped identify the most important items and reduced the number of questions to a smaller subset. To ensure sufficient breadth of coverage across different domains of harm, the final selection of items was not solely based on the results from the PCA, but also included questions that captured a wide range of potential negative impacts.

The refined set of questions were reviewed by Nation Centre for Social Research's (NatCen) Questionnaire Development and Testing Hub. The decision was made to retain questions that were most frequently reported during the initial testing phase, along with questions about more severe consequences , such as crime, violence and significant financial loss.

The final list for the GSGB pilot phase consisted of 14 items, 6 of which were adapted from the Problem Gambling Severity Index (PGSI) and the Diagnostic and Statistical Manual of the American Psychiatric Association (DSM-IV) due to the similarity of some of the statements being used. Additional questions were also included to assess whether people had accessed various support service access due to gambling (mental health, financial advice) as further indicators of potentially negative experiences. One of the more severe harms covered was that of suicide attempts and suicide ideation caused by one’s own gambling. Questions relating to suicide attempts and ideation were adapted from the Adult Psychiatric Morbidity Study (AMPS).

Questions relating to adverse consequences from gambling included in GSGB pilot survey

Health

These are the health questions included in included in GSGB pilot survey:

Financial

These are the finance questions included in included in GSGB pilot survey:

Relationships

These are the relationship questions included in included in GSGB pilot survey:

1 Question derived from PGSI.
2 Question derived from DSM-IV screening tool.

Pilot study

In January 2022, the Gambling Commission conducted a pilot survey of questions about the adverse consequences from gambling as part of the Gambling Survey for Great Britain (GSGB) development. Respondents who had gambled in the past 12 months answered questions about adverse consequences experienced due to their own gambling. Respondents who said that someone close to them gambled answered similar questions about harms resulting from someone else's gambling.

Over 1,000 participants aged 16 and above took part in the survey. Items that referred to serious negative consequences from gambling, even if experienced only once, were classified as 'Type 1' or ‘severe’ harms (for example violence or abuse). 'Type 2' harms describe adverse consequences that can have a more cumulative and gradual negative impact on people's lives (for example reduced spending on everyday items).

Respondents were asked to indicate the presence of 'Type 1' harms using a binary 'yes' or 'no' response, while 'Type 2' harms were reported on a three-point scale: 'Not at all,' 'A little,' or 'A lot.' Questions that were taken from the Problem Gambling Severity Index (PGSI) and Diagnostic and Statistical Manual of the American Psychiatric Association (DSM-IV) retained their original response format (PGSI: 'Never', 'Sometimes', 'Fairly often', 'Very often'; DSM-IV: 'Never', 'Sometimes', 'Most of the time', 'Almost always').

Adverse consequences from own gambling

As expected, endorsement of 'Type 1' or severe harms was low, with between 2 and 3 participants reporting experience of each harm. For other adverse consequences (that is, 'Type 2' harms), the distribution of responses across the response options deviated from the expected pattern. Typically, for scaled response options where the measured experience is relatively rare, endorsement rates tend to decrease as severity or frequency increases. Contrary to this expectation, the pilot study data showed a higher proportion of past year gamblers who reported experiencing each impact 'a lot' (2 to 3 percent) than those who reported experiencing each impact 'a little' (1 to 2 percent). These findings indicated that the response options may not have effectively captured the frequency or severity of gambling-related harm, and highlighted the need for further refinement of the response format.

Adverse consequences experienced from someone else’s gambling

Endorsement rates for 'Type 1' or severe harms caused by someone else’s gambling were generally low. One participant reported losing something of significant financial value and 9 participants reported a breakdown in relationships. For items referring to other potential negative consequences (that is, 'Type 2' harms), the percentage of respondents who reported experiencing each harm 'a little' ranged from 2 percent (seven participants), for 'increased use of credit', to 7 percent (22 participants) for 'increased conflict or arguments.' Approximately 2 percent of respondents reported experiencing other adverse consequences (that is, 'Type 2' harms) 'a lot.' Findings also indicated some evidence of under-reporting; while 40 percent of respondents said they lived with someone who gambles, only 29 percent said that someone close to them gambles. This discrepancy may be due to a lack of clarity in the definition of gambling in the question or to the method of filtering respondents into the 'harm from others' questions.

