With this document you can:

This box is not visible in the printed version.

Exploring demographic differences in adverse consequences from gambling

Examining whether associations between 'at-risk' gambling behaviour (measured using the PGSI) and adverse consequences vary across demographic groups.

Published: 16 July 2026

Last updated: 16 July 2026

This version was printed or saved on: 16 July 2026

Online version: https://www.gamblingcommission.gov.uk/report/exploring-demographic-differences-in-adverse-consequences-from-gambling

Executive summary

This analysis contributes to Theme 3 of our evidence roadmaps, which highlights the need to understand how gambling-related harms are experienced and who may be most at risk. To assess the prevalence of gambling harm within the population, the Gambling Survey for Great Britain (GSGB) includes a set of survey questions that focus on some of the negative impacts from gambling. The questions distinguish between severe consequences, which are clearly and unequivocally harmful (for example, relationship breakdown or experiences of violence), and potential adverse consequences which vary in severity and can have more cumulative effects on people’s lives (for example, reduced spending on everyday items).

Our previous research suggests that some people may experience potential and severe adverse consequences despite scoring between 0 and 2 (indicating either ‘non-problem’ or ‘low-risk’ gambling) on the Problem Gambling Severity Index (PGSI). The aim of this report was to explore how associations between PGSI scores and potential and severe adverse consequences vary across different demographic groups.

Key findings

After controlling for other demographics and PGSI scores, participants from ethnic minority backgrounds, those living in lower-income households, and younger participants were more likely to report both potential and severe adverse consequences than their respective comparison groups.

For severe consequences, these demographic differences were broadly similar across the PGSI scale.

For potential adverse consequences, differences by ethnicity and age were most pronounced at lower PGSI scores and reduced as scores increased:

These findings reflect an important distinction between at-risk gambling behaviour, which is captured by the PGSI, and the consequences that people experience.

Similar PGSI scores may correspond to different levels of adverse consequences depending on individuals’ wider circumstances, and so relying on PGSI scores alone risks masking important differences in vulnerability across the population. It is therefore important to monitor the prevalence of adverse consequences alongside behavioural risk indicators, such as the PGSI, to ensure we gain a better understanding of gambling-related harm.

Introduction

Gambling-related harm can have wide-ranging consequences for people's lives, affecting their finances, their relationships, and their health. One of the Gambling Commission's key priorities is to develop our understanding of how these harms are experienced and who is most at risk (the Commission, 2025b).

The Problem Gambling Severity Index (PGSI) (Ferris & Wynne, 2001 (opens in new tab)) is often used to assess patterns of gambling associated with increased risk. It consists of 9 questions and classifies participants into one of 4 categories based on their total score: 'non-problem gambling' (0), 'low risk' (1 to 2), 'moderate risk' (3 to 7), and 'problem gambling' (8 or more). While the PGSI is a valid tool for assessing gambling risk, it was not designed to measure gambling-related harm. A review commissioned by GambleAware noted that several questions within the PGSI focus on gambling behaviour (for example, 'Have you bet more than you could afford to lose?'), rather than the consequences of that behaviour (Ipsos, 2023 (opens in new tab)). As such, the PGSI provides only limited insight into the negative impacts that gambling can have on people's lives.

To address this gap, we developed a set of survey questions that assess the adverse consequences people might experience from gambling across 3 domains: resources, relationships, and health (the Commission, 2024a; Wardle et al., 2018 (opens in new tab)). The questions distinguish between severe consequences, which are clearly and unequivocally harmful (for example, relationship breakdown or violence), and potential adverse consequences, which vary in severity and could have more cumulative effects on people's lives (for example, reducing spend on everyday items).

The aim of this report was to explore whether some people may be more susceptible to potential and severe adverse consequences at lower levels of gambling risk (as indicated by the PGSI). This builds on secondary analysis of Year 2 (2024) Gambling Survey for Great Britain (GSGB) data, which highlighted a mismatch between PGSI scores and risk of adverse consequences (Gambling Commission, 2025a): we found that 2 percent of participants who scored 0 on the PGSI (categorised as 'non-problem' gambling) reported potential adverse consequences affecting their financial resources. Conversely, 9 percent of those scoring 8 or more on the PGSI (categorised as 'problem gambling') reported no adverse consequences to their finances. These findings suggest that gambling risk and adverse consequences are not always aligned within the population, and that relying on PGSI scores alone is unlikely to capture the full picture of gambling-related harm. This raises the possibility that reliance on the PGSI risks understanding harm at lower levels of gambling risk and overstating it at higher levels, particularly if susceptibility to harm varies across the population.

