Ways of extending use (and increasing value):
Where people of different ethnic groups live, and the residential moves that shape the diversity of neighbourhoods, are a reflection of residential choice and constraint. Debate in Britain over the past decade has focused on explanations of internal migration based on ethnic conflict; particularly, the idea that residential decision making is driven by desires to increase physical separation between ethnic groups. This paper demonstrates increased residential ethnic mixing as a result of internal migration, and argues that ‘white flight’ and ‘white avoidance’ are inappropriate descriptions of residential mobility for ethnic groups in the UK. Indeed, the 2011 census reveals that white and non-white groups are moving away from the most diverse neighbourhoods, on balance, at similar rates: there is dispersal from minority concentrations, as was seen in the 1991 and 2001 censuses. Furthermore, Whites are more attracted than non-Whites to the most diverse neighbourhoods.
This all points to support for theories of spatial assimilation. However, using tables that have not been available in previous censuses, the paper analyses the age and socio-economic class profile of internal migration of ethnic groups from areas in which they are concentrated. As expected, the general profile is for highest rates of (out) migration from diverse areas for Managers and Professionals but there are some interesting subtleties to the patters. For example, non-White managers and professionals are moving out of non-white concentrations at twice the rate of non-White Routine workers whereas for Whites, Managers and Professionals and Routine workers are leaving these areas, on balance, at similar rates. The results raise the questions of whether there are particular residential constraints for minorities in lower socio-economic groups, and lead to the paper’s conclusion that it is such potential constraints – including discrimination – that should be the focus of further research.
The extent of inequality experienced by ethnic minorities in the UK has been well documented in the literature. While studies have shown the persistent disadvantage faced by ethnic minorities in a range of socio-economic outcomes compared with the White British, few studies have examined how ethnic minority outcomes vary at the local level. This paper draws on data from the 2001 and 2011 Censuses to examine the spatial patterns of ethnic inequality across local authority districts in England and Wales through an area classification framework. Ethnic inequality is measured as the absolute difference in the proportion of White British people and people from ethnic minority groups who experience disadvantage in employment, education, health and housing. The analysis suggests that ethnic inequalities are widespread but manifest in different ways in different localities in line with differences in local ethnic, social and economic structures.
This paper evidences persistent minority disadvantage and gender inequality in the labour market using data from the 1991, 2001 and 2011 censuses. Patterns of labour market participation, unemployment and types of employment across the eighteen ethnic groups detailed in the 2011 Census are described (for ages 25-49); and changes in employment for the seven groups that can be derived in a reasonably consistent way across the 1991, 2001 and 2011 Censuses are tracked. The paper shows that the White ethnic groups (with the marked exception of the Gypsy and Irish Traveller group) were in a more advantaged position in the labour market compared with other ethnic groups throughout the 1990s and 2000s. This advantage is apparent in terms of higher rates of economic activity for both men (where only Indian men had a similar rate) and women (where only Black Caribbean women had a similar rate) and in lower rates of unemployment (where only Chinese, Indian and Other Asian men had a similar rate to men in the White groups, and only Chinese women had the same rate as women in the White groups).
Women had lower rates of economic activity than men in all ethnic groups. However, the difference in rates was largest for Bangladeshi, Pakistani, Arab and Gypsy & Irish Traveller groups. Although Pakistani and Bangladeshi women were the least likely groups to be in the labour market throughout the period, they experienced the most significant rise in rates of economic activity since 1991, largely due to an increase in part-time working. The paper concludes that, given the implications of socio-economic disadvantage for other social realms including health and housing, both academia and policy should be attentive to ethnic inequalities in employment.
A distinctive element of CoDE’s work with the UK censuses has been the production of a series of Census Briefings on Dynamics of Diversity, in collaboration with the Joseph Rowntree Foundation (JRF). This presentation will reflect on: how census data can be carefully presented for public and policy audiences; how the Briefings format has engaged new audiences with census data; what was necessary in terms of CoDE and JRF’s collaboration to make this use of census data possible; and, the value of the collaboration and Briefings for JRF’s work to influence policy, practice and public debate and effect social change.