Cognitive testing

Cognitive testing was conducted with 14 participants to assess their understanding of the survey questions. While some participants found the questions slightly uncomfortable or intrusive, most reported that the questions were clear, straightforward, and covered an appropriate range of gambling-related harms. Some areas for improvement were also highlighted, such as the need to define certain terms and to clarify that the questions relate specifically to the effects of gambling.

Peer review

Two external academics, Professor Robert Williams and Dr. Rachel Volberg, were invited to evaluate the Gambling Commission's methodology for developing the harms questions and to make recommendations for refining the items. Professor Williams and Dr Volberg commended the survey as well-designed and coherent, and agreed that the items captured a wide range of gambling-related harms. They also supported the inclusion of questions about harm caused by someone else's gambling.

Professor Williams and Dr Volberg suggested several areas for improvement. They emphasised the need to ensure a balanced representation of questions across different domains of gambling related harm. Additionally, they recommended including a question related to productivity, such as absenteeism from school or work, to broaden the range of consequences captured by the survey. They also discussed the unexpected performance of the scaled response options, in which fewer respondents reported experiencing each consequence 'a little' than 'a lot'. To address this issue, they recommended modifying response options to ensure they are clear and evenly spaced in terms of severity or frequency. Professor Williams and Dr. Volberg also suggested conducting experimental research to compare the results obtained using binary (yes or no) questions with those obtained using scaled response options. This comparison would help determine the most effective way to capture the experience of gambling-related harm.

Testing binary versus scaled response options (Experimental phase)

In order to address the issues identified in the pilot survey, the experimental statistics phase evaluated the performance of two different response formats for questions relating to other potential adverse consequences of gambling. Specifically, the aim was to compare responses when these consequences were measured using a 4-point scale (never, occasionally, fairly often, very often) and binary 'yes' and 'no' options. A question about absence from work or studying was included following suggestions from Dr Volberg and Professor Williams. Over 4,000 participants were administered questions using the scaled response options (Condition A), and over 2,000 participants were administered questions using binary response options (Condition B).

The experimental phase tested different response options to the items that used a scaled response format. The decision was made to remove the item that was adapted from the Diagnostic and Statistical Manual of the American Psychiatric Association (DSM-IV) (such as, ‘Asked others to provide money to help with a desperate financial situation caused by gambling’) as it was thought to overlap with an item from the Problem Gambling Severity Index (PGSI) (such as, ‘Have you borrowed money or sold anything to get money to gamble’). Given that 3 of the items were derived from a validated scale (such as, the PGSI), we did not test responses to these items.

The following 6 items were included in the experimental phase:

Each item was asked to respondents who had gambled in the past 12 months and to those who knew someone who had gambled. To address possible under-reporting of ‘consequences from someone else’s gambling', the screening question was refined to remind participants of the different types of gambling and to answer this question even if people close to them only gambled occasionally. More than half (57 percent) of participants said they knew someone who gambled. This figure was closer to past-year gambling participation rates, suggesting that under-reporting was successfully addressed.

Response patterns for the 4-point scale were as expected, with fewer respondents reporting experiencing each harm 'very often' than 'occasionally'. Scaled responses showed better convergent validity than binary responses and were significantly associated with several measures of well-being, including the Short Warwick-Edinburgh Mental Well-being Scale (SWEMWBS), and alcohol intake. For most questions, 'yes' responses in Condition B (binary responses) yielded similar endorsement rates to 'fairly often’ or ‘very often' responses in Condition A (scaled responses). These findings suggest that the 4-point answer scale provided a more valid measure of harms, and more analytical opportunities than the binary 'yes' or 'no' option. The 4-point scale also demonstrated improvements over the prior 3-point scale used in the pilot study, with logical sequences of endorsement.

The results of the experimental phase had important implications for the development of the survey questions. In particular, the decision was made to retain the 4-point response format for questions about potential adverse consequences from gambling, given its improved data quality and ability to capture a more detailed gradient of responses. It was recommended that the Gambling Commission conduct further research to identify the point at which endorsement of scaled items reflects negative impacts of gambling. In particular, further work was needed to clarify whether impacts experienced ‘occasionally’ were considered harmful. To address these recommendations, we conducted further analyses of experimental survey data, and initiated follow-up interviews with survey participants to provide insight into how harm is experienced and reported.

