Equality across different areas of life in the UK: 2011 to 2020

• We have taken a new approach to analysing equality to enable comparison of the outcomes of different population groups across multiple areas of life (crime, mortality, life satisfaction, digital exclusion and wealth).

• By introducing explanatory factors in stages, this approach provides more nuanced insights into which factors affect the relationship between the population group and the outcome.

• Across the five areas of life in this article, for most groups, there was no consistent pattern in terms of which group was likely to experience a poorer outcome and the extent to which explanatory factors affected the outcomes; this was the case when considering age, sex, ethnic group, and region.

• However, this was not the case for disabled people who were more likely to experience poorer outcomes than non-disabled people across all the outcome areas even after adjusting for a range of socio-demographic factors.

• Similarly, people who identify as bisexual experienced poorer outcomes compared with those who identify as heterosexual or straight across three areas of life and this was unaffected by adjusting for explanatory factors; data on sexual orientation were only available for crime, life satisfaction and wealth.

• While aiming to be as consistent as possible to enable comparison across outcomes, challenges included differences in the data sources used, geographical coverage and time periods of these data sources, and the information available within them, highlighting the need for better data for analysing equalities.

This release is part of the Equality Data Programme (EDP) announced by the Cabinet Office in 2020. Its purpose is to improve the quality of data and evidence within government about the inequalities that different groups may face.

The initial aim of this programme of work was to explore the characteristics and circumstances associated with better or worse outcomes, in relation to:

• crime

• wealth

• mortality

• life satisfaction

• digital exclusion

We used linear and binary logistic regression to investigate the relationship between specific characteristics of interest (age, disability, ethnicity, region, sex, and sexual orientation) and each different life outcome.

To identify groups more likely to experience poorer outcomes across multiple areas of life, we took a consistent approach wherever possible across topic areas and data sources, in relation to methodology, reference groups and analytical breakdowns used. With the latter, consistency was sometimes sacrificed in favour of greater granularity in the breakdowns presented.

The choice of characteristics and circumstances to include in our models was guided by published literature and topic expertise, though was constrained by the availability of relevant information within each data source. Please refer to the datasets published alongside this release for details of the factors included at each stage of the modelling and further information.

We welcome comments on the benefits of this approach to exploring equality and any suggested improvements.

Stages of regression models

For each characteristic of interest, we ran regression models in stages. The first stage was to fit an unadjusted model, to provide an estimate of the overall relationship between each characteristic and each outcome. Including only the characteristic and outcome allowed us to observe the relationship between them without taking any other characteristics or circumstances into account.

At the second stage, referred to as the basic model, we adjusted for confounders of our characteristic of interest. Confounders are factors thought to affect both the characteristic and outcome of interest. For example, an individual’s age is likely to affect both their disability status as well as their likelihood of mortality. The basic model allows us to see whether there is any statistically significant difference in the outcome between different population groups, after adjusting for confounding factors.

In the final stage, referred to as the fully adjusted model, we adjusted for possible explanatory factors (mediators) that may affect the relationship between the characteristic and the outcome. Where adjusting for these additional factors reduces any differences in the likelihood of experiencing the outcome seen in the basic model, this indicates that these factors are associated with some of these differences. Where these factors have no effect on the differences seen in the basic model (that is, the relationship is not affected by mediating factors), this indicates that there may be other factors not included in the model that are associated with the differences. There could also be a direct relationship between the characteristic and the outcome.

Data sources

We have used a range of different data sources, deemed to be the best available to explore the outcomes of interest. While trying to be as consistent as possible across the analysis, differences between data sources in terms of geographical coverage, time periods covered and availability of relevant characteristic information, has resulted in some unavoidable inconsistencies. In particular, the analysis of life satisfaction uses data covering January to December 2020 when a range of coronavirus (COVID-19) restrictions were in place.

Other outcomes are explored using data up to the end of March 2020, before most coronavirus restrictions were in place, or using linked data from the 2011 Census. The patterns observed should be considered in this context.

This analysis illustrates one method of comparing outcomes for different population groups across multiple areas of life to provide a more holistic picture. We hope that providing more insight into specific factors associated with more or less advantageous outcomes between groups will provide valuable information for policy and practice. We see this as an initial first step, with further research needed to identify the specific explanatory factors contributing most to particular life outcomes for different groups.

Definitions for each outcome metric are included in the notes to figures and in Section 10: Glossary.

Across the outcomes included in this analysis, the findings in relation to age generally showed that, relative to those aged 30 to 34 years, people in older age groups:

• had higher odds of dying (Figure 1)

• were more likely to be digitally excluded (Figure 2)

• had more wealth on average (Figure 3)

After adjusting for a range of factors, those in the oldest age groups (70 years and over) were no more or less likely to report low life satisfaction than those aged 30 to 34 years (Figure 4). Those in older age groups were less likely to be victims of crime relative to those aged 25 to 34 years (Figure 5).

Analysis of the Public Health Data Asset showed that in the year after the 2011 Census in England and Wales, the likelihood of mortality generally increased with increasing age as would be expected (for the definition used to measure mortality see Section 10: Glossary). However, while they were still more likely to die, the odds of mortality were lower for all those over the age of 34 years compared with those aged 30 to 34 years after adjusting for:

• sex

• country of birth

• disability status

• ethnicity

• English language proficiency

• housing tenure

• Index of Multiple Deprivation (IMD) decile

• marital status

• region

• religion

• socio-economic status

• urban-rural status

This demonstrates that while age is important in the likelihood of mortality, other factors are also important.

Figure 1: Those aged 35 years and over had higher odds of mortality compared with those aged 30 to 34 years, but this relationship is affected by other socio-demographic characteristics

Odds ratios of mortality by age group in England and Wales compared with 30- to 34-year-olds, 27 March 2011 to 26 March 2012

Source: Office for National Statistics - Public Health Data Asset 

Notes:

1. 95% confidence intervals are displayed on the chart, see the Glossary for details.
2. Basic models are adjusted for sex. Fully adjusted models are adjusted for sex, country of birth, disability status, English language proficiency, ethnicity, housing tenure, Index of Multiple Deprivation decile, marital status, region, religion, socio-economic status and urban-rural status.
3. An individual on the Public Health Data Asset was deemed to have died if they were recorded on the 2011 Census and subsequently linked to a death registration record with a date of death within a year of the 2011 Census (between 27 March 2011 and 26 March 2012).

Download this image Figure 1: Those aged 35 years and over had higher odds of mortality compared with those aged 30 to 34 years, but this relationship is affected by other socio-demographic characteristics

.svg (5.2 MB) .xlsx (20.8 kB)

We also considered how age is related to digital exclusion, defined as people who had not used the internet within the last three months or had never used it.