DataShine is an automated approach to mapping the many 2011 UK Census Quick Statistics tables that been made available, producing maps within a web interface with user-specifiable colours and a conventional cartography that constrains the data colouring (choropleth) to building areas, allowing the street network and natural features to add context to a data-based map and increase its relatability to the real world. This best-of-both‐worlds approach retains map topography while reducing the overemphasis effects inherent in conventional choropleth mapping.
We outline key technical considerations in the construction of DataShine, mixing on-the-fly raster mapping imagery with vector-based metadata display on a neal-real-time basis, including optimisation and simplification considerations, and detail the algorithm used to determine the data catagorising strategy that assigns colours to areas. Careful choice of colour ramp intervals allow for discovery of variations and patterns in Census data across both local and regional extents. We outline additional functionality that has been added to the web interface to aid our key end-‐users, including targeted data download, overlaying of administrative areas and simple view re-‐creation.
In addition, we look at an approach to visualising some of the small-area origin-destination matrices that have also been made available, using a vector straight-line approach where opacity, width and colour are used to show the data clearly and simply.
This paper describes the sources, methods and results of two related and highly applied projects on historical census data funded by government bodies for policy purposes. The first, for the European Union, is creating time series 1951 to 2011 for the total populations of the 8,941 Wards of Great Britain as used by the 2011 census. The second, for the Greater London Authority (GLA), is redistricting diverse data from censuses 1801-1961, creating consistent data for the current London Boroughs, for London’s Central Activity Zone (CAZ) as defined by the GLA, and for the overall GLA area; focused on London’s industrial structure, converting diverse historical classifications to Standard Industrial Classification 2007 (SIC). Both projects required very varied historical reporting areas data to be redistricted to a single set of modern units therefore requiring both modern and historical units boundary data; and preferably that the historical units be the most detailed.
The Great Britain Historical GIS project has been working for over twenty years with historical census data, and our focus is real world use. Our experience shows that to achieve significant non-academic 'impact' research must come up to the present, and achieving long runs of data requires new skills: disinterring obscure statistical datasets; locating even more obscure boundary maps; even negotiating copyright. Also, that there is a far greater requirement among both policy makers and the general public for local time series than for maps or other purely cross-sectional presentations and analyses. Further, policy makers almost always require time series for modern units, even though it is generally easier to redistrict modern small area data to less detailed historical units. This means that GIS techniques must be used to redistrict these diverse data sets to a single constant geography.
In Westminster, space to build new homes is at a premium. Much of the 19th and 20th century public housing is no longer fit for purpose and there is an oversupply of apartments unsuitable for families. Surveys reveal however, that residents fear that housing renewal will change the character of their area, bringing new people and fears of division.
The Intelligence and Analysis team was approached to model the possible outcomes for two housing renewal developments in the borough, using census data as a base, and local intelligence from surveys and development plans where possible. In particular housing commissioners needed to:
An excel-based calculator was developed that could be used anywhere in the borough. The model uses as its basis ward-based population yields from the Census (based on the size and tenure of property). Depending on the stage of the project, information about existing residents or proposed builds may be inputted into the calculator as it becomes available (where these are not known, ward averages are used as default).
Although the product is, in effect, a constantly updated ‘living document’, housing commissioners can start to understand the possible outcomes of the proposed developments, numbers to be re-housed, the mobility of new and old residents and the service impact for schools and other services.
The 2011 Census of England and Wales was the first to employ a bespoke set of output zones for the release of workplace-related data. These ‘workplace zones’ enabled the Office for National Statistics (ONS) to publish 22 tables relating to the characteristics of workplaces and workers at the workplace zone level, compared to just four at the output area level for the 2001 Census. Compared to output areas, workplace zones provide higher spatial resolution in areas where workers and workplaces are highly concentrated and greater levels of aggregation in areas where they are sparsely distributed.
