Cookies on the Gambling Commission website

The Gambling Commission website uses cookies to make the site work better for you. Some of these cookies are essential to how the site functions and others are optional. Optional cookies help us remember your settings, measure your use of the site and personalise how we communicate with you. Any data collected is anonymised and we do not set optional cookies unless you consent.

Set cookie preferences

You've accepted all cookies. You can change your cookie settings at any time.

Skip to main content

Report

Gambling Survey for Great Britain Year 2 topic report: Investigating the profiles of those who gamble more frequently

This topic report uses data from Year 2 of the GSGB to explore the association between engagement in gambling activities, frequency of gambling, and risk.

  1. Contents
  2. Composition of the gambling frequency groups

Composition of the gambling frequency groups

A comparison was made of the socio-demographic profiles of each gambling frequency group. The full set of socio-demographic characteristics included in the analysis were: age, sex, ethnicity, tenure, qualifications, economic activity, household income, marital status, whether there were children present in the household, deprivation indicators, and region.

Methods

The first step was to run a series of tables showing the weighted profile of each socio-demographic characteristic for each of the four categories of gambling frequency group. These are shown in Tables 1 to 9.

A multinomial regression model was then used to investigate whether there were differences in the socio-demographic profiles of frequency groups. Like all regression models, these models regress a number of explanatory variables (in this case, the socio-demographic variables) onto an outcome variable (the four categories of frequency group). The resulting model provides insight into how each of the explanatory variables relates to the outcome, when all other explanatory variables are held constant. This means the model can be used to isolate the impact of specific explanatory variables. This is useful; quite often the relationship between two measures (for example, employment status and frequency of gambling) may in part be explained by underlying differences in a third variable (such as age). Such a model allows us to tease apart the relationships between frequency of gambling, age, and employment.

The multinomial model was run on weighted data. Any demographic characteristics not significantly associated with the frequency group was removed from the model, meaning the model only retained the characteristics that had the most significant relationship with frequency group. These were: age, sex, ethnicity, tenure, qualifications, marital status, whether there were children present in the household, deprivation indicators, and region. Household income and economic activity were not significantly related to frequency group once the other demographic factors had been considered and were removed from the model.

Multinomial models are used when the outcome variable has distinct categories. These models make no assumptions about the underlying order of the categories, hence are suitable in instances such as this where the outcome variable is not only split by frequency, but also by activity (the weekly gambling group is split into those who only buy tickets for lottery draws and those who also participate in all forms of gambling).

The output from a multinomial model is interpreted by comparing each of the outcome variable categories to a baseline group. The baseline group for the gambling frequency variable was the 12-month group (that is, those who had gambled on any activity, either online or in person, less frequently than the past 4 weeks but in the past 12 months). For each demographic characteristic, the profiles of the weekly gambling on any activity or activities in addition to lotteries, weekly lottery-only, and four-month groups are therefore compared to the 12-month group and tested to identify where there are significant differences in profile.

Results

The output from the multinomial model in the following section cross refers to information that can be found in the accompanying set of data tables, specifically table B.1. We summarise the weighted profile of each socio-demographic characteristic included in the model, before describing the direction of the relationships as confirmed by the model output.

Age

Table 1 shows how the weekly lottery-only group has an older age profile: 35 percent of this group were aged over 65 years whereas between 18 to 22 percent of the other groups were 65 and over. Similarly, the weekly lottery-only group has the lowest proportion of people in the younger age bracket. 8 percent of this group were aged 18 to 34 years, whereas between 29 to 34 percent of other age groups were aged between 18 to 34.