Using data from January to March 2020 across the UK, we found adjusting for a range of factors made little difference to the higher likelihood of being digitally excluded among those aged 45 years and over, compared with those aged 25 to 34 years. This suggests that age may be an important factor in explaining digital exclusion. However, there may be other factors at play that were not included in the modelling.

Additionally, there may be a generational effect where the current generation of older people may be less likely to use the internet than future generations who have grown up with it.

Figure 2: The odds of being digitally excluded compared with those aged 25 to 34 years increased with age after fully adjusting for a range of factors

Odds ratios of adults who were digitally excluded (adults who have not used the internet within the last three months)​ by age in the UK, January to March 2020 

Source: Office for National Statistics - Labour Force Survey

Notes:

1. 95% confidence intervals are displayed on the chart, see the Glossary for details.
2. Basic models are adjusted for sex only. Fully adjusted models are adjusted for sex, disability status, educational attainment, ethnicity, household composition, housing tenure, marital status, region and socio-economic status.
3. In this analysis anyone that had not used the internet within the last three months (including those who had never used the internet) was defined as digitally excluded.

Download this image Figure 2: The odds of being digitally excluded compared with those aged 25 to 34 years increased with age after fully adjusting for a range of factors

.svg (2.4 MB) .xlsx (19.5 kB)

A similar pattern also emerged when exploring how a range of factors affected the relationship between age and individual wealth. Using data for the period from April 2018 to March 2020 in Great Britain, we examined differences in average total individual wealth in a fully adjusted model, which adjusted for:

• sex

• disability status

• economic activity status

• educational attainment

• ethnicity

• household composition

• housing tenure

• industry type

• sexual orientation

• socio-economic status

We found these factors had little effect on the difference in average wealth of those aged 35 years and over compared with those aged 30 to 34 years. This suggests that for these age groups, age is likely to be a factor in levels of wealth, reflecting a general pattern of accumulating wealth through a person’s working life, which gradually reduces through retirement.

Fully adjusting for these characteristics did, however, affect age groups aged below 30 years. Before adjustment, these age groups had significantly less wealth on average than those aged 30 to 34 years, and this was largely unaffected by adjusting for sex.

However, after adjusting for additional factors, adults aged under 25 years had more wealth on average than those aged 30 to 34 years. Those aged 25 to 29 years had a similar level of wealth to those aged 30 to 34 years. The findings in relation to adults aged under 25 years are interesting and may benefit from further exploration.

Figure 3: After fully adjusting for a range of factors, all age groups other than those aged 25 to 29 years had higher average (mean) total individual wealth than those aged 30 to 34 years

Difference in average (mean) total individual wealth by age in Great Britain, April 2018 to March 2020

Source: Office for National Statistics - Wealth and Assets Survey

Notes:

1. Only people aged 19 years and over or aged 16 to 18 years and not in full-time education have been included in the analysis.
2. 95% confidence intervals are displayed on the chart, see the Glossary for details.
3. Basic models are adjusted for sex only. Fully adjusted models are adjusted for sex, disability status, economic activity status, educational attainment, ethnicity, household composition, housing tenure, industry type, sexual orientation and socio-economic status.
4. In this analysis, we use the mean as a measure of average wealth. Mean wealth is the total wealth of all individuals divided by the number of individuals. Total wealth is the sum of net wealth in property, private pensions, financial and physical items.

Download this image Figure 3: After fully adjusting for a range of factors, all age groups other than those aged 25 to 29 years had higher average (mean) total individual wealth than those aged 30 to 34 years

.svg (4.5 MB) .xlsx (20.6 kB)

In 2020 across the UK, the likelihood of reporting low life satisfaction differed by age group (see Section 10: Glossary for the definition of low life satisfaction). The effect of adjusting for various factors on the relationship between age group and low life satisfaction also varied. These factors were:

• sex

• disability status

• educational attainment

• employment status

• ethnicity

• housing tenure

• marital status

• region

• sexual orientation

• socio-economic status

• type of interview at survey

After adjusting for these factors, the estimated odds of reporting low life satisfaction were higher for those aged 35 to 64 years compared with those aged 30 to 34 years.

This suggests that other factors may be important in understanding these differences.

In contrast, after adjusting for these factors, the estimated odds of reporting low life satisfaction were lower for those aged 65 to 69 years when compared with those aged 30 to 34 years. They were comparable for those aged 70 years and over. This indicates that these factors affect the likelihood of reporting low life satisfaction among these older age groups.

Figure 4: After fully adjusting for a range of factors, adults aged 35 to 64 years were more likely to report low life satisfaction than those aged 30 to 34 years 

Odds ratios of adults who reported low life satisfaction by age in the UK compared with 30-to-34-year-olds, 2020

Source: Office for National Statistics - Annual Population Survey

Notes:

1. 95% confidence intervals are displayed on the chart, see the Glossary for details.
2. Basic models are adjusted for sex only. Fully adjusted models are adjusted for sex, disability status, educational attainment, employment status, ethnicity, housing tenure, marital status, region, sexual orientation, socio-economic status and type of interview conducted at survey.
3. In this analysis, low life satisfaction was defined as a score of 0 to 4 inclusive, on a scale of 0 to 10, where 0 is “not at all” and 10 is “completely”, in response to how satisfied people are with their life overall.

Download this image Figure 4: After fully adjusting for a range of factors, adults aged 35 to 64 years were more likely to report low life satisfaction than those aged 30 to 34 years 

.svg (4.4 MB) .xlsx (20.3 kB)

Across England and Wales in the year ending March 2020, those aged 44 years and over were significantly less likely to be a victim of crime compared with those aged 25 to 34 years, before taking any other factors into account. Victim of crime excludes fraud, sexual offences and computer misuse, see Section 10: Glossary for the definition of a victim of crime. These patterns persisted after adjusting for sex. The patterns seen here may have been different if other types of crime, for example, sexual offences had been included in the analysis.

After also adjusting for explanatory factors in the fully adjusted model, only those aged 65 years and over still had a lower likelihood of being a victim of crime compared with those aged 25 to 34 years. The factors in the fully adjusted model were:

• sex

• disability status

• educational attainment

• employment status

• ethnicity

• household composition

• household income

• housing tenure

• Index of Multiple Deprivation (IMD) decile

• marital status

• region

• sexual orientation

• socio-economic status

• urban-rural status

Adjusting for these factors resulted in those aged 45 to 64 years being as likely to be a victim of crime as those aged 25 to 34 years. Whereas, for those aged 16 to 24 years, adjusting for these factors resulted in their having a higher likelihood of experiencing crime compared with those aged 25 to 34 years.