NS produces various area-based classifications, such as the Output Area Classification (OAC) and the rural-urban classification, as well as specific geographies, such as built-up areas. Each classification and geography has its own focus, with OAC, for example, being based on the characteristics of residential households and places of residence. To date, it has not been possible to produce an equivalent area-based classification for workplaces and workers at the small area level due to a lack of sufficient variables and appropriate geography. The release of workplace zones and workplace data now make this feasible.
This paper outlines the methods used to create a new geodemographic classification of workplaces zones for England and Wales using 2011 Census data. Although the methods implemented are broadly similar to those used previously by OAC, the very different distributions and issues associated with mapping and measuring workers and workplaces at this new level of geography have necessitated extensive exploratory analysis and required important conceptual and methodological decisions to be made. The paper explores findings from this analysis and presents illustrative results of the classification. Finally, it discusses the challenges and opportunities involved in extending the geographical coverage of workplace zones and the classification to the whole of the UK.
Commuting to work is an activity that is carried out relatively frequently, often daily, by the vast majority of individuals in employment. It is, therefore, a relatively important part of many people’s routine behaviour, with the nature of an individual’s commute impacting upon their lifestyle, both directly and indirectly.
The paper aims to do three things. First, it is designed to contribute new analyses of the relationships between commuting behaviour and patterns and individual predictor variables. Second, it endeavours to fill a substantial gap in the literature related to the systematic, simultaneous and quantitative analysis of several important sociodemographic and geographic variables and their relationships with commuting over different distances using particular modes of travel. Third, it attempts to shed light on the temporal changes in commuting behaviour between 1991, 2001 and 2011.
Binary logistic regression is the statistical method that has been used to analyse the variations in commute distance and mode of transport disaggregated for those variables which past research has suggested are important. The analyses are being carried out using microdata from each of the Individual Samples of Anonymised Records (I-SARs) from the last three censuses after individuals who ‘work mainly at or from home’ have been excluded.
The findings from this research will inform the formulation of some policy suggestions for implementation by regional or local governments or any other organisations with a responsibility to supply and maintain transport networks for those who require them.
Mike Coombes and Tony Champion, CURDS, Newcastle University: Using flow and stock data from Censuses to compare trends in British city regions
City decline has attracted growing interest in several countries which have not experienced the high level of net in-migration recently seen in Britain. The fact that absolute population decline has become rare among UK cities has been confirmed by Census data which leads to than emphasis here on identifying relative decline. Looking at the city regions housing 74 larger cities of the UK, multi-dimensional analyses utilise diverse Census and other datasets to track longer term trends.
The analyses placed particular emphasis on forms of decline likely to increase the risk of poverty. This leads to a focus on the labour market and trends in the ‘more and better jobs’ that are central to poverty reduction. The labour market focus led to analysis by city region so that change in local job availability can be linked to change in local employment rate.
The preliminary findings included some surprises, such as that net in-commuting to city regions has begun to decline. Then a synthesis index of relative decline is developed, and a set of other variables are used to model the resulting city region index scores. The results suggest that the risk of a city region experiencing relative decline in the last decade was reduced if the city had:
So far as these results are significant for policy, their implications are stark. A city cannot really alter the characteristics that are relevant to most of these four factors.
Jane Falkingham, Maria Evandrou, James Robards and Athina Vlachantoni, University of Southampton: The prevalence of informal care and its association with health: longitudinal research using census data for England and Wales
Informal or unpaid caregiving is increasingly significant in the context of an ageing population and pressures related to the funding of care for older people. At the 2011 Census informal caring prevalence was higher than at 2001 and more people provided 20 hours or more care per week. Using a 1% sample of England and Wales 2011 Census records matched to the 2001 Census responses from the same individuals, the Office for National Statistics Longitudinal Study, this paper investigates informal carers at 2011 with reference to their caring role at 2001.
Using a typology of caring status, the paper examines the characteristics associated with providing care in 2011 and investigates the relationship between caring and health, taking into account selection into the caring role at 2011 depending on preceding health. Multivariate analyses to predict informal caring at 2011 among the carers at 2001 show that those providing 20 hours or more care in 2001 were the most likely to be caring at 2011, suggesting that past provision of care is crucial in predicting future caring.
