open menu

Northern Ireland: approved projects

Public policy and ‘peace’ walls in Belfast: developing baseline indicators

Peace walls are physical structures separating communities in Northern Ireland in interface areas. The first was constructed 1969 and despite the more recent successes of the peace process, such segregation still forms part of the daily reality for large parts of Northern Ireland. About 100 of these walls and barriers remain in place (though estimates on this number vary), and they have come to symbolise the ongoing gulf between the aspirations of the peace process and the implementation of peace in practice. It is against this backdrop that our project takes place.

The rationale for the project has been driven by the Northern Ireland Executive’s recently published community relations strategy document, Together: Building a United Community (May 2013), which sets itself an ambitious target of removing all peace walls by 2023. However, given that previous research has shown that 69% of residents living in close proximity to these walls believe they are still necessary, much work would need to be undertaken with these communities for the NI Executive to be able to achieve its policy objective.

The thrust of this research project is to produce a report (and subsequent academic papers and presentations) that will highlight the socio-demographic and socio-economic characteristics of the areas designated as proximate to the peace walls. This will inevitably entail comparison with less deprived areas, thus adding to the more general debate on social inequalities. We will establish a series of baseline indicators related to those living in closest proximity to the walls, engage with senior policy-makers and practitioners connected to the NI Executive's Peace Walls Implementation Strategy and other stakeholders who have been tasked with progressing the implementation of the peace walls initiative over the next 10 years. We will also lead a series of roundtable workshops with academics and senior civil servants, establish a database which brings together relevant data on interface areas, and produce a number of evidence-based policy briefs designed to encourage wider participation and consultation in policy development.

Project Team:

Cathy Gormley-Heenan (Ulster University, Principal investigator)

Jonny Byrne (Ulster University, co-investigator)

Duncan Morrow (Ulster University, co-investigator)

Brendan Sturgeon (Ulster University, co-investigator)

Michael Rosato (Ulster University, co-investigator)

Sally Cook (Ulster University, co-investigator)


Factors associated with decreased representation in higher education (completed)

This project has been completed. Read the lay summary here.

Widening participation in higher education by students from those groups which are currently under-represented, in particular students from disadvantaged backgrounds and those with disabilities and learning difficulties, is a key strategic goal for the Department for Employment and Learning (DEL). Accurate information about the nature and scale of such under-representation within higher education and understanding the factors associated with participation in higher education (HE) is essential in devising strategies and action plans to bring about significant change.

This project, a collaboration between researchers in DEL and the Administrative Data Research Centre Northern Ireland, aims to provide evidence to better inform the identification of groups at risk, to identify patterns of disadvantage when applying to higher education and to provide potential explanations for under-representation of certain groups. It provides longitudinal information, linking information from the 2011 Census and student enrolment data from the Higher Education Statistics Agency (HESA) and will look at the impact of a range of individual, household and area information related to participation in HE.  

Multi-level modelling will be used to estimate the effects of a range of socio-demographic factors measured at the individual, household and area levels on the probability of initial enrolment of Northern Ireland residents for an undergraduate degree at a UK HE institution. Estimates of the representation of various groups of policy interest in the first year undergraduate cohort will also be derived.

Findings from the project will be used to inform departmental policies that aim to widen participation in HE by providing additional support for prospective students from under-represented groups. The scope of the project will be outlined and progress reported.

Project team:

Michael MacNeill (Department of Employment and Learning)

David Wright (Administrative Data Research Centre Northern Ireland, Queen's University Belfast)

Stephen Donnelly (Department of Employment and Learning)

David Patton (Department of Employment and Learning)

Dermot O'Reilly (Administrative Data Research Centre Northern Ireland, Queen's University Belfast)

Laura Smyth (Department of Employment and Learning)


Sociodemographic characteristics, educational attainment and self-reported health status of farmers in Northern Ireland

This research, led by members of the Department of Agriculture and Rural Development (DARD), will use the Administrative Data Research Network to access linked data from Northern Ireland’s Agricultural Census 2010 and Population Census 2011 to provide a range of social and demographic information on the farming community. The study aims to provide information on the characteristics of farmers in Northern Ireland in order to promote equality and good relations in the administration of grants, subsidies and schemes for the farming community. Of particular interest is self-reported health status and educational attainment of farmers. Models will be developed to investigate associations between these variables and farm business size, farm activity type, land type, remoteness and socio-demographic profiles of farmers.

The findings of this study will be used to inform government policy and will assist in resource targeting and outcome monitoring of both existing and future DARD policies and programmes. Should particular groups or areas be identified in which farmers have poorer health (over and above what would be expected based on demographics and other factors), recommendations will be made to target several current assistance programmes (e.g. Farm Family Health Checks Programme) towards these areas and groups. Similarly, information on the educational attainment, economic status and working hours of farmers will provide evidence of need, establish robust baseline information and aid in targeting and monitoring projects aimed at raising skills, encouraging diversification, business creation and development in the farming community. The scope of the project will be outlined and progress reported.

Project Team:

Patricia McDowell (Department of Agriculture and Rural Development)

Conor McCormack (Department of Agriculture and Rural Development)

David Wright (Administrative Data Research Centre Northern Ireland, Queen's University Belfast)

Paul Caskie (Department of Agriculture and Rural Development)


Using data linkage to obtain accurate population estimates of migrants and their needs for and use of mental health care in Northern Ireland

Northern Ireland (NI) has experienced increased levels of migration in recent years, yet it is still difficult to gain an accurate and comprehensive account of the health of migrants. The recent Racial Equality Strategy for NI highlighted a gap in research on migrants despite international research which shows that migrants have typically poorer physical and mental health than the settled population. This study aims to assess the strengths and caveats of different data sources in accurately identifying and quantifying migrants and their characteristics, and to identify migrants’ use and need for mental health care services.