Table 1: Age profile by gambling frequency group

Table 1: Age profile by gambling frequency group
Frequency of activity and type (either online or in person) At least weekly on any activity in addition to lottery draws* (percentage) At least weekly but only on lottery draws* (percentage) Past 4 weeks but not in the past week (percentage) Past 12 months but not in the past 4 weeks (percentage) Total of everyone who gambled in past 12 months (percentage)
18 to 34 30% 8% 31% 34% 26%
35 to 64 52% 58% 51% 46% 52%
65 and over 18% 35% 18% 20% 22%
Number of total participants 1,780 2,707 4,661 2,407 11,555
* Lottery draws include both National Lottery and charity lottery draws

After controlling for other demographic characteristics, the weekly lottery-only group still has a notably older age profile than the other groups. Individuals within this group were significantly more likely to belong in an older age bracket; they were nearly six times more likely to be aged 65 and over than aged 18 to 34, and nearly three times more likely to be aged 35 to 64 years than aged 18 to 34, when compared with the baseline 12 month group. The other groups also had a higher likelihood of being aged 35 to 64 than the baseline group, however, the differences were smaller.

Gender

The group with the highest proportion of women was the 12 month group, 57 percent of this group were women, followed by the four-week group (54 percent), then the weekly lottery group (48 percent). The weekly gambling on any activity or activities in addition to lotteries group contains the lowest proportion of women (36 percent). Essentially, as gambling frequency increased, the proportion of women gambling decreased. That said, a third of participants who gambled weekly on activities in addition to lotteries were women. This is shown in Table 2.

Table 2: Gender profile by gambling frequency group

Table 2: Gender profile by gambling frequency group
Frequency of activity and type (either online or in person) At least weekly on any activity in addition to lottery draws* (percentage) At least weekly but only on lottery draws* (percentage) Past 4 weeks but not in the past week (percentage) Past 12 months but not in the past 4 weeks (percentage) Total of everyone who gambled in past 12 months (percentage)
Male (percentage) 64% 52% 46% 43% 50%
Female and/or non-binary (percentage) 36% 48% 54% 57% 50%
Number of total participants 1,780 2,707 4,661 2,407 11,555
* Lottery draws include both National Lottery and charity lottery draws

This gender split across groups was also seen in the model output. In all groups, participants were significantly more likely to be male, compared with the baseline group, reflecting the bivariate profile (whereby the 12 month group contained the highest proportion of women).

Ethnicity

The weekly lottery-only group contained the lowest proportion of people belonging to a non-white background at eight percent. The proportion for other groups ranged between 12 to 15 percent.

Table 3: Ethnic background by gambling frequency group

Table 3: Ethnic background by gambling frequency group
Frequency of activity and type (either online or in person) At least weekly on any activity in addition to lottery draws* (percentage) At least weekly but only on lottery draws* (percentage) Past 4 weeks but not in the past week (percentage) Past 12 months but not in the past 4 weeks (percentage) Total of everyone who gambled in past 12 months (percentage)
Black, Asian, Mixed, Other, and missing 14% 8% 12% 15% 12%
White 86% 92% 88% 85% 88%
Number of total participants 1,780 2,707 4,661 2,407 11,555
* Lottery draws include both National Lottery and charity lottery draws

The model indicates that, after controlling for the other demographic characteristics, participants within the weekly lottery-only and the 4 week groups were significantly more likely to be from a white ethnic background, compared with the baseline group. There were no significant differences between the baseline group and the weekly gambling on any activity or activities in addition to lotteries group.

Tenure

The weekly lottery-only group contained the highest proportion of people who own their homes outright at 39 percent versus 22 to 29 percent for other groups. People in the weekly gambling on any activity or activities in addition to lotteries group were more likely than those in other groups to be renting from a housing association (24 versus 12 percent in all other groups).

Table 4: Tenure by gambling frequency group

Table 4: Tenure by gambling frequency group
Frequency of activity and type (either online or in person) At least weekly on any activity in addition to lottery draws* (percentage) At least weekly but only on lottery draws* (percentage) Past 4 weeks but not in the past week (percentage) Past 12 months but not in the past 4 weeks (percentage) Total of everyone who gambled in past 12 months (percentage)
Own it outright 22% 39% 27% 29% 29%
Buying it with the help of a mortgage or loan 30% 36% 39% 35% 36%
Rent it from a housing association or local council 24% 12% 12% 12% 14%
Rent it from another type of landlord 18% 10% 18% 20% 17%
Live here rent-free 5% 2% 3% 4% 4%
Number of total participants 1,780 2,707 4,661 2,407 11,555
* Lottery draws include both National Lottery and charity lottery draws

After controlling for the other demographic characteristics, significant differences in the tenure profile of the groups remained in the model. The weekly gambling on any activity or activities in addition to lotteries group was significantly more likely to rent from a housing association or be living rent-free (individuals in this group were twice as likely to rent from a housing association or live rent free than they were to be an owner occupier, when compared to the tenure profile of the baseline 12-month group). Both the weekly lottery-only group and the four-week group members were more likely to be buying with a mortgage, compared with the baseline 12-month group.