This report examined how associations between PGSI scores and potential and severe adverse consequences vary across different demographic groups. We conducted secondary analysis of Year 3 (2025) GSGB data to address the following research question: do associations between 'at-risk' gambling behaviour (measured using the PGSI) and potential and severe adverse consequences vary across demographic groups? Specifically, do some demographic groups report adverse consequences at lower levels of gambling risk?

Method

Measuring adverse consequences from gambling

The 2025 Gambling Survey for Great Britain (GSGB) collected data from adults aged 18 years and older living in Great Britain (N=20,775). Fieldwork was carried out between January 2025 and January 2026. Further details, including the strengths and weaknesses of the methodology, can be found in the GSGB technical report (Gambling Commission, 2024b).

The following questions were used to assess potential adverse consequences from gambling (that is, consequences which vary in severity and can have more cumulative effects)1:

How often, in the last 12 months, has gambling led you to:

  1. reduce or cut back spending on everyday items such as food, bills, and clothing (potential adverse consequence to resources)
  2. use savings or increase use of credit, such as credit cards, overdrafts, and loans (potential adverse consequence to resources)
  3. experience conflict or arguments with friends, family, or work colleagues (potential adverse consequence to relationships)
  4. feel isolated from other people, left out, or feel completely alone (potential adverse consequence to relationships)
  5. lie to family, or others, to hide the extent of gambling (potential adverse consequence to relationships)
  6. be absent or perform poorly at work or study (potential adverse consequence to resources).

Response options were 'Never', 'Occasionally', 'Fairly often', and 'Very often'. Item endorsement was defined as responding 'Occasionally', 'Fairly often', or 'Very often'. A binary indicator was then derived to identify participants who reported one or more potential adverse consequences. Findings from our recent analysis support the use of 'one or more' potential adverse consequences as a valid population-level indicator of gambling-related harm (Gambling Commission, 2026).

To assess severe consequences from gambling (that is, those that are unequivocally harmful), participants were asked the following:

In the last 12 months:

  1. Have you lost something of significant financial value such as your home, business, car, or been declared bankrupt because of your own gambling? (severe consequence to resources).
  2. Has your relationship with someone close to you, such as a spouse, partner, family member or friend broken down because of your own gambling? (severe consequence to relationships).
  3. Have you experienced violence or abuse because of your own gambling? (severe consequence to relationships).
  4. Have you committed a crime in order to finance gambling or to pay gambling debts? (severe consequence to resources).

Response options to these questions were 'Yes' or 'No'.

Statistical analysis

Regression analyses were used to examine whether some demographic groups are more likely than others to report potential and severe adverse consequences, and whether demographic differences varied across Problem Gambling Severity Index (PGSI) scores. The following variables were included in regression models as predictors: age, sex, ethnicity, household income, educational attainment, and PGSI score2. Results from regression models were used to estimate predicted probabilities of potential and severe adverse consequences, by demographic group and PGSI score3.

1 Potential adverse consequences were also captured using 3 PGSI items:

To avoid confounding the outcome and predictor variables in subsequent analyses, responses to these items were used only in the calculation of PGSI scores and were not included as indicators of adverse consequences.

2 Logistic regression models were conducted separately for 'one or more' potential adverse consequences and severe adverse consequences. Predictor variables included PGSI score, ethnicity, household income quintile, educational attainment, age, and sex. Age was included as a binary variable (under 35 versus 35 and over) to allow comparisons with the other variables. Models also included interaction terms between PGSI score and each demographic characteristic to test whether associations between PGSI score and adverse consequences differed across groups. Regression analyses included participants who had gambled in the past 12 months, and who provided valid responses to PGSI and demographic questions.
Models also included interaction terms between PGSI score and each demographic characteristic to test whether associations between PGSI score and adverse consequences differed across groups. Regression analyses included participants who had gambled in the past 12 months, and who provided valid responses to PGSI and demographic questions.