Figure 5: Those aged 35 to 44 years were as likely to be a victim of crime as those aged 25 to 34 years, before and after adjusting for a range of factors

Odds ratios of adults who were victims of crime at least once within the last 12 months (excludes fraud, sexual offences and computer misuse)​ by age in England and Wales, year ending March 2020

Source: Office for National Statistics - Crime Survey for England and Wales

Notes:

1. 95% confidence intervals are displayed on the chart, see the Glossary for details.
2. Basic models are adjusted for sex. Fully adjusted models adjust for sex, disability status, educational attainment, employment status, ethnicity, household composition, household income, Index of Multiple Deprivation decile, marital status, region, sexual orientation, socio-economic status, tenure and urban-rural status.
3. A victim of crime is someone who had experienced at least one personal crime or had been a resident in a household that was a victim of at least one household crime in the last 12 months, irrespective of whether they reported these incidents to the police (excludes fraud, sexual offences and computer misuse).

Download this image Figure 5: Those aged 35 to 44 years were as likely to be a victim of crime as those aged 25 to 34 years, before and after adjusting for a range of factors

.svg (1.9 MB) .xlsx (19.2 kB)

Across the areas of life covered in this analysis, there was no consistent pattern between males and females in terms of which group was likely to experience a poorer outcome.

Considering the estimated odds of being a victim of crime, in the year ending March 2020 there was no difference between men and women, either before or after adjusting for a range of factors. The measure used for victims of crime does not cover fraud, computer misuse or sexual offences. If these had been included, it may have revealed different patterns. For more information on this please refer to Section 10: Glossary.

Before considering any other factors, men were less likely to be digitally excluded than women between January and March 2020. Men were also less likely than women to report low life satisfaction during 2020. However, after adjustment, there was no difference between men and women in the likelihood of experiencing these outcomes. This suggests that the factors included play an important part in shaping men’s and women’s outcomes in these areas of life (see the accompanying datasets for the full details of the factors included.)

Between April 2018 and March 2020 in Great Britain, men had significantly more wealth on average than women both before and after adjusting for other factors. This indicates that the relationship between sex and wealth is not affected by the factors included in our analysis and other factors not included in this analysis may be affecting this relationship.

Figure 6: On average, men had more total individual wealth than women, even after adjusting for a range of factors 

Difference in average (mean) total individual wealth by sex in Great Britain, April 2018 to March 2020

Source: Office for National Statistics - Wealth and Assets Survey

Notes:

1. Only people aged 19 years and over or aged 16 to 18 years and not in full-time education have been included in the analysis.
2. 95% confidence intervals are displayed on the chart, see the Glossary for details.
3. Basic models are adjusted for age only. Fully adjusted models are adjusted for age, disability status, economic activity status, educational attainment, ethnicity, household composition, housing tenure, industry type, sexual orientation and socio-economic status.
4. In this analysis, we use the mean as a measure of average wealth. Mean wealth is the total wealth of all individuals divided by the number of individuals. Total wealth is the sum of net wealth in property, private pensions, financial and physical items.

Download this image Figure 6: On average, men had more total individual wealth than women, even after adjusting for a range of factors 

.svg (1.1 MB) .xlsx (18.8 kB)

For the year ending 26 March 2012 in England and Wales, males had a lower likelihood of dying compared with females before adjusting for any other factors. After adjusting for age, males were more likely to die than females. The estimated odds that a male will die compared with a female were higher again after adjusting for:

• age

• country of birth

• disability status

• ethnicity

• English language proficiency

• housing tenure

• Index of Multiple Deprivation (IMD) decile

• marital status

• region

• religion

• socio-economic status

• urban-rural status

This suggests that these factors are not accounting for the differences between the groups and that sex, or other factors not included in the analysis, are affecting this relationship.

Figure 7: After adjusting for age, males had a higher likelihood of death compared with females

Odds ratios of mortality by sex compared with females in England and Wales, 27 March 2011 to 26 March 2012

Source: Office for National Statistics - Public Health Data Asset

Notes:

1. 95% confidence intervals are displayed on the chart, see the Glossary for details.
2. Basic models are adjusted for age. Fully adjusted models are adjusted for age, country of birth, disability status, English language proficiency, ethnicity, housing tenure, Index of Multiple Deprivation decile, marital status, region, religion, socio-economic status and urban-rural status.
3. An individual on the Public Health Data Asset was deemed to have died if they were recorded on the 2011 Census and subsequently linked to a death registration record with a date of death within a year of the 2011 Census (between 27 March 2011 and 26 March 2012).

Download this image Figure 7: After adjusting for age, males had a higher likelihood of death compared with females

.svg (820.3 kB) .xlsx (18.8 kB)

Across all the life outcomes analysed in this research, disabled people were more likely to experience poorer outcomes than non-disabled people. They were more likely to:

• be digitally excluded (Figure 8)

• report low life satisfaction (Figure 10)

• be victims of crime (Figure 11)

Overall, they were also more likely to die (Figure 9). They also had lower average wealth than non-disabled people (Figure 12). Explanatory factors were shown to contribute to the differences seen but other factors not included in this analysis also appear to be important.

The higher likelihood that disabled people will experience digital exclusion compared with non-disabled people, was lower after adjusting for age and sex (basic model) (Figure 8). This was also the case when considering mortality (Figure 9). In the mortality analysis, we looked separately at outcomes for disabled people according to whether they said their activities were limited a little or a lot.

The increased likelihood of digital exclusion among disabled people was also lower again after adjusting for the following factors:

• age

• sex

• educational attainment

• household composition

• housing tenure

• marital status

• region

• socio-economic status

Similarly, the increased likelihood of mortality among both groups of disabled people was lower again after adjusting for:

• age

• sex

• housing tenure

• Index of Multiple Deprivation (IMD) decile

• marital status

• region

• religion

• socio-economic status

• urban-rural status

However, in all cases, disabled people were still more likely to experience poorer outcomes than non-disabled people.

This suggests that the factors included in this analysis are associated with the differences between disabled and non-disabled people in these outcomes. However, there may also be other factors that are important to these outcomes that were not included in this analysis.

Figure 8: The odds of being digitally excluded were higher for disabled adults than non-disabled adults even after adjusting for a range of factors

Odds ratios of adults who were digitally excluded (adults who have not used the internet within the last three months) by disability​ status in the UK, January to March 2020

Source: Office for National Statistics - Labour Force Survey

Notes:

1. 95% confidence intervals are displayed on the chart, see the Glossary for details.
2. Basic models are adjusted for age and sex. Fully adjusted models are adjusted for age, sex, educational attainment, household composition, housing tenure, marital status, region and socio-economic status.
3. In this analysis anyone that had not used the internet within the last three months (including those who had never used the internet) were defined as digitally excluded.