The study uses a cross-sectional design with data linkage drawing on the 2011 NI Census and health service data from the Business Service Organisation. The largest migrant groups (Polish, other eastern European,  Chinese, south Asian, American/Canadian and other) will be compared to the settled majority by linking data on psychotropic medication (antidepressants; hypnotics and anxiolytics; drugs used in psychoses and related disorders) with data on individual migrant and migration factors, socio-economic factors, and neighbourhood characteristics including migrant population density, neighbourhood level of racist hate crimes as well as data on religious population density and sectarian hate crimes related to the specific context of NI. Furthermore, remaining unlinked data will be used to identify the characteristics of those migrants who did not respond to the Census or who did not register with primary care.

Findings will be helpful to understand the quality and accuracy of available datasets in researching migrant populations in NI; to compare the mental health status of the largest migrant groups to that of the majority population; to identify migrants’ access and use of primary mental health care; to inform policy on migrant and ethnic minority equality; and to inform and support the activities of migrant community-based organisations.

Want to know more? Check out this short video and impact case study (pdf).

Project team:

Tania J. Bosqui (Administrative Data Research Centre Northern Ireland, Queen's University Belfast)

Anne Kouvonen (Administrative Data Research Centre Northern Ireland, Queen's University Belfast)

Michael Donnelly (Queen's University Belfast)

Dermot O’Reilly (Administrative Data Research Centre Northern Ireland, Queen's University Belfast)

John Moriarty (Administrative Data Research Centre Northern Ireland, Queen's University Belfast)

Ciara Close (Administrative Data Research Centre Northern Ireland, Queen's University Belfast)


Key predictors of education attainment in Northern Ireland

This project explores which factors are most important for educational attainment. These factors include characteristics of individual pupils, such as their gender or age relative to their classmates, their family context and their family’s history in education and the wider context of the school they attend and the area where they live and attend school.

This is the first research study to link data from the School Census and School Leavers’ Survey to Northern Ireland Census records. The study cohort comprises all persons aged 14-16, or in year 10, 11 and 12 of school in 2011 (N = 60,000). The School Leavers’ Survey allows for analysis of key outcomes, including GCE grades, school non-attendance and leaving school without qualifications. Planning of services can be improved through knowledge of the factors which cause different individuals to drop out of school or to under-perform academically. By better understanding these causes, policy makers can better improve and better target messages aimed at encouraging greater engagement with education.

Approved: February 2016

Data currently in preparation

Project Team:

John Moriarty (Administrative Data Research Centre Northern Ireland, Queen's University Belfast)

David Wright (Administrative Data Research Centre Northern Ireland, Queen's University Belfast)

Dermot O’Reilly (Administrative Data Research Centre Northern Ireland, Queen's University Belfast)

Allen Thurston (Queen's University Belfast)


Past and current same sex couples in Northern Ireland: an exploratory analysis

Currently in Northern Ireland same sex couples who live together in a committed relationship have no recognised legal status. In the instance that they have registered their civil partnership in another part of the UK, many rights and responsibilities may then be accessed in N.I. but areas falling under the category of ‘transferred’ such as social security and financial support remain inaccessible. This has resulted in substantial disparity in the rights of same sex couples in N.I. compared to both same sex couples in other parts of the UK, and also married couples. Unfortunately, this is not the only area where persons of a same sex orientation face inequality; LGBT (lesbian, gay, bisexual and transgendered) persons are likely to experience discrimination and social exclusion in all areas including education, occupation, and health.

Despite the dearth of research literature pertaining to this population in N.I. some concerning trends are evident. Same sex orientated young people tend to report higher incidence of school bullying, homophobic bullying, reduced academic achievement, and social isolation at school, which often resulted in early school leaving. Obviously, where there is reduced academic achievement and early school departures there will be negative consequences in relation to adult occupational choices, and thus income. Living with harassment, discrimination, and social stigma, plus threats of, and actual violence to, both the person and their home is likely to have negative implications for one’s health, with this pattern evidenced in the available N.I. literature. Mental health issues are an area of major concern for this sample, with substance abuse, eating disorders, and depression common. It has also been shown that same sex and transgendered people have higher than average rates of attempted and completed suicide and self-harm.

This project may help to reduce biases, prejudices and harassment in addition to informing interested parties in the voluntary and public sector with an overall aim of addressing these issues on behalf of this subsample of the population. This project will use data from the 2011 Census to establish the sample and their profile. Tracing the sample retrospectively will permit analyses of the 2001 Census data and allow for detailed examination of the transitions occurring over the intervening ten year period. The main aim of the study is to establish a comprehensive profile of those living or having lived in a same sex relationship in Northern Ireland. Outcomes of interest include socioeconomic and socio-demographic outcomes from census records, education, family structure, age structure, economic activity, employment and real level indicators of deprivation (NIMDM); health, self-reported health from the Census data and mortality rates and causes, all-cause mortality, external causes, suicides, avoidable causes (preventable, amenable) from GRO death statistics.

Project team:

Ulster University


Insights from addressing data quality and under-coverage in administrative sources

The census of the United Kingdom (UK), taken every ten years, provides a wealth of information on the size and attributes of the population; among the detailed demographic data collected is a record of the address of usual residents at that time. This presents a rare and