With regard to the health of carers, compared to non-carers (1) there are similarly low odds of reporting bad or very bad health at the 2011 Census among those providing light (1-19 hours per week) or heavy (20+ hours per week) care at 2011 who were not caring 10 years before and (2) light informal carers at 2001 not caring 10 years later have no difference in the odds of reporting bad or very bad health.
The aim of this research, involving data linkage and health outcomes, is to gain a full understanding of the impact of both fertility histories and childlessness on health outcomes mid-life. The research draws on and extends work on reproductive histories and life-course outcomes. We aim to extend this area of research specifically for Scotland based on Scottish Census data (1991-2011), namely the Scottish Longitudinal Study (SLS) linked to health data from the NHS Scottish Morbidity Records (SMR).
Where the census health measures, including the new 2011 Census health condition question on mental health, are the research outcomes and the explanatory information is from Census socio-economic data captured along with the SMR02 Maternity Inpatient and Day Case dataset and the SMR04 Mental Health Inpatient and Day Case dataset.
The SLS allows follow-up to mid-life for specific female SLS birth cohorts from the 1991 Census. From preliminary modelling we find high birth parity to be an important factor in relation to self-reported mental health conditions. For limiting long-term illness birth parity is initially important but not once socio-economic variables are controlled for. Preliminary modelling also highlights that relationship status, single, married or cohabiting, to be important over that of legal marital status as recorded at Census.
Fran Darlington and Paul Norman, University of Leeds and Dimitris Ballas, University of Sheffield: Changing patterns of ethnic health and socio-economic status between 1991, 2001 and 2011: What can Census Microdata tell us?
Ethnic inequalities in health, although widely observed, are not fully understood. In the context of increasing ethnic diversity, the pathways by which these disparities are maintained, widened or even narrowed must be further examined. Theories of selective sorting between area-types and social classes may help explain changing health gradients in England. More importantly, as opportunities and propensities for either internal migration or social mobility vary between ethnic groups, theories of selective sorting may be revealing as to changing ethnic health gradients.
Cross-sectional and longitudinal data from the Census illuminate different aspects of this picture. First, through analysing the cross-sectional Samples of Anonymised Records (SARs) from 1991, 2001 and 2011 it is possible to establish whether propensity to migrate varies between ethnic groups according to health and wider socio-demographic information and model whether health is differently explained by migrant status, distance of move and socio-demographic attributes for different ethnic groups. This indicates whether selective sorting a) can operate and b) varies between ethnic groups.
Second, by analysing the Office for National Statistics Longitudinal Study (LS) it is possible to how transitions between deprivation quintiles and social classes for movers (internal migrants) and stayers (non-migrants) influences health gradients between ethnic groups. This reveals how the movement of differently healthy groups between area types and social classes influences health gradients and crucially, whether this influence varies between ethnic groups.
Combining results from these two sources of census data help illustrates how selective sorting may vary between ethnic groups owing to different propensities for migration or contrasting socioeconomic experiences between ethnic groups, and how this can differently influence overall and ethnic-specific health gradients.
Zhiqiang Feng and Dawn Everington, University of St Andrews, and Kevin Ralston and Chris Dibben, University of Edinburgh: A longitudinal analysis of health effects of NEET experiences in Scotland, 2001-2011
This paper investigates whether experiences of young people who are not in employment, education or training (NEET) are associated with adverse long-term outcomes in health. We used the Scottish Longitudinal Study (SLS), which collates information from the 1991, 2001, and 2011 censuses as well as from vital events, for a 5.3% representative sample of the Scottish population. Linked health data such as hospital admissions and prescribing in general practice are also available. We followed around 10,000 young people who were aged 16-19 in 2001 up to 2011. We explored whether NEET young people in 2001 displayed higher risks of poor physical and mental health in the follow-up period. Poor physical health is measured by less than good health, and limiting long term illness from the 2011 census and poor mental health is measured by prescription of anti-depressant and anti-anxiety medicine. We used descriptive and modelling approaches in our analysis.