Download this image Figure 8: The odds of being digitally excluded were higher for disabled adults than non-disabled adults even after adjusting for a range of factors

.svg (949.8 kB) .xlsx (18.8 kB)

Figure 9: The odds of mortality were higher for disabled people than non-disabled people, even after adjusting for other factors

Odds ratios of mortality by disability status compared with non-disabled people in England and Wales, 27 March 2011 to 26 March 2012

Source: Office for National Statistics - Public Health Data Asset

Notes:

1. 95% confidence intervals are displayed on the chart, see the Glossary for details.
2. Basic models are adjusted for age and sex only. Fully adjusted models are adjusted for age, sex, housing tenure, Index of Multiple Deprivation decile, marital status, region, religion, socio-economic status and urban-rural status.
3. An individual on the Public Health Data Asset was deemed to have died if they were recorded on the 2011 Census and subsequently linked to a death registration record with a date of death within a year of the 2011 Census (between 27 March 2011 and 26 March 2012).
4. Disability was defined according to the 2011 Census question; "Are your day-to-day activities limited because of a health problem or disability which has lasted, or is expected to last, at least 12 months? - Include problems related to old age" (Yes, limited a lot; Yes, limited a little; and No). 

Download this image Figure 9: The odds of mortality were higher for disabled people than non-disabled people, even after adjusting for other factors

.svg (1.3 MB) .xlsx (19.0 kB)

A similar pattern is also evident when looking at people reporting low life satisfaction in 2020 across the UK. The inclusion of explanatory factors resulted in lower estimated odds of disabled people reporting low life satisfaction compared with non-disabled people. However, they were still more likely to report it than non-disabled people. This again suggests that these other factors play an important part in disabled people’s life satisfaction, but that it may also be affected by other factors.

Figure 10: The odds of reporting low life satisfaction were higher for disabled adults than non-disabled adults

Odds ratios of adults who reported low life satisfaction by disability status compared with non-disabled adults in the UK, 2020

Source: Office for National Statistics - Annual Population Survey

Notes:

1. 95% confidence intervals are displayed on the chart, see the Glossary for details.
2. Basic models are adjusted for age and sex only. Fully adjusted models are adjusted for age, sex, educational attainment, employment status, housing tenure, marital status, region, socio-economic status and type of interview conducted at survey.
3. In this analysis, low life satisfaction was defined as a score of 0 to 4 inclusive, on a scale of 0 to 10, where 0 is “not at all” and 10 is “completely”, in response to how satisfied people are with their life overall.

Download this image Figure 10: The odds of reporting low life satisfaction were higher for disabled adults than non-disabled adults

.svg (946.5 kB) .xlsx (18.8 kB)

A different pattern was seen when looking at victims of crime in England and Wales in the year ending March 2020. Before adjusting for any other factors, disabled people were more likely to be victims of crime compared with non-disabled people.

The estimated odds of being a victim of crime were higher for disabled people than non-disabled people after adjusting for age and sex in the basic model and this remained the case in the fully adjusted model. This suggests that it is important to consider age and sex when exploring differences in victimisation between disabled and non-disabled adults.

Figure 11: Disabled adults were more likely to be a victim of crime than non-disabled adults, before and after adjusting for a range of factors 

Odds ratios of adults who were a victim of crime at least once within the last 12 months (excludes fraud, sexual offences and computer misuse)​ by disability status in England and Wales, year ending March 2020

Source: Office for National Statistics - Crime Survey for England and Wales

Notes:

1. 95% confidence intervals are displayed on the chart, see the Glossary for details.
2. Basic models are adjusted for age and sex. Fully adjusted models adjust for age, sex, educational attainment, employment status, household composition, household income, housing tenure, Index of Multiple Deprivation decile, marital status, region, socio-economic status and urban-rural status.
3. A victim of crime is someone who had experienced at least one personal crime or had been a resident in a household that was a victim of at least one household crime in the last 12 months, irrespective of whether they reported these incidents to the police (excludes fraud, sexual offences and computer misuse).

Download this image Figure 11: Disabled adults were more likely to be a victim of crime than non-disabled adults, before and after adjusting for a range of factors 

.svg (913.5 kB) .xlsx (18.8 kB)

Between April 2018 and March 2020, disabled people in Great Britain had significantly less average (mean) wealth than non-disabled people before adjustment.

After adjusting for age and sex, the difference in average wealth between disabled and non-disabled people was higher, indicating that for any given age and sex, disabled people on average had less wealth than non-disabled people. Compared with the basic model, the difference in average wealth is lower in the fully adjusted model, which adjusts for:

• age

• sex

• economic activity status

• educational attainment

• household composition

• housing tenure

• industry type

• socio-economic status

This suggests that these other factors also affect the relationship between average wealth and disability.

Figure 12: Disabled people had less wealth on average than non-disabled people even after controlling for a range of factors

Difference in average (mean) total individual wealth by disability status in Great Britain, April 2018 to March 2020

Source: Office for National Statistics - Wealth and Assets Survey

Notes:

1. Only people aged 19 years and over or aged 16 to 18 years and not in full-time education have been included in the analysis.
2. 95% confidence intervals are displayed on the chart, see the Glossary for details.
3. Basic models are adjusted for age and sex only. Fully adjusted models are adjusted for age, sex, economic activity status, educational attainment, household composition, housing tenure, industry type and socio-economic status.
4. In this analysis, we use the mean as a measure of average wealth. Mean wealth is the total wealth of all individuals divided by the number of individuals. Total wealth is the sum of net wealth in property, private pensions, financial and physical items.

Download this image Figure 12: Disabled people had less wealth on average than non-disabled people even after controlling for a range of factors

.svg (1.0 MB) .xlsx (18.7 kB)

We have ensured that we are as consistent as possible across our analysis. However, we used different data sources to capture specific life outcomes. This means that the level of detail we can present for ethnic group breakdowns varies by source. To provide the greatest insight and most detailed breakdowns possible for each source, the consistency across the analysis has been affected. The results should be considered in this context.

Across the different life outcomes included in this analysis, we found no consistent pattern in the experiences of different ethnic groups. Similarly, the extent to which the explanatory factors affected the outcomes varied. This is perhaps to be expected and may reflect the diversity of characteristics and experiences of different ethnic groups.

Life satisfaction

In 2020 across the UK, those identifying within the Black, African, Caribbean or Black British ethnic group were more likely to report low life satisfaction than those identifying within the White ethnic group. This is before adjusting for any other factors and after adjusting for age and sex only.