Evaluating 'occasionally' responses

Quantitative analyses

Data collected from step 3 of the experimental phase of the Gambling Survey for Great Britain (GSGB) was used to examine differences in Problem Gambling Severity Index (PGSI) scores and mental wellbeing across participants' responses to items about adverse consequences1. Participants who had gambled in the past 12 months and had provided valid responses to impact questions, and PGSI and mental wellbeing scales, were included in the analyses (PGSI: 2,222 participants; SWEMWBS: 2,210 participants). Due to low base sizes, 'very often' and 'fairly often' responses were combined for analysis.

To check the validity of the results, we calculated the probability value (p-value) which is a method of understanding how likely is it that the results are due to chance. We used a scale of 0 to 1 where a lower number suggested the results were unlikely to be due to chance.

Results showed that participants who reported experiencing each impact 'fairly often’ or ‘very often' had significantly higher PGSI scores, on average, compared with those who answered 'occasionally' or 'never'. Similarly, those who 'occasionally' experienced each impact had significantly higher PGSI scores than those who had 'never' experienced each impact (Figure 3) (p less than .01 in all comparisons).

Results showed that participants who reported experiencing each potential adverse consequence 'fairly often’ or ‘very often' had significantly higher PGSI scores, on average, compared with those who answered 'occasionally' or 'never'. Similarly, those who 'occasionally' experienced each potential adverse consequence had significantly higher PGSI scores than those who had 'never' experienced each consequence (Figure 3) (p-value less than .01 in all comparisons).

We also found that mental wellbeing scores (Short Warwick-Edinburgh Mental Well-being Scale (SWEMWBS)) were significantly lower (indicating poorer mental wellbeing) amongst those who had experienced each adverse consequence 'occasionally' compared to those who had 'never' experienced each consequence (p-value less than .05 in all comparisons). For 4 of the 6 items ('reduced spend on every day items', 'use of savings', 'lie to family', 'absent from work'), mental wellbeing scores did not differ significantly between those who answered 'very or fairly often' and 'occasionally' (Figure 4). All analyses should be interpreted with caution due to low base sizes amongst those who reported each potential adverse consequence ‘very or fairly often’ and ‘occasionally’2.

Analyses were repeated to examine mental wellbeing scores amongst people who knew someone who gambles (1,660 participants). Participants who reported each potential adverse consequence ‘occasionally’ had lower SWEMWBS scores (indicating poorer mental wellbeing) relative to those who had ‘never’ experienced each potential adverse consequence. For 4 items (‘conflict’, ‘isolated’, ‘lie to family’, and ‘absent from work’), SWEMWBS scores did not differ significantly between those who had experienced these consequences ‘occasionally’ compared with those who had experienced each consequence ‘very or fairly often’ (Figure 5). All analyses should be interpreted with caution due to the low base sizes amongst those who reported each consequence ‘very or fairly often’ and ‘occasionally’3.

Figure 3. Mean PGSI score by response to questions about potential adverse consequences amongst people who had gambled in the past year

Figure 3: Mean PGSI score by response to impact questions amongst people who had gambled in the past year

(Unweighted base = 2,222). All comparisons were significant with p-values less than .01. Error bars represent standard error of the mean.

Mean PGSI score by response to potential adverse consequences questions amongst people who had gambled in the past year.
Response Never (mean) Occasionally (mean) Very or fairly often (mean) Never (standard error of the mean) Occasionally (standard error of the mean) Very or fairly often (standard error of the mean)
Spending on everyday items 0.50 6.79 15.09 0.04 0.54 1.04
Savings 0.50 6.38 13.50 0.04 0.59 1.00
Conflict or arguments 0.61 7.05 14.89 0.04 0.85 0.90
Isolated 0.56 6.05 14.37 0.04 0.67 0.95
Lie to family 0.52 5.65 13.16 0.04 0.58 0.85
Absent from work 0.60 9.07 14.82 0.04 0.91 0.92

Figure 4. Mean mental wellbeing score (SWEMWBS) by response to potential adverse consequences questions amongst people who had gambled in the past year

Figure 4: Mean mental wellbeing score (SWEMWBS) by response to impact questions amongst people who had gambled in the past year

(Unweighted base = 2,210). Lower scores indicate poorer mental wellbeing. Significant differences are denoted by different letters. Error bars represent standard error of the mean.