Covariates include a number of individual socioeconomic characteristics and local area characteristics in the models. Our research found that around 6% of the cohort members have reported less than good health, and 7% reported limiting long term illness, while around 30% have been prescribed with anti-depressant and anti-anxiety drugs. The NEET status in 2001 appears to be associated with poor general health and limiting long term illness: young NEET people were over 50% more likely to report poor health and limiting long term illness. Also the NEET experiences are associated with poor mental health with the odds of poor mental health is over 60% higher among NEET people than the odds among non-NEET people. The effect of NEET experiences appears to be consistent for men and for women. Policy intervention is necessary in assisting NEET young people to re-engage in education or employment.
Richard Wiseman, UK Data Service Census Support: Accessing UK census aggregate data using InFuse
The Economic and Social Research Council (ESRC) has funded UK Data Service Census Support and its predecessor Census Programme for over twenty years to develop and deliver innovative online services to provide access to outputs from current and historic UK censuses to the UK academic community.
This presentation focuses on the development of InFuse, our interface for accessing UK census aggregate data. We have created a single, integrated, standards-compliant dataset from the thousands of separate tables produced by the three UK census agencies. This allows users to select data across the UK in a single interface, and allows users to retrieve data by topic instead of by table. We will discuss our current development, as well as describing future plans, challenges and opportunities.
Geodemographic classifications provide summary indicators of the social, economic, demographic and built characteristics of small areas. Within the UK, there is a lineage of freely available geodemographic classifications covering small to larger geographies that have been built from census data outputs. The 2011 Output Area Classification (2011 OAC) is the most recent example of this. It provides an update to the successful 2001 OAC methodology, and summarises the social and physical structure of neighbourhoods using data from the 2011 UK Census.
The processes used to build the 2011 OAC can be summarised as the selection of an initial set of input data, standardisation, refining of input data based upon correlation and sensitivity analysis and finally, k-means clustering to form the geodemographic classification. These finalised clusters were then described using a range of illustrative materials. The key methodological advancements focused on testing an enhanced range of data manipulation methods prior to clustering and the use of open source software. Different data handling procedures were tested to assess the impact they had on cluster solutions; while the majority of operations were performed in the statistical programming language R, with all the code used being made freely available online.
This updated methodological approach was guided based on engagement with users, including a stakeholder consultation exercise, carried out at UCL in collaboration with ONS. These results helped to inform the classification methodology and identify what the key user requirements for an updated OAC were. The final 2011 OAC consisted of a three-tiered hierarchy, with the UK population being divided into one of 8 Supergroups, 26 Groups and 76 Subgroups. These clusters provide a clear and easy way of interpreting the socio-demographics of the UK.
Recent UK census outputs have been produced on a usual residence basis with each individual being represented at their usual place of residence, and this remains the focus for the majority of 2011 published tables using output areas which are themselves explicitly structured around residential locations. However, the 2011 ONS census has also introduced a range of statistics for workplace populations (individuals at their usual places of work) using a new geography of workplace zones, in addition to key information for other population bases such as students at their home addresses.
A conventional difficulty with spatial models of population data has been that the restriction of most statistics to the residential population base has limited the mapping and spatial modelling of population to applications for which the residential location of population is meaningful. This may be entirely appropriate, for example, when planning primary school catchment areas or analysing residential segregation, but is likely to be of very limited value for estimating populations for planning of daytime services or emergency response.
The ESRC-funded Population24/7 project (2009-11, www.esrc.ac.uk/my-esrc/grants/RES-062-23-1811/read) began to develop ways of estimating time-specific population distributions using a variety of data sources to augment the census residential population with information about daytime population activities, including development of the SurfaceBuilder247 modelling tool. Initial applications have explored population exposure to flood hazard at different times but were restricted to 2001 census data and various intercensal estimates.
This presentation introduces the first major implementation of the approach using 2011 ONS census data, demonstrating how the new census outputs, including the new workplace datasets and open lookup tables, can be combined to produce a much richer series of time-specific population estimates for small areas.