However, we found no significant difference in the likelihood of reporting low life satisfaction among those identifying within the Black, African, Caribbean or Black British ethnic group compared with those within the White ethnic group after adjusting for:

• age

• sex

• educational attainment

• employment status

• housing tenure

• marital status

• region

• socio-economic status

• type of interview conducted at survey

This suggests that these factors are important in understanding the differences seen between these groups.

This analysis reflects a period during which a range of coronavirus (COVID-19) restrictions were in place, so the patterns seen here may be different to those that would have been seen before the coronavirus pandemic.

Crime

In the year ending March 2020 across England and Wales, compared with the White British ethnic group, only those identifying within the Pakistani or Bangladeshi ethnic groups or within the Mixed or Multiple ethnic group were more likely to be a victim of crime before adjusting for any other factors (Figure 13).

After adjusting for age and sex, the estimated odds of being a victim of crime were lower among those identifying within the Chinese or Other Asian ethnic group than among the White British ethnic group. This suggests that at any given age and sex, those identifying within the Chinese or Other Asian ethnic group are less likely to be a victim of crime than those within the White British ethnic group.

We then adjusted for the following factors:

• age

• sex

• educational attainment

• employment status

• household composition

• household income

• housing tenure

• Index of Multiple Deprivation (IMD) decile

• marital status

• region

• socio-economic status and

• urban-rural status

We found that those identifying within the Chinese or Other Asian ethnic group were still less likely to be a victim of crime compared with those identifying within the White British ethnic group.

This was also the case for those identifying within the Black, African, Caribbean or Black British ethnic group and the Other White ethnic group. These findings would benefit from further detailed research into the patterns observed.

Figure 13: After adjusting for age and sex, adults identifying within the Chinese or Other Asian ethnic group were less likely to be a victim of crime than those identifying within the White British ethnic group

Odds ratios of adults who were a victim of crime at least once within the last 12 months (excludes fraud, sexual offences and computer misuse)​ by ethnic group in England and Wales, year ending March 2020

Source: Office for National Statistics - Crime Survey for England and Wales

Notes:

1. 95% confidence intervals are displayed on the chart, see the Glossary for details.
2. Basic models are adjusted for age and sex. Fully adjusted models are adjusted for age, sex, educational attainment, employment status, household composition, household income, housing tenure, Index of Multiple Deprivation decile, marital status, region, socio-economic status and urban-rural status.
3. A victim of crime is someone who had experienced at least one personal crime or had been a resident in a household that was a victim of at least one household crime in the last 12 months, irrespective of whether they reported these incidents to the police (excludes fraud, sexual offences and computer misuse).

Download this image Figure 13: After adjusting for age and sex, adults identifying within the Chinese or Other Asian ethnic group were less likely to be a victim of crime than those identifying within the White British ethnic group

.svg (3.0 MB) .xlsx (19.7 kB)

Digital exclusion

From January to March 2020 in the UK, before adjustment, people identifying within the Bangladeshi and Pakistani ethnic groups had the same likelihood of being digitally excluded as those identifying within the White ethnic group.

However, after adjusting for age and sex, the likelihood of those within the Bangladeshi and Pakistani ethnic groups being digitally excluded was higher compared with those identifying within the White ethnic group. This remained the case after adjusting for the following factors:

• age

• sex

• educational attainment

• household composition

• housing tenure

• marital status

• region

• socio-economic status

People identifying within the Black, African, Caribbean or Black British ethnic group were significantly less likely to be digitally excluded before adjustment. However, after adjusting for age and sex, they were significantly more likely to be digitally excluded than those identifying within the White ethnic group.

Together, these patterns suggest that holding age and sex constant, those identifying within the Bangladeshi, Pakistani and Black, African, Caribbean or Black British ethnic groups are more likely to be digitally excluded than their White counterparts.

For all other ethnic groups, once age, sex and factors listed previously have been adjusted for, there were no significant differences compared with those identifying within the White ethnic group.

Figure 14: After adjusting for age and sex, the odds of being digitally excluded for someone identifying within a Bangladeshi or Pakistani ethnic group were higher than for someone identifying within a White ethnic group

Odds ratios of adults who were digitally excluded (adults who have not used the internet within the last three months)​ by ethnic group in the UK, January to March 2020

Source: Office for National Statistics - Labour Force Survey

Notes:

1. 95% confidence intervals are displayed on the chart, see the Glossary for details.
2. Basic models are adjusted for age and sex. Fully adjusted models are adjusted for age, sex, educational attainment, household composition, housing tenure, marital status, region and socio-economic status.
3. In this analysis anyone that had not used the internet within the last three months (including those who had never used the internet) were defined as digitally excluded.

Download this image Figure 14: After adjusting for age and sex, the odds of being digitally excluded for someone identifying within a Bangladeshi or Pakistani ethnic group were higher than for someone identifying within a White ethnic group

.svg (2.7 MB) .xlsx (18.6 kB)

Mortality

In the year ending 26 March 2012 in England and Wales, people identifying within all other ethnic groups had a lower likelihood of mortality compared with those identifying within the White British ethnic group. This is before adjusting for any socio-demographic factors. The exception was for people identifying within the White Irish ethnic group who had higher odds of death compared with those within the White British group.

However, after adjusting for age and sex, people identifying within the Gypsy or Irish Traveller, White and Black Caribbean, and White Irish ethnic groups had a significantly higher likelihood of mortality compared with people identifying within the White British ethnic group.

For those identifying within the White and Black Caribbean ethnic group, this difference was no longer evident after adjusting for:

• age

• sex

• English language proficiency

• housing tenure

• Index of Multiple Deprivation (IMD) decile

• marital status

• region

• religion

• socio-economic status

• urban-rural status

For those identifying within the White Irish ethnic group, after adjusting for these factors they had a lower likelihood of mortality. However, they still had higher odds of death when compared with those identifying within the White British ethnic group.

After adjusting for these factors, those identifying within the Gypsy or Irish Traveller ethnic group still had higher odds of death compared with those identifying within the White British ethnic group.

Figure 15: After adjusting for a range of factors, those identifying within the Gypsy or Irish Traveller and White Irish ethnic groups had higher odds of mortality than those identifying within the White British ethnic group

Odds ratios of mortality by ethnicity compared with the White British group in England and Wales, 27 March 2011 to 26 March 2012

Source: Office for National Statistics - Public Health Data Asset

Notes:

1. 95% confidence intervals are displayed on the chart, see the Glossary for details.
2. Basic models are adjusted for age and sex only. Fully adjusted models are adjusted for age, sex, English language proficiency, housing tenure, Index of Multiple Deprivation decile, marital status, region, religion, socio-economic status and urban-rural status.
3. An individual on the Public Health Data Asset was deemed to have died if they were recorded on the 2011 Census and subsequently linked to a death registration record with a date of death within a year of the 2011 Census (between 27 March 2011 and 26 March 2012).