Mean mental wellbeing score (SWEMWBS) by response to impact questions amongst people who had gambled in the past year
Question Very or fairly often (mean) Occasionally (mean) Never (mean) Very or fairly often (standard error) Occasionally (standard error) Never (standard error) Statistical comparisons
Spending on everyday items 21.17 20.64 23.10 0.82 0.46 0.09 SWEMWBS scores for people who responded 'never' were significantly higher than for those who responded 'occasionally' or 'Fairly or very often'.
Savings 20.98 20.84 23.09 0.83 0.47 0.09 SWEMWBS scores for people who responded 'never' were significantly higher than for those who responded 'occasionally' or 'fairly or very often'.
Conflict or arguments 19.79 22.00 23.04 0.71 0.67 0.09 SWEMWBS scores for people who responded 'never' were significantly higher than those who responded 'occasionally' or 'fairly or very often'. The mean SWEMWBS scores for people who responded 'occasionally' were significantly higher than those who responded 'fairly or very often'.
Isolated 19.23 21.38 23.10 0.46 0.68 0.09 SWEMWBS scores for people who responded "never" were significantly higher than those who responded "occasionally" or "fairly or very often". The mean SWEMWBS scores for people who responded “occasionally” were significantly higher than those who responded "fairly or very often."
Lie to family 20.30 21.06 23.09 0.53 0.53 0.09 SWEMWBS scores for people who responded 'never' were significantly higher than for those who responded 'occasionally' or 'fairly or very often'.
Absent from work 20.44 19.64 23.07 0.83 0.65 0.09 SWEMWBS scores for people who responded 'never' were significantly higher than for those who responded 'occasionally' or 'fairly or very often'.

Figure 5. Mean mental wellbeing score (SWEMWBS) by response to potential adverse consequences questions amongst people who know someone who gambles

Figure 5. Mean mental wellbeing score (SWEMWBS) by response to impact questions amongst people who know someone who gambles

(Unweighted base = 1,660). Lower scores indicate poorer mental wellbeing. Significant differences are denoted by different letters. Error bars represent standard error of the mean.

Mean mental wellbeing score (SWEMWBS) by response to potential adverse consequences questions amongst people who know someone who gambles.
Question Very or fairly often (mean) Occasionally (mean) Never (mean) Very or fairly often (standard error) Occasionally (standard error) Never (standard error) Statistical comparisons
Spending on everyday items 19.32 21.21 22.86 0.71 0.71 0.10 SWEMWBS scores for people who responded 'fairly or very often' were significantly lower than for those who responded 'occasionally' or 'never'. The mean scores for those who responded 'occasionally' were significantly lower than for those who responded 'never'.
Savings 19.38 21.74 22.84 0.65 0.82 0.10 SWEMWBS scores for people who responded 'fairly or very often' were significantly lower than for those who responded 'occasionally' or 'never'. SWEMWBS scores for those who responded 'occasionally' were significantly lower than for those who responded 'never'.
Conflict or arguments 21.24 21.09 22.89 0.62 0.46 0.10 SWEMWBS scores for people who responded 'never' were significantly higher than for those who responded 'occasionally' or 'fairly or very often'.
Isolated 20.60 19.96 22.88 0.77 0.60 0.10 SWEMWBS scores for people who responded 'never' were significantly higher than for those who responded 'occasionally' or 'fairly or very often'.
Lie to family 21.13 20.94 22.87 0.60 0.57 0.10 SWEMWBS scores for people who responded 'never' were significantly higher than for those who responded 'occasionally' or 'fairly or very often'.
Absent from work 21.14 19.56 22.83 0.73 0.92 0.10 SWEMWBS scores for people who responded 'never' were significantly higher than for those who responded 'occasionally' or 'fairly or very often'.

Qualitative analysis

To supplement the analysis conducted with Step 3 data and gain further insight into how participants had understood the response option 'occasionally', we conducted follow-up qualitative research in collaboration with National Centre for Social Research (NatCen). Participants (16 participants) who reported 'occasional' or multiple instances of gambling-related harm, and those who had experienced adverse consequences due to someone else’s gambling, were invited to attend follow-up interviews.