Marital status

Individuals in the weekly lottery-only group were less likely to be single, (20 percent versus between 36 to 39 percent for other groups), and were more likely to be married (62 percent versus 44 to 49 percent for other groups).

Table 5: Marital status by gambling frequency group

Table 5: Marital status by gambling frequency group
Frequency of activity and type (either online or in person) At least weekly on any activity in addition to lottery draws* (percentage) At least weekly but only on lottery draws* (percentage) Past 4 weeks but not in the past week (percentage) Past 12 months but not in the past 4 weeks (percentage) Total of everyone who gambled in past 12 months (percentage)
Single 39% 20% 36% 38% 33%
Married or Civil Partnership 44% 62% 49% 49% 51%
Separated or Separated Civil Partnership 3% 2% 2% 2% 2%
Divorced or Dissolved Civil Partnership 9% 10% 8% 7% 8%
Widowed 4% 5% 4% 4% 4%
Number of total participants 1,780 2,707 4,661 2,407 11,555
* Lottery draws include both National Lottery and charity lottery draws

The model shows there were significant differences in marital status, and that these were present despite controlling for other socio-demographic characteristics. Those in the weekly gambling on any activity or activities in addition to lotteries group were significantly less likely to be married than the baseline group, whilst those in the weekly lottery-only group were significantly more likely to be married than the baseline. There were no significant differences in marriage rates between the four-week and 12-month group.

Household type

The weekly lottery-only group was less likely to contain people living in households with children (25 percent versus 31 to 32 percent for other groups). The weekly gambling on any activity/activities in addition to lotteries group contained more people living in multi-adult households (55 percent versus 40 to 50 percent for other groups).

Table 6: Household composition by gambling frequency group

Table 6: Household composition by gambling frequency group
Frequency of activity and type (either online or in person) At least weekly on any activity in addition to lottery draws* (percentage) At least weekly but only on lottery draws* (percentage) Past 4 weeks but not in the past week (percentage) Past 12 months but not in the past 4 weeks (percentage) Total of everyone who gambled in past 12 months (percentage)
All households with children 32% 25% 32% 31% 30%
All single adult households 15% 15% 14% 14% 14%
All 2 adult households 28% 43% 35% 35% 36%
All households with multiple (2 or more) adults 55% 40% 50% 49% 49%
Number of total participants 1,780 2,707 4,661 2,407 11,555
* Lottery draws include both National Lottery and charity lottery draws

Within the multi-variate model, presence of children in the household was significantly related to gambling frequency group. Both the weekly gambling on any activity or activities in addition to lotteries and weekly lottery-only groups were less likely to have children present in the household when compared with the 12-month group.

Qualifications

There appears to be a strong relationship between gambling frequency and qualifications. People in the weekly gambling on any activity or activities in addition to lotteries group were less likely to have any qualifications, with 29 percent of this group saying they had no qualifications, compared with 20 percent in the weekly lottery-only group and 16 to 17 percent in the four-week and 12-month groups. This pattern is also present when looking at the proportion of participants with a degree and/or higher degree whereby those who gambled weekly, and gambled on things other than lottery draws alone, were less likely to have a degree than those in other groups.