Download this image Figure 15: After adjusting for a range of factors, those identifying within the Gypsy or Irish Traveller and White Irish ethnic groups had higher odds of mortality than those identifying within the White British ethnic group

.svg (5.1 MB) .xlsx (20.5 kB)

Wealth

Between April 2018 and March 2020 in Great Britain, people identifying within most ethnic groups had less wealth on average than those identifying within the White British ethnic group before any adjustment. The exception to this was those identifying within the Chinese, and White and Asian ethnic groups. These groups had comparable levels of wealth to those identifying within the White British ethnic group both before and after adjusting for a range of factors.

Across most ethnic groups, after adjusting for age and sex these differences were smaller to varying extents. The effect of additionally adjusting for economic activity status, educational attainment, household composition, housing tenure, industry type and socio-economic status also had varying results.

In the case of people identifying within the Black Caribbean, Other Black, Other Mixed, Other White, and White and Black Caribbean ethnic groups, these adjustments resulted in comparable levels of wealth to those identifying within the White British ethnic group. This suggests that these factors affect the differences seen.

Across those identifying within the remaining ethnic groups, although adjusting for these additional factors in most cases resulted in smaller differences in wealth compared with those identifying within the White British ethnic group, they still had less wealth on average. This indicates that there may be other factors not included in this analysis that are affecting these differences.

Figure 16: Adjusting for a range of factors reduced wealth differences for those identifying within some, but not all, ethnic groups compared with those identifying within the White British ethnic group

Difference in average (mean) total individual wealth by ethnicity in Great Britain, April 2018 to March 2020

Source: Office for National Statistics - Wealth and Assets Survey

Notes:

1. Only people aged 19 years and over or aged 16 to 18 years and not in full-time education have been included in the analysis.
2. 95% confidence intervals are displayed on the chart, see the Glossary for details.
3. Basic models are adjusted for age and sex only. Fully adjusted models are adjusted for age, sex, economic activity status, educational attainment, household composition, housing tenure, industry type and socio-economic status.
4. In this analysis, we use the mean as a measure of average wealth. Mean wealth is the total wealth of all individuals divided by the number of individuals. Total wealth is the sum of net wealth in property, private pensions, financial and physical items.

Download this image Figure 16: Adjusting for a range of factors reduced wealth differences for those identifying within some, but not all, ethnic groups compared with those identifying within the White British ethnic group

.svg (4.5 MB) .xlsx (19.4 kB)

Across the regions and countries of the UK, outcomes across the areas of life included in our analysis varied. Analysis of differences in average wealth by region has not been included (see Section 11: Data sources and quality for more information.) With the exception of mortality and life satisfaction, in most cases, adjusting for explanatory factors had little impact on the likelihood of experiencing the outcome relative to the reference group. This indicates that these factors were not affecting the patterns seen.

In 2020, any differences in levels of reporting of low life satisfaction across areas of the UK before adjustment were no longer evident after adjusting for the following factors:

• age

• sex

• disability status

• educational attainment

• ethnicity

• sexual orientation

Similarly, there were no differences in the likelihood of reporting low life satisfaction across the regions and countries of the UK compared with the South East. This was after adjusting for:

• age

• sex

• disability status

• educational attainment

• ethnicity

• sexual orientation

• employment status

• housing tenure

• socio-economic status

• type of interview conducted at survey

Between January and March 2020 across the UK, people in all regions aside from the East of England, London and the South West were more likely to be digitally excluded compared with the South East. This was not affected by adjusting for any of the additional factors included in this analysis.

Figure 17: Those living outside of the south and east of England were significantly more likely to be digitally excluded in the UK compared with those living in the South East

Odds ratios of adults who were digitally excluded (adults who have not used the internet within the last three months)​ by region in the UK, January to March 2020

Source: Office for National Statistics - Labour Force Survey

Notes:

1. 95% confidence intervals are displayed on the chart, see the Glossary for details.
2. Basic models are adjusted for age, sex, disability status, educational attainment and ethnicity. Fully adjusted models are adjusted for age, sex, disability status, educational attainment, ethnicity, household composition, housing tenure and socio-economic status.
3. In this analysis anyone that had not used the internet within the last three months (including those who had never used the internet) were defined as digitally excluded.

Download this image Figure 17: Those living outside of the south and east of England were significantly more likely to be digitally excluded in the UK compared with those living in the South East

.svg (3.5 MB) .xlsx (18.9 kB)

In the year ending March 2020 in England and Wales, people living in all regions except the East Midlands, East of England and London were less likely to be victims of crime, compared with people living in the South East. This was both before and after adjusting for a range of different factors.

London was the only region where fully adjusting for a range of factors affected the outcome. Before adjustment, people living here were more likely to be a victim of crime compared with people living in the South East, whereas after adjustment there was no longer any difference.

Figure 18: The odds of being a victim of crime for adults in most English regions and Wales was lower than for those living in the South East 

Odds ratios of adults who were a victim of crime at least once within the last 12 months (excludes fraud, sexual offences and computer misuse)​ by region in England and Wales, year ending March 2020

Source: Office for National Statistics - Crime Survey for England and Wales

Notes:

1. 95% confidence intervals are displayed on the chart, see the Glossary for details.
2. Basic models are adjusted for age, sex, disability status, educational attainment, ethnicity and sexual orientation. Fully adjusted models are adjusted for age, sex, disability status, educational attainment, ethnicity, sexual orientation, employment status, household composition, household income and housing tenure, Index of Multiple Deprivation decile, socio-economic status and urban-rural status.
3. A victim of crime is someone who had experienced at least one personal crime or had been a resident in a household that was a victim of at least one household crime in the last 12 months, irrespective of whether they reported these incidents to the police (excludes fraud, sexual offences and computer misuse).

Download this image Figure 18: The odds of being a victim of crime for adults in most English regions and Wales was lower than for those living in the South East 

.svg (3.0 MB) .xlsx (18.6 kB)

In the year ending 26 March 2012, people living in all regions in England and Wales other than the West Midlands, East of England and London, had a higher likelihood of mortality compared with people in the South East region before any adjustments.

After adjusting for age, disability status, ethnicity, religion and sex, there was a lower likelihood of mortality across England and Wales in all regions and countries, other than Yorkshire and The Humber and the North West, when compared with the South East. This suggests that other factors not included in this analysis may be important in these areas.

After further adjusting for housing tenure, Index of Multiple Deprivation (IMD) decile, socio-economic status and urban-rural status, the pattern was altered such that people in all regions of England and in Wales had a lower likelihood of mortality compared with people in the South East.