Participants described 'occasional' harms using a range of frequencies and impacts. Some participants perceived occasional harms to be less impactful than more frequent harms, while others said that even infrequent harms could lead to substantial negative impacts, such as serious financial hardship or strained relationships. The potential negative impacts of ‘occasional’ harms aligns with findings from our quantitative analyses, in which we found an association between ‘occasional’ adverse consequences and lower mental wellbeing. Overall, our findings suggest that 'occasional' adverse consequences from gambling may indicate some degree of gambling-related harm.

1 Analyses were conducted using univariate Analysis of Variance (ANOVAs) with PGSI/SWEMWBS as dependent variables and each question as independent variables. For each question, pairwise comparisons using Least Squared Differences (LSD) were conducted to identify significant differences in PGSI or SWEMWBS scores amongst those who had answered ‘never’, ‘occasionally’, and ‘fairly or very often’. LSD was selected for pairwise comparisons due to the small number of comparisons needed for each item, and to optimise power to detect significant differences between groups.

2 The number of participants who answered ‘very often’ or ‘fairly often’ to questions about potential adverse consequences from one’s own gambling varied by question but ranged from 35 participants (‘reduced spend on everyday items’) to 60 participants (‘lie to family’). The number of participants who answered ‘occasionally’ ranged from 45 participants (‘absent from work’) to 104 participants (‘reduced spend on everyday items’). These numbers represent the weighted base sizes that were used in the ANOVA.

3 The number of participants who answered ‘very often’ or ‘fairly often’ to questions about potential adverse consequences from someone else’s gambling varied by question but ranged from 22 participants (‘absent from work’) to 41 participants (‘conflict or arguments). The number of participants who answered ‘occasionally’ ranged from 30 participants (‘absent from work’) to 94 participants (‘conflict or arguments’). These numbers represent weighted base sizes that were used in the ANOVA.

Further validation of survey questions

Further analyses were conducted to examine the reliability and incremental validity of the adverse consequences items over and above the Problem Gambling Severity Index (PGSI). To measure internal consistency and determine the reliability of the data, we used Cronbach’s alpha which works on a scale on 0 to 1 where a value closed to 1 means a higher reliability. Reliability analysis showed that items assessing severe consequences and other potential adverse consequences (including the 3 items taken from the PGSI), demonstrated good internal consistency, suggesting that all questions assessed the same underlying construct (Cronbach's alpha = 0.894). To assess the incremental validity of the measure, we examined whether the number of items endorsed predicted variance in mental wellbeing scores over and above that accounted for by the PGSI, amongst people who had gambled in the past year. A two-step hierarchical linear regression was conducted with PGSI scores entered in step 1 of the model and the number of items endorsed, excluding PGSI items, entered in step 2. Mental wellbeing scores (assessed using the SWEMWBS (Short Warwick-Edinburgh Mental Well-being Scale)) were included as the outcome variable.

The regression model showed that mental wellbeing was significantly associated with PGSI scores, and the number of harms endorsed. In step 1 of the model, higher PGSI scores were associated with lower scores on the SWEMWBS (indicating lower mental wellbeing). In step 2 of the model, including the number of adverse consequences endorsed significantly improved model fit (probability value (p-value) less than .001); lower SWEMWBS scores were associated with experiencing a greater number of impacts.

Overall, our findings suggest that the selected items provide a reliable and valid measure of the potential adverse consequences of gambling. While we do not aim to develop a psychometric tool, it is important to ensure that the questions adequately capture the key aspects of gambling-related harm. By demonstrating good psychometric properties, such as internal reliability and incremental validity, we can be confident that the selected questions provide a robust foundation for monitoring trends in gambling-related harm and identifying risk factors.

Table 1: Hierarchical regression analysis of mental wellbeing scores (SWEMWBS) predicted by PGSI and number of other potential adverse consequences endorsed (2,255 participants).