Table 7: Highest qualification by gambling frequency group

Table 7: Highest qualification by gambling frequency group
Frequency of activity and type (either online or in person) At least weekly on any activity in addition to lottery draws* (percentage) At least weekly but only on lottery draws* (percentage) Past 4 weeks but not in the past week (percentage) Past 12 months but not in the past 4 weeks (percentage) Total of everyone who gambled in past 12 months (percentage)
None or Missing 29% 21% 16% 17% 19%
Below degree level 39% 42% 39% 34% 39%
Degree level or above 32% 37% 45% 49% 42%
Number of total participants 1,780 2,707 4,661 2,407 11,555
* Lottery draws include both National Lottery and charity lottery draws

These differences were also apparent in the model output. Individuals in the 12-month group were most likely to hold a degree-level qualification, they were twice as likely to have a degree than those in the weekly gambling on any activity or activities in addition to lotteries group. Those in the weekly gambling on any activity or activities in addition to lotteries group were significantly less likely to have any qualifications (either a degree or other qualifications below degree level) than those within the 12-month group. Individuals in the weekly lottery-only group were also significantly less likely to hold a degree-level qualification than those in the 12-month group, although there were no differences in the likelihood of holding other non-degree qualifications between the weekly lottery-only group and the 12-month group. The four-week group were significantly more likely to hold non-degree qualification than the 12-month group.

Index of Multiple Deprivation2

The weekly gambling on any activity or activities in addition to lotteries group contained the highest proportion of people living in deprived local areas, with 27 percent of this group living in the most deprived 20 percent of local areas, compared with 16 to 18 percent in the other groups.

Table 8: Deprivation quintile by gambling frequency group

Table 8: Deprivation quintile by gambling frequency group
Frequency of activity and type (either online or in person) At least weekly on any activity in addition to lottery draws* (percentage) At least weekly but only on lottery draws* (percentage) Past 4 weeks but not in the past week (percentage) Past 12 months but not in the past 4 weeks (percentage) Total of everyone who gambled in past 12 months (percentage)
20% most deprived 27% 16% 18% 17% 19%
2nd 23% 20% 21% 21% 21%
3rd 20% 22% 21% 22% 21%
4th 16% 22% 21% 20% 20%
20% least deprived 14% 21% 19% 21% 19%
Number of total participants 1,780 2,707 4,661 2,407 11,555
* Lottery draws include both National Lottery and charity lottery draws

The same pattern was apparent in the model output. The weekly gambling on any activity or activities in addition to lotteries group contained a significantly higher proportion of people living in the most deprived areas (top 20 percent), even after controlling for regional differences in the profiles of the groups. Individuals in the weekly gambling on any activity or activities in addition to lotteries group were twice as likely to be in the most deprived quintile than they were in the least deprived quintile, compared with the individuals in the 12-month group. The deprivation profiles of the weekly lottery-only, four-week and 12-month groups were not significantly different.

Region and/or Nation

Those in the weekly gambling on any activity or activities in addition to lotteries group were more likely to be from Scotland (11 percent versus eight percent in all other groups) or the North West (14 percent versus 10 to12 percent across the other groups), and less likely to be from the South East (11 percent versus 14 to 16 percent for other groups).

Table 9: Region by gambling frequency group

Table 9: Region by gambling frequency group
Frequency of activity and type (either online or in person) At least weekly on any activity in addition to lottery draws* (percentage) At least weekly but only on lottery draws* (percentage) Past 4 weeks but not in the past week (percentage) Past 12 months but not in the past 4 weeks (percentage) Total of everyone who gambled in past 12 months (percentage)
East Midlands 8% 9% 8% 7% 8%
East of England 9% 10% 10% 10% 10%
London 10% 9% 12% 14% 11%
North East 5% 5% 5% 4% 5%
North West 14% 12% 12% 10% 12%
Scotland 11% 8% 8% 8% 9%
South East 11% 14% 15% 16% 14%
South West 8% 10% 9% 9% 9%
Wales 5% 6% 5% 4% 5%
West Midlands 10% 9% 8% 10% 9%
Yorkshire and The Humber 10% 10% 9% 8% 9%
Number of total participants 1,780 2,707 4,661 2,407 11,555
* Lottery draws include both National Lottery and charity lottery draws

The model also identified regional differences across groups. The weekly gambling on any activity or activities in addition to lotteries group, weekly lottery-only group and the four-week group were all significantly less likely to contain people from London compared with the baseline 12-month group. The weekly gambling on any activity or activities in addition to lotteries group and weekly lottery-only group were additionally less likely than the 12-month group to contain people living in the South East, and both the weekly lottery-only group and four-week group were additionally less likely than the 12-month group to contain people living in the West Midlands.