3 Predicted probabilities were calculated by combining the model's estimated coefficients on the log-odds scale, holding all other characteristics at their reference levels (White ethnicity, household income in the third quintile or above, holds an educational qualification, female, aged 35 and over). The total was converted to a probability using the inverse-logit function. The following PGSI anchor points were used to summarise findings: PGSI = 1, PGSI = 3 and PGSI = 8 for potential adverse consequences, and PGSI = 3 and PGSI = 8 for severe consequences. Predicted probabilities were calculated for variables showing main effects and/or interactions in regression models.

Results

The sample consisted of 12,194 participants who had gambled in the past 12 months. Of these, 11.7 percent reported ‘one or more’ potential adverse consequences, 2.7 percent reported severe consequences from gambling, and 2.3 percent reported both potential and severe adverse consequences. The most frequently reported consequences were ‘reducing or cutting back spending on everyday items’ (55.2 percent), ‘lying to family or others’ (50.4 percent), and using savings or increased use of credit (50.0 percent). Among those reporting severe consequences (n=138), 64.2 percent reported ‘relationship breakdown’, 39.8 percent reported ‘losing something of significant financial value’, 42.8 percent reported ‘violence or abuse’, and 33.1 percent said that they had ‘committed a crime’ due to gambling.

Among participants who scored 0 on the Problem Gambling Severity Index (PGSI) (categorised as ‘non-problem gambling’) (n=9,747), 3.6 percent reported ‘one or more’ potential adverse consequences from gambling. Among those scoring 8 or above on the PGSI (categorised as problem gambling) (n=335), 92.7 percent reported ‘one or more’ potential adverse consequences. Tables showing the percentage of participants reporting adverse consequences from gambling, by PGSI score and demographic group, are provided in supplementary material (Table S1 and Table S2).

Potential adverse consequences

After controlling for PGSI scores and other demographic characteristics, the odds of reporting ‘one or more’ potential adverse consequences were significantly higher among participants aged under 35, people from ethnic minority backgrounds, and those living in lower-income households, compared with their respective comparison groups4. The association with age and ethnicity varied by PGSI score: differences between groups were most pronounced among those scoring lower on the PGSI, and reduced as PGSI scores increased.

The odds of reporting potential adverse consequences did not differ between male and female participants, or between those with and without an educational qualification. Results from the regression analysis can be found in supplementary Table S3.

Predicted probability of potential adverse consequences by PGSI score and demographic group

Table 1 shows the predicted probability of reporting potential adverse consequences at PGSI scores of 1, 3 and 8, by ethnicity, household income, and age. These scores were selected as anchor points because they correspond to conventional thresholds used to classify gambling risk (PGSI scores of 1, 3 and 8 represent the lower bound of ‘low-risk’, ‘moderate risk’, and ‘problem gambling’ categories, respectively).

At a PGSI score of 1, participants from ethnic minority backgrounds were more than twice as likely as White participants to report potential adverse consequences from gambling (11.8 percent compared with 5.5 percent). This difference became less pronounced as PGSI scores increased: at a score of 3, participants from ethnic minority backgrounds were 1.3 times more likely than White participants to report potential adverse consequences (32.3 percent compared with 24.0 percent), and at a PGSI score of 8 this pattern was no longer observed (see Table 1 and Figure 1).

Similarly, participants aged under 35 had higher predicted probability of potential adverse consequences at a PGSI score of 1, compared with those aged 35 and over (7.5 percent compared with 5.5 percent). This difference was not observed at PGSI scores of 3 or 8.

For those scoring 1 or 3 on the PGSI, the predicted probabilities of potential adverse consequences were slightly higher among participants in lower-income households, compared with those in higher-income households (6.4 percent compared with 5.5 percent at PGSI = 1; 25.5 percent compared with 24.1 percent at PGSI = 3). These differences were no longer observed at a PGSI score of 8.

Figure 1: Predicted probability of reporting ‘one or more’ potential adverse consequences for White and Ethnic minority participants from PGSI scores 0 to 10

Dashed reference lines show values at PGSI = 1 (low risk), PGSI = 3 (moderate risk), and PGSI = 8 (problem gambling).

Figure 1: Predicted probability of reporting ‘one or more’ potential adverse consequences for White and Ethnic minority participants from PGSI scores 0 to 10

The graph presents results for Ethnicity shown in Table 1.