Figure 19: After fully adjusting for a range of factors, people living in all regions of England and in Wales had lower odds of mortality than those living in the South East 

Odds ratios of mortality by region in England and Wales compared with the South East, 27 March 2011 to 26 March 2012 

Source: Office for National Statistics - Public Health Data Asset

Notes:

1. 95% confidence intervals are displayed on the chart, see the Glossary for details.
2. Basic models are adjusted for age, sex, disability status, ethnicity and religion. Fully adjusted models are adjusted for age, sex, disability status, ethnicity, religion, housing tenure, Index of Multiple Deprivation decile, socio-economic status and urban-rural status.
3. An individual on the Public Health Data Asset was deemed to have died if they were recorded on the 2011 Census and subsequently linked to a death registration record with a date of death within a year of the 2011 Census (between 27 March 2011 and 26 March 2012).

Download this image Figure 19: After fully adjusting for a range of factors, people living in all regions of England and in Wales had lower odds of mortality than those living in the South East 

.svg (3.0 MB) .xlsx (19.5 kB)

Data on sexual orientation were only available in the sources used for the analysis of crime, life satisfaction and wealth. Across these outcomes, people who identify as bisexual were more likely to be a victim of crime (Figure 20) and report low life satisfaction (Figure 21) and had less wealth on average (Figure 22) than those who identify as heterosexual or straight. People who identify as gay or lesbian were also more likely to be victims of crime than those identifying as heterosexual or straight.

Differences in the likelihood of being a victim of crime and reporting low life satisfaction were unaffected by the inclusion of any factors in either the basic or fully adjusted models. This indicates that there may be other factors not included in the analysis that are associated with these differences.

Figure 20: Even after adjusting for a range of factors, people who identify as bisexual, gay or lesbian were more likely to be victims of crime than those who identify as heterosexual or straight

Odds ratios of adults who were a victim of crime at least once in the last 12 months (excludes fraud, sexual offences and computer misuse)​ by sexual orientation compared with heterosexual or straight people in England and Wales, year ending March 2020

Source: Office for National Statistics - Crime Survey for England and Wales

Notes:

1. 95% confidence intervals are displayed on the chart, see the Glossary for details.
2. Basic models are adjusted for age and sex. Fully adjusted models are adjusted for age, sex, educational attainment, employment status, household composition, household income, housing tenure, Index of Multiple Deprivation decile, marital status, region, socio-economic status and urban-rural status.
3. A victim of crime is someone who had experienced at least one personal crime or had been a resident in a household that was a victim of at least one household crime in the last 12 months, irrespective of whether they reported these incidents to the police (excludes fraud, sexual offences and computer misuse).

Download this image Figure 20: Even after adjusting for a range of factors, people who identify as bisexual, gay or lesbian were more likely to be victims of crime than those who identify as heterosexual or straight

.svg (1.3 MB) .xlsx (18.1 kB)

Figure 21: People who identify as bisexual were more likely to report low life satisfaction compared with those who identify as heterosexual or straight, even after adjusting for a range of factors

Odds ratios of adults who reported low life satisfaction by sexual orientation in the UK compared with people that identified as heterosexual or straight, 2020 

Source: Office for National Statistics - Annual Population Survey

Notes:

1. 95% confidence intervals are displayed on the chart, see the Glossary for details.
2. Basic models are adjusted for age and sex. Fully adjusted models are adjusted for age, sex, educational attainment, employment status, housing tenure, marital status, region, socio-economic status and type of interview conducted at survey.
3. In this analysis, low life satisfaction was defined as a score of 0 to 4 inclusive, on a scale of 0 to 10, where 0 is “not at all” and 10 is “completely”, in response to how satisfied people are with their life overall.

Download this image Figure 21: People who identify as bisexual were more likely to report low life satisfaction compared with those who identify as heterosexual or straight, even after adjusting for a range of factors

.svg (1.3 MB) .xlsx (19.0 kB)

Between April 2018 and March 2020 in Great Britain, people who identify as bisexual also had less wealth on average than those who identify as heterosexual or straight before adjustment. This difference in average wealth was reduced after adjusting for age and sex. However, those who identify as bisexual still had less wealth. This difference in average wealth was broadly similar after adjusting for:

• age

• sex

• economic activity status

• educational attainment

• household composition

• housing tenure

• industry type

• socio-economic status

Figure 22: People who identify as bisexual had less wealth when compared with those who identify as heterosexual or straight, even after adjusting for a range of factors

Difference in average (mean) total individual wealth by sexual orientation in Great Britain, April 2018 to March 2020

Source: Office for National Statistics - Wealth and Assets Survey

Notes:

1. Only people aged 19 years and over or aged 16 to 18 years and not in full-time education have been included in the analysis.
2. 95% confidence intervals are displayed on the chart, see the Glossary for details.
3. Basic models are adjusted for age and sex. Fully adjusted models are adjusted for age, sex, economic activity status, educational attainment, household composition, housing tenure, industry type and socio-economic status.
4. In this analysis, we use the mean as a measure of average wealth. Mean wealth is the total wealth of all individuals divided by the number of individuals. Total wealth is the sum of net wealth in property, private pensions, financial and physical items.

Download this image Figure 22: People who identify as bisexual had less wealth when compared with those who identify as heterosexual or straight, even after adjusting for a range of factors

.svg (1.3 MB) .xlsx (19.0 kB)

Adult

Unless otherwise indicated, an adult is defined as someone aged 16 years and over.

Average wealth

In this analysis, we used the mean as a measure of average wealth. Mean wealth is the total wealth of all individuals divided by the number of individuals. Total wealth is the sum of net wealth in property, private pensions, financial and physical items.

Confidence intervals

Confidence intervals use the standard error to derive a range in which we think the true value is likely to lie. A confidence interval gives an indication of the degree of uncertainty of an estimate and helps to decide how precise a sample estimate is. It specifies a range of values likely to contain the unknown population value. These values are defined by lower and upper limits. The width of the interval depends on the precision of the estimate and the confidence level used. A greater standard error will result in a wider interval; the wider the interval, the less precise the estimate is.

Digital exclusion

In this analysis anyone that had not used the internet within the last three months or had never used the internet were defined as digitally excluded.

While recognising that digital skills are as important for digital inclusion as internet usage, for the purpose of this analysis, digital exclusion has been defined based only on internet usage. This reflects the availability of relevant information within the data source. If other characteristics related to digital exclusion that were not available in the data source were to be included in an analysis of this kind, they may produce different results.

Disability

To define disability in the analysis for crime, digital exclusion, life satisfaction and wealth, we refer to the Government Statistical Service (GSS) harmonised “core” definition.