Hierarchical regression analysis of mental wellbeing scores (SWEMWBS) predicted by PGSI and number of other potential adverse consequences endorsed
Model Type B (coefficient) SE B (coefficient) t-value p-value
Model 1 Constant 23.09 0.09 247.60 less than 0.001
Model 1 PGSI score -0.14 0.03 -4.74 less than 0.001
Model 2 Constant 23.10 0.09 248.21 less than 0.001
Model 2 PGSI score -0.01 0.05 0.27 0.780
Model 2 Harms (including "occasionally") -0.40 0.11 -3.49 less than 0.001

Note. SWEMWBS = Short Warwick-Edinburgh Mental Well-being Scale. SE = Standard Error. Model 1: Adj. R2 = 0.009. Model 2: Adj. R2 =0.014. R² change = .005 (p < .001).

Limitations

The inclusion of survey questions relating to the adverse consequences of gambling will provide important insight into the effects of gambling on individuals and their friends and family. However, it is important to recognise some of the limitations inherent in survey research. Firstly, because the survey relies on self-reporting, responses are determined by personal perceptions and recollections, which can be influenced by factors such as social desirability and recall bias. As a result, the accuracy of information on adverse consequences from gambling largely depends on respondents' subjective experiences and their willingness to share sensitive information. Ideally, a range of subjective and objective methods is needed to provide robust insight into the impacts of gambling on individuals and society. Secondly, while the Gambling Survey for Great Britain (GSGB) invited people to take part in the survey using a random-probability method of sampling, participants ultimately decided whether or not they would like to take part. This may limit the generalisability of the findings to the wider population due to potential differences between those who chose to participate and those who declined.

Finally, it is important to recognise the complexity of accurately measuring adverse consequences from gambling. The potential effects of gambling can manifest in many ways and are influenced by a range of personal, social and environmental factors. Capturing this complexity in a survey format is inherently challenging, and therefore our questions may not include all aspects of gambling-related harms. To help mitigate these limitations, the GSGB has been rigorously tested through both quantitative and qualitative methodologies and incorporates recommendations from a wide range of stakeholders, including people with lived experience of gambling harms. Furthermore, our survey questions were developed based on existing frameworks of harm, which helps to ensure that they are conceptually sound and capture a range of adverse consequences associated with gambling. This comprehensive approach helps to ensure that the survey provides a valid and reliable estimate of the adverse consequences from gambling across representative samples of the population. A full review of the strengths and limitations of the overall GSGB methodology is available.

Implications for reporting the negative impacts of gambling

Through careful consideration of findings from our development work, we have made the following decisions about how gambling-related harms should be reported in our annual official statistics.

Consultations with stakeholders

Consultations with stakeholders, including those with lived experience of gambling-related harm, have highlighted the variability in people’s experiences of gambling and its consequences. To reflect this variability, we will use the term ‘adverse consequences from gambling’ when reporting findings from the new survey questions. This terminology aligns with established frameworks of gambling-related harm. For example, ‘adverse consequences’ is used by Langham et al. (2016) to describe the range of harms experienced from gambling, and within PGSI definitions of low-risk, moderate-risk, and problem-gambling. While the term is often used interchangeably with ‘negative consequences’ (which is the term used in the ICD-11 definition of gambling disorder), we opted for ‘adverse consequences’ to maintain consistency with the harms frameworks that guided the development of our survey questions.

Distinguishing consequences

We distinguish between ‘severe adverse consequences’, and ‘other potential adverse consequences’ experienced due to gambling.12

Annual report

The annual report highlights the percentage of people who had gambled in the past year who experienced at least one severe adverse consequence from gambling (such as, significant financial loss, relationship breakdown, violence or abuse, and committing a crime). The percentage of people experiencing each of these harms is also be reported.

Reporting severe adverse consequences separately

Due to the unequivocal negative impact of severe adverse consequences, these are reported separately from other potential adverse consequences.

Recording frequency of responses

When presenting the prevalence of ‘other potential adverse consequences’ from gambling, we decided not to provide an aggregate figure. Instead, for each question, we report the frequency of each of response to each question (such as, 'occasionally,' 'fairly often,' and 'very often’).

Recording use of support services

The report includes the percentage of people who had gambled in the past year who had accessed support services (such as, relationship counselling, gambling support, financial advice, and mental health services).