Table 1: Predicted probability of reporting ‘one or more’ potential adverse consequences at PGSI = 1 (low risk), PGSI = 3 (moderate risk) and PGSI = 8 (problem gambling), by demographic group

Table 1: Predicted probability of reporting ‘one or more’ potential adverse consequences at PGSI = 1 (low risk), PGSI = 3 (moderate risk) and PGSI = 8 (problem gambling), by demographic group
Demographic group Probability at PGSI = 1 (percentage) Probability at PGSI = 3 (percentage) Probability at PGSI = 8 (percentage)
Ethnicity
Ethnic minority 11.8% 32.3% 91.9%
White 5.5% 24.1% 95.6%
Household income quintile
Higher income (third quintile or above) 5.5% 24.1% 95.6%
Lower income (second quintile or below) 6.4% 25.5% 94.0%
Age
35 years and over 5.5% 24.1% 95.6%
Under 35 years 7.5% 24.0% 90.4%

Severe consequences

After adjusting for PGSI scores and other demographic characteristics, the odds of reporting severe consequences from gambling were significantly higher among participants from ethnic minority backgrounds, those living in lower-income households, and those aged under 35, compared with their respective comparison groups (such as White ethnicity, participants living in higher-income households, and aged 35 years or older)5. The odds of reporting severe consequences did not differ by sex or educational qualification. There were also no significant interactions between PGSI score and any demographic characteristics, indicating that between-group differences did not increase or decrease across varying levels of gambling risk.

Results from regression analyses can be found in supplementary Table S4.

Predicted probability of severe consequences by PGSI score and demographic group

Table 2 shows the predicted probability of severe consequences from gambling at PGSI scores of 3 and 8, by ethnicity, household income, and age. Results for household income are also shown in Figure 2.

At a PGSI score of 3, participants from ethnic minority backgrounds were 3.5 times more likely to report severe consequences, compared with White participants (2.4 percent, compared with 0.7 percent).

The predicted probability of severe consequences was over 2 times higher for those in lower income households, and 1.7 times higher for younger participants, compared with those in higher income households and older participants, respectively (Lower income: 1.5 percent; Higher income: 0.7 percent; Aged under 35: 1.2 percent; Aged 35 and over: 0.7 percent).

Similar between-group differences were observed at a PGSI score of 8.

Figure 2: Predicted probability of reporting severe consequences for people living in higher and lower income households across PGSI scores of 0 to 10

Dashed reference lines show values at PGSI = 3 (moderate risk) and PGSI = 8 (problem gambling).

Figure 2: Predicted probability of reporting severe consequences for people living in higher and lower income households across PGSI scores of 0 to 10

The graph presents results for Household income shown in Table 2.

Table 2: Predicted probability of reporting severe adverse consequences at PGSI = 3 (moderate risk) and PGSI = 8 (problem gambling), by demographic group

Table 2: Predicted probability of reporting severe adverse consequences at PGSI = 3 (moderate risk) and PGSI = 8 (problem gambling), by demographic group
Demographic group Probability at PGSI = 3 (percentage) Probability at PGSI = 8 (percentage)
Ethnicity
Ethnic minority 2.4% 9.0%
White 0.7% 3.0%
Household income quintile
Higher income (third quintile or above) 0.7% 3.0%
Lower income (second quintile or below) 1.5% 6.9%
Age
35 years and over 0.7% 3.0%
Under 35 years 1.2% 4.0%

4 Reference categories were female, age 35 years or older, higher household income (quintile 3 or above), White ethnicity, and has an educational qualification.

5 Some subgroup analyses, particularly for severe consequences (Table S2) and corresponding regression estimates (Table S4), are based on relatively small cell counts. These estimates should therefore be interpreted with caution, as reflected in wider confidence intervals.

Discussion

This report presents secondary analysis of 2025 Gambling Survey for Great Britain (GSGB) data to examine whether some people may be particularly susceptible to gambling-related harm at lower levels of gambling risk (as measured using the Problem Gambling Severity Index (PGSI)). After controlling for other demographic characteristics and PGSI scores, we found that participants from ethnic minority backgrounds, those living in lower-income households, and younger participants were more likely to report both potential and severe adverse consequences than their respective comparison groups. For potential adverse consequences, differences by ethnicity and age were most pronounced at lower PGSI scores and reduced as scores increased. For severe consequences, between-group differences were broadly similar across the PGSI scale.