To define disability in the mortality analysis, we refer to the self-reported answers to the 2011 Census question, "Are your day-to-day activities limited because of a health problem or disability which has lasted, or is expected to last, at least 12 months? - Include problems related to old age", which are:

• Yes, limited a lot

• Yes, limited a little

• No

This is slightly different to the current GSS harmonised "core" definition. The GSS definition is designed to reflect the definitions that appear in legal terms in the Disability Discrimination Act 1995 and the subsequent Equality Act 2010.

Ethnicity

Ethnicity is generally considered multifaceted and includes a combination of:

• common ancestry

• elements of culture

• identity

• religion

• language

• physical appearance

• other factors

For all outcomes in this analysis, ethnicity was self-reported by the respondents to the census or specific survey used. The answer options available to respondents living in the different nations of the UK differ slightly. This placed constraints on the level of aggregation of ethnic groups that could be used for a topic if the geographical coverage of the data source was Great Britain or the UK. Sample size also influenced the level of aggregation of ethnic groups that could be used for a topic.

Life satisfaction

Life satisfaction is one of our four measures of personal well-being. Respondents are asked to evaluate, on a scale of 0 to 10, where 0 is “not at all” and 10 is “completely”, how satisfied they are with their life overall.

Life satisfaction thresholds

Thresholds are used to present the distribution of the personal well-being data. For the life satisfaction question, response categories are grouped as below. In this analysis we have focused on the “low” threshold:

• 0 to 4 out of 10 - low well-being

• 5 to 6 out of 10 - medium well-being

• 7 to 8 out of 10 - high well-being

• 9 to 10 out of 10 - very high well-being

Mortality

An individual on the Public Health Data Asset was deemed to have died if they were recorded on the 2011 Census and subsequently linked to a death registration record with a date of death within a year of the 2011 Census (between 27 March 2011 and 26 March 2012).

Reference category

A reference category is identified as a category within a variable that all other categories can be compared with. In most instances, the category with the highest population was identified as the reference category. For example, the sample of females was higher than males, therefore “Female” was chosen as the reference category for the variable “Sex”.

Odds ratio

An odds ratio for a particular group describes the relative difference in the likelihood of an outcome in that group compared with a reference category, which in this analysis was based on majority count. An odds ratio higher than 1 indicates a greater likelihood of that outcome, while an odds ratio less than 1 indicates a lower likelihood.

Victim of crime

A victim of crime is someone who, when responding to the Crime Survey for England and Wales, had experienced at least one personal crime or had been a resident in a household that was a victim of at least one household crime in the last 12 months, irrespective of whether they reported these incidents to the police. Household crimes may have happened to anyone in the household, while personal crimes are only counted if they relate to the individual being interviewed.

It does not distinguish between victims of a single crime and multiple victimisations. Experiences of crime therefore do not capture the total burden of crime. Victims of fraud or computer misuse are excluded from this analysis. Additionally, the analysis does not include those who have experienced sexual offences because of under-reporting in the survey and the resulting unreliability in estimates.

Uncertainty and quality

Statistical significance of differences between groups when compared with the reference category were determined through p-values smaller than 0.05 (the 5% level). We have also used confidence intervals to represent uncertainty around the estimated odds ratios and the estimates of differences in wealth. More information is available in our Uncertainty and how we measure it guide.

The statistics presented are estimates and, as with all estimates, there is a level of uncertainty associated with them. Given this, sample-based estimates will occasionally indicate a difference between population groups when there is in fact no systematic difference between them. Such findings are known as "false positives". If we were able to repeatedly draw different samples from the population, then, for a single comparison, we would expect 5% of findings with a p-value below a threshold of 0.05 to be false discoveries. However, if multiple comparisons are conducted, then the probability of making at least one false discovery will be greater than 5%.

Weighting

For the analysis on mortality, we used a sample of people for which there had been a successful linkage between the 2011 Census and the Public Health Data Asset. As not everyone on the 2011 Census was able to be linked to the Public Health Data Asset, we created and applied an inverse probability weight to the successfully linked sample to bring the weighted totals and proportions back in line with those measured by the 2011 Census.

For the digital exclusion analysis, we created a specific weight to account for the relatively high level of missingness in the internet usage variable. In the Labour Force Survey, data for non-responding individuals in responding households are rolled forward from the previous calendar quarter. As the question on internet usage is only asked in the first calendar quarter, non-responding individuals in January to March are missing data on internet usage. It was considered unlikely that non-responding individuals within responding households occurred at random. Therefore, a specific weight for those who had provided a valid answer to the internet usage question was needed to bring the weighted totals and proportions of this sample back in line with that of the UK population.

For the analysis on other topics, the weights provided with the data sources were used.

Crime Survey for England and Wales (CSEW)

More information on the Crime Survey for England and Wales (CSEW) and its methodology is available in our Crime and justice methodology.

Labour Force Survey (LFS)

More quality and methodology information on the Labour Force Survey (LFS) including creation, uses and evaluation can be found in our Labour Force Survey Quality and Methodology Information report.

Annual Population Survey (APS)

The analysis on life satisfaction is based on Annual Population Survey (APS) data covering adults aged 16 years and over in the UK. The life satisfaction question can only be answered in person and proxy responses are not accepted. More quality and methodology information on the APS including creation, uses and evaluation can be found in our Annual Population Survey Quality and Methodology Information report.

Public Health Data Asset (PHDA)

The Public Health Data Asset (PHDA) combines the 2011 Census records, death registrations, Hospital Episode Statistics (HES) and primary care records retrieved from the General Practice Extraction Service (GPES) Data for Pandemic Planning and Research (GDPPR). Information about these data sources, how they have been linked, and the methods used for previous publications can be found in our Deaths involving COVID-19 by religious and ethnic group methodology.

Wealth and Assets Survey (WAS)

More information on data collection, methods, strengths and limitations is available in the Wealth and Assets Survey Quality and Methodology Information report and our Measuring wealth on an individual level methodology.

Within the wealth analysis, geographical region was not controlled for in the modelling, because while there is an association between region and wealth, it is unclear whether region determines wealth or wealth determines region. In the case of the latter, controlling for region would distort the model estimates.

This analysis is the first stage of a broader programme of work to improve the data and evidence to understand what contributes to better or worse outcomes for different population groups across a range of areas of life.

Using various individual and area characteristics, we have shown how outcomes vary across different groups of people, some of whom are experiencing poorer outcomes across multiple areas of life. It has also provided a starting point to better understand the characteristics and circumstances that may affect these outcomes.

As part of future work, we will be developing an Equality Data Asset with the Integrated Data Service. This will securely link demographic data from Census 2021 to other relevant information on outcomes across a range of areas of life. It will provide a rich resource for exploring barriers to opportunity for different groups of people in more detail, enabling decision-makers and communities to see where inequality exists and consider how best to address it.