Recording suicidal ideation

Due to the need to examine associations between gambling participation and suicide, the percentage of people who experienced suicidal ideation due to one’s own gambling are reported separately from other adverse consequences.

While these form the core statistics, we will also explore associations between the negative impacts of gambling and various demographics, Problem Gambling Severity Index (PGSI) scores, mental wellbeing, and the types and number of gambling activities that people engage in. In future reports, we will also report the prevalence of negative impacts by each type of gambling activity.

1Severe adverse consequences refer to those which negatively impact people’s lives, even if experienced once.

2Other potential adverse consequences refer to experiences that can have cumulative and gradual negative impacts on people’s lives, and which may affect people to differing degrees (for example, reduced spending on everyday items).

Conclusions

This technical report describes the development and validation of survey questions within the Gambling Survey for Great Britain (GSGB) to assess adverse consequences from gambling . Following stakeholder input and rigorous testing, we selected and validated a set of survey items that capture adverse consequences caused by one's own gambling and someone else's gambling. These items cover the areas of resources, relationships and health and are consistent with established frameworks of gambling-related harm (Wardle et al., 2018). The selected items demonstrated good reliability and validity and capture unique variance in mental wellbeing beyond that captured by the Problem Gambling Severity Index (PGSI).

As we gather more data, we will continue to refine and develop our approach to assessing adverse consequences from gambling to ensure that our methods remain reliable and robust. The ongoing monitoring and assessment of potential adverse consequences from gambling is in line with our key research priorities and is an important step in understanding how gambling can affect individuals, and their family and friends.

Appendix A: Final questions included in GSGB

Other potential adverse consequences due to own gambling (response options: Very often, Fairly often, Occasionally, or Never)

The next few questions are about the impact that gambling can have on some people. Please answer as honestly as you can. All of the answers you provide will be entirely confidential. 

Thinking about your own gambling, how often in the last 12 months has your own gambling led you to:

Items derived from Problem Gambling Severity Index (PGSI) (response options: Almost always, Most of the time, Sometimes, Never)

In the last 12 months, how often:

Adverse consequences experienced due to own gambling (response options: Yes or No)

In the last 12 months:

Use of support services (response options: Yes or No)

In the last 12 months, has your own gambling led you to seek to help, support or information online, in-person or by telephone from: 

Suicidal thoughts and attempts

In the last 12 months have you ever thought about taking your own life, even though you would not actually do it? (response options: Yes or No)

In the last 12 months, have you made an attempt to take your life, by taking an overdose of tablets or in some other way? (response options: Yes or No)

To what extent, if at all, was this related to your gambling? (Response options: Not at all, a little, a lot).

Other potential adverse consequences experienced due to someone else’s gambling (response options: Very often, Fairly often, Occasionally, or Never)

The next few questions are about the impact that someone else’s gambling may have had on you, whether you live with them or not. Please answer as honestly as you can. 

Thinking about someone else’s gambling, in the last 12 months:

Severe adverse consequences experienced due to own gambling (response options: Yes or No)

Has your relationship with someone close to you such as a spouse, partner, family member or friend broken down because of someone else’s gambling? 

Have you lost something of significant financial value such as your home, business, car or been declared bankrupt because of someone else’s gambling? 

Have you experienced violence or abuse because of someone else’s gambling? 

Have you committed a crime in order to finance someone else’s gambling or to pay their gambling debts? 

Accessing support services due to someone else’s gambling (response options: Yes or No)

In the last 12 months, has someone else’s gambling led you to seek help, support or information online, in-person or by telephone from:

Appendix B: Timeline

Timeline of the key phases in the development of questions assessing the adverse consequences of gambling of gambling

  1. June to July 2023

    Qualitative research: To understand how participants had answered survey questions, interviews were conducted with 16 participants who reported 'occasional' or multiple negative impacts from their own or someone else’s gambling.

  2. October 2022 to May 2023

    Experimental phases Step 1 and 2 focused on refining survey questions and data collection methods. Step 3 aimed to finalise the survey design and prepare for full implementation.
  3. Peer review: Two academic experts, Professor Robert Williams and Dr. Rachel Volberg, were invited to evaluate the methodology and make recommendations for refining the questions.

  4. February to March 2022

    Cognitive Testing: We conducted interviews with 14 participants to assess their understanding of the survey questions.