Summary of differences between frequency groups

Examining the demographic and socio-economic profile of people who gamble according to gambling frequency highlights some marked differences between these groups. First, those who gamble weekly on any activity in addition to lotteries have a distinct profile. This group is significantly different to all other groups in terms of area deprivation (people in this group are more likely to live in more deprived areas), having a greater proportion of people with a lack of educational qualifications, and higher likelihood of living in accommodation rented from a housing association. Each of these factors is significantly related to group membership, even when the other factors are taken into account (for example, even if we were to remove any differences that exist between groups in terms of educational qualifications or tenure, people who gamble weekly on any activity will still be more likely to live in areas of higher deprivation). This suggests these factors have an additive impact. In short, this group appears to have greater socio-economic disadvantage than those in the baseline 12-month group.

Second, when looking at those who gambled weekly but only on lotteries, differences between this group and others were driven by age, the number of children in the household, and ethnicity. People in this group tended to be older, from White backgrounds and to have fewer children living in the household. This group were also more likely to be married. Notably, this group did not vary from the baseline group according to most socio-economic features. The main difference between these two groups of people who gamble weekly is whether they do something other than lotteries alone. Those who gamble weekly on any activities have a profile consistent with being more socially and economically disadvantaged than people who gamble less frequently on any activities, or those who gamble at the same frequency but only on lotteries.

Variables removed from the model

The following section cross refers to information that can be found in the accompanying set of data tables, specifically table C.1.

Economic occupation and household income had no significant relationship with frequency group once the other socio-economic measures had been controlled for and were removed from the model. Whilst there are some occupational differences seen across frequency groups, these are likely being driven by the differences in age profile across groups – as the key difference was the noticeably higher proportion of retired people in the weekly lottery-only group. Occupation therefore dropped out of the model once the differences in age profile were controlled for. Household income is correlated with both area level deprivation and qualifications, these two measures had stronger relationships with group membership than income, meaning income dropped out of the model also.

Problem Gambling Severity Index

Table 10 shows the difference in PGSI scores by frequency group, both grouped score and mean score. These, as expected, indicate that the weekly gambling on any activity or activities in addition to lotteries group is far more likely to have a PGSI score of eight or more. This is followed by the four-week group (but there is quite a gap between the two). The weekly lottery-only and 12-month groups are very similar in terms of PGSI profile.

Table 10: PGSI score by gambling frequency group

Table 10: PGSI score by gambling frequency group
Frequency of activity and type (either online or in person) At least weekly on any activity in addition to lottery draws* (percentage) At least weekly but only on lottery draws* (percentage) Past 4 weeks but not in the past week (percentage) Past 12 months but not in the past 4 weeks (percentage) Total of everyone who gambled in past 12 months (percentage)
PGSI score 0 46% 84% 79% 85% 75%
PGSI score 1 to 2 23% 11% 15% 11% 15%
PGSI score 3 to 7 14% 3% 4% 2% 5%
PGSI score 8 or more 17% 1% 2% 2% 4%
Mean score 3.26 0.42 0.61 0.49 1.01
Standard deviation 5.34 1.58 2.04 2.11 3.04
Number of total participants 1,780 2,707 4,661 2,407 11,555
* Lottery draws include both National Lottery and charity lottery draws

The differences in PGSI scores of people who gamble on a weekly basis is explored further in the next section by looking in turn at the associations between weekly participation in each activity and PGSI score.

References

2 The Index of Multiple Deprivation (IMD) is a measure of relative deprivation applied to small geographic areas, that ranks them according to levels of local deprivation. Each of the four constituent nations of the United Kingdom creates its own index. Whilst there are small differences in the approach used to measure deprivation in each nation, broad themes include income, employment, health, education, crime, barriers to housing and services, and the lived environment.

Previous section
Investigating the profiles of those who gamble more frequently - Gambling frequency groups
Next section
Investigating the profiles of those who gamble more frequently - Analysis of weekly activities and PGSI scores
Is this page useful?
Back to top