These findings reflect an important distinction between at-risk gambling behaviour, which is captured by the PGSI, and the consequences that people experience. Although higher PGSI scores are generally associated with increased risk of harm, the strength of this association depends on people's wider social and financial circumstances. Relying on PGSI scores alone is likely to mask important differences in vulnerability across the population, and so using a range of markers of harm may help gambling operators identify and support people who are most at risk.

Our findings align with previous research showing that people on lower incomes and those from ethnic minority backgrounds experience harm at comparatively lower levels of gambling frequency and spend (Raybould, Larkin and Tunney, 2021 (opens in new tab); Wheaton, Collard and Nairn, 2024 (opens in new tab)). The increased risk of harm in these groups can be partly explained by differences in financial resilience, stigma, and barriers to accessing support. For example, people on lower incomes may have reduced capacity to absorb financial losses. There is also evidence that gambling is more stigmatised within some ethnic minority communities, which could contribute to harm and discourage people from seeking support (Wheaton, Collard and Nairn, 2024 (opens in new tab)).

Finally, results from this secondary analysis are consistent with the 'prevention paradox' of gambling-related harm, which states that, at a population level, a substantial proportion of harm arises among people who are not classified as high risk (Browne and Rockloff, 2018 (opens in new tab)). Our findings show that only a small proportion of people who scored 0 on the PGSI (categorised as 'non-problem' gambling) reported adverse consequences from gambling. However, because most people who gambled in the past 12 months scored 0 on the PGSI, this group may still account for a meaningful share of gambling-related harm overall. This highlights the need to ensure that harm prevention strategies reach the wider population of people who gamble, rather than focusing solely on those identified as higher risk.

Limitations

Several limitations should be considered when interpreting these findings. Firstly, the analysis draws on cross-sectional data and so we are unable to establish causal relationships between demographic characteristics, PGSI scores, and adverse consequences. Longitudinal research is needed to clarify the direction of these associations and the pathways through which harm develops over time. Secondly, due to low base sizes, participants from ethnic minority backgrounds were grouped within a single category, which is likely to have masked meaningful group differences in gambling behaviour and experiences of harm. We plan to address this in future work by pooling GSGB data across Years 1 to 3 to examine gambling behaviour and adverse consequences among individual ethnic groups.

Conclusion

This report contributes to Theme 3 of our evidence priority roadmaps, which focuses on understanding gambling-related harm and vulnerability (Gambling Commission, 2025b). We found that adverse consequences from gambling were most prevalent among younger participants, those from ethnic minority backgrounds, and people living in lower-income households, and that differences by age and ethnicity were most pronounced at lower Problem Gambling Severity Index (PGSI) scores. Taken together, these findings highlight the importance of monitoring adverse consequences alongside behavioural risk indicators, such as the PGSI, to ensure a complete understanding of gambling-related harm.

Supplementary information

Percentage of participants reporting potential adverse consequences, and severe consequences, by demographic group and Problem Gambling Severity Index (PGSI) category

Tables S1 and S2 show the percentage of participants who reported ‘one or more’ potential adverse consequences and any severe consequences from gambling6. Percentages are broken down by PGSI risk category (0, 1 to 2, 3 to 7, and 8 or more) and demographic group (ethnicity, household income, educational qualification, age, and sex).

Among those who scored 0 on the PGSI, potential adverse consequences were more prevalent among younger participants, males, people from ethnic minority backgrounds, those living in lower-income households, and among people who did not hold an educational qualification. Among people scoring 1 to 2 on the PGSI, similar differences were observed for ethnicity, income, and educational qualification. However, demographic differences in rates of potential adverse consequences were less pronounced among those scoring 3 to 7 and 8 or more on the PGSI (Table S1).

Across all PGSI categories, rates of severe consequences were higher among participants from ethnic minority backgrounds, compared with white participants. Participants in the highest PGSI risk category (PGSI 8 or more) were more likely to report severe consequences if they were living in lower income households, did not hold an educational qualification, and were male. Rates of severe consequences were similar for those aged under 35 and over 35 across all PGSI categories (Table S2).

6 Tables S1 and S2 are descriptive only; no statistical significance testing was conducted for these subgroup comparisons