  5. January to February 2022

    National Centre for Social Research (NatCen) pilot survey: In collaboration with NatCen, we tested a refined set of 14 questions relating to the negative impacts of gambling with over 1,000 participants.

  6. November to December 2021

    Stakeholder Consultation: We conducted a consultation with stakeholders to gather ideas on the design and content of the new survey.

  7. June 2020 to June 2021

    Initial pilot testing: An initial set of 27 questions relating to the negative impacts of gambling were tested in multiple waves of our online tracker survey.

Appendix C: References

Browne, M., Goodwin, B. C., & Rockloff, M. J. (2018). Validation of the Short Gambling Harm Screen (SGHS): A Tool for Assessment of Harms from Gambling (opens in new tab) Journal of gambling studies, 34(2), 499 to 512.

Browne, M., Rawat, V., Greer, N., Langham, E., Rockloff, M., & Hanley, C. (2017). What is the harm? Applying a public health methodology to measure the impact of gambling problems and harm on quality of life (PDF) (opens in new tab) Journal of Gambling Issues, 36, 28 to 50.

Ferris, J., & Wynne, H. (2001). The Canadian Problem Gambling Index: Final Report (PDF) (opens in new tab). Ottawa: Canadian Centre on Substance Abuse.

Fisher S. (2000). Developing the DSM-IV-DSM-IV criteria to identify adolescent problem gambling in non-clinical populations (opens in new tab). Journal of gambling studies, 16(2-3), 253–273.

GambleAware. (2023). PGSI extended summary. Retrieved June 19, 2024, from https://www.gambleaware.org (opens in new tab).

Gambling Commission. (2023). Developing survey questions capturing gambling-related harms. Retrieved June 19, 2024, from https://www.gamblingcommission.gov.uk 

Gambling Commission. (2023). Evidence gaps and priorities 2023 to 2026. Retrieved June 19, 2024, from https://www.gamblingcommission.gov.uk.

Gambling Commission. (2023). Gambling participation and the prevalence of problem gambling survey: Experimental statistics stage. Retrieved June 19, 2024, from https://www.gamblingcommission.gov.uk

Gambling Commission. (2023). Participation and prevalence stakeholder engagement report. Retrieved June 19, 2024, from https://www.gamblingcommission.gov.uk.

Ipsos. (2018). Mixed Mode Surveys: Enhancing Quality Using Online & Face-to-Face (opens in new tab). Retrieved June 19, 2024, from https://www.ipsos.co.uk.

Langham, E., Thorne, H., Browne, M., Donaldson, P., Rose, J., & Rockloff, M. (2016). Understanding gambling related harm: a proposed definition, conceptual framework, and taxonomy of harms (PDF) (opens in new tab) BMC public health, 16, 80.

Latvala, T., Browne, M., Rockloff, M., & Salonen, A. H. (2021). 18-Item version of the Short Gambling Harm Screen (SGHS-18): Validation of screen for assessing gambling-related harm among Finnish population (opens in new tab). International Journal of Environmental Research and Public Health, 18(21), 11552.

Li, E., Browne, M., Rawat, V., Langham, E., & Rockloff, M. (2017). Breaking bad: Comparing gambling harms among gamblers and affected others (opens in new tab), 33(1), 223 to 248.

Marionneau, V., Egerer, M., & Raisamo, S. (2023). Frameworks of gambling harms: a comparative review and synthesis (opens new tab)(PDF) Addiction Research & Theory, 31(1), 69 to 76.

Price, A., Hilbrecht, M., & Billi, R. (2021). Charting a path towards a public health approach for gambling harm prevention. (opens in new tab) (PDF), Journal of Public Health: From Theory to Practice, 29, 37 to 53.

Sturgis, P. (2024)  Assessment of the Gambling Survey for Great Britain (GSGB) (PDF) (opens in new tab), London School of Economics and Political Science, London, UK.

Volberg, R.A. & Williams, R.J. (2012). Developing a Short Form of the PGSI (PDF) (opens in new tab), Report to the Gambling Commission.

Wardle, H., Reith, G., Best, D., McDaid, D., & Platt, S. (2018). Measuring gambling-related harms: a framework for action (PDF) (opens in new tab), Gambling Commission.