In this paper we respond to Kapoor’s (2013) concern that debates about ethnic segregation in the UK have paid too little attention to the interaction of ethnic and socio-economic segregation. Following the publication of the 2001 and now 2011 Census data, considerable discussion has been given to patterns of ethnic segregation within the UK. The evidence contributes to debates about integration, however it also risks promoting the idea that what we measure is voluntary segregation, arising from the outcome of residential choices and a preference to live with one's ethno-cultural peers. In reality, ethnic and social segregation overlap and are easily confounded. In this paper we use an area typology to assess whether minority ethnic groups are disproportionately concentrated in neighbourhoods of greatest economic disadvantage in England and Wales, and evaluate how those concentrations have changed between 1991 and 2011. We consider the isolation of the ethnic groups from both the White British and from each other, and identify the groups affected by the persistence of economic disadvantage. The analysis shows that patterns of ethnic segregation intersect strongly with patterns of socio-economic segregation, with inequalities in the labour market suggested as a strong contributing factor.
Births to non-UK born mothers have played an important contributory role in the recent increase in fertility in the UK (Tromans et al. 2009; ONS, 2013). Total fertility rates (TFRs) among overseas born women tend to much higher than those of UK-born women (and often higher than TFRs in the country of origin) (Dorman, 2014). However, TFRs calculated for non-UK born women can be potentially misleading since the timing of childbearing is often related to the migration event. Studying the interrelation between the timing of fertility and migration in England and Wales has been difficult because of the hitherto lack of appropriate data sources which contain both information on date of arrival and fertility histories.
The 2011 Census asked non-UK born respondents the month and year of first residence in the UK. By combining these data with linked information on childbearing available within the ONS Longitudinal Study this paper contributes to the literature by estimating the fertility of immigrants before and after migration. The paper provides new insights by (1) estimating the fertility of immigrant groups to England and Wales before migration, (2) examining whether there is evidence of an acceleration of childbearing around the time of the migration and (3) identifying which migrants are likely to have a birth within the first five years after migration to England and Wales. Our findings suggest important timing changes in childbearing associated with the migration event which mean that standard period measures of fertility among these groups can be a misleading indicator of overall life time fertility. These timing changes appear to differ according to the type of sending country. We discuss these findings in terms of their implications for understanding overall fertility trends in the UK.
Immigration is a key issue both for policy makers and the general public. According to the January 2015 MORI issues index, over a third of people believe immigration is the most important issue facing Britain. Over the past three years ONS and Home Office researchers have worked on an innovative joint programme of census analysis producing new insight into the social and economic characteristics of migrants, and informing the public debate.
In addition to the longstanding question on country of birth, the 2011 Census included important new questions on people’s nationality (according to the passport they held) as well as the length of time migrants had been resident in the UK. The joint programme of work has utilised these questions and identified differences in socio-economic outcomes, such as employment and housing tenure, by both length of residence in the UK and country of origin. Additionally the new question permits analysis of the degree to which migrants from different countries acquire British citizenship.
This cross government research programme utilised the specialist knowledge of both ONS and HO analysts and resulted in four published outputs which received media attention from newspapers such as The Guardian and The Independent, as well as The Express and The Daily Mail. Demonstrating the success of the programme of work in generating and communicating clear messages about migrant groups, and the relevance of these messages to the public debate on migration.
This session will discuss the key findings and presentation of this series of analysis. Such findings are key to both the public debate and policy development, and anyone with an interest in migration will be interested to learn more about these important findings and the innovative cross government work that produced them.
The 2011 Census of Population included new questions on language ability, passports held, length of time in the UK, short-term migration and national identity. These provide an alternative perspective to the established country of birth and ethnic group questions for examining the ethnic and cultural diversity of localities and for measuring social and economic exclusion.
This paper investigates the associations between migration, ethnicity, socio-economic status and national identity. The Scottish referendum and the rise of UKIP have shown that economic marginalisation may be associated with some forms of nationalism. Variations in types of national identification across the UK will be related to socio-economic structure, mobility, ethnic composition and measures of social and economic integration. The paper will also explore how these factors may have influenced spatial patterns of voting in the 2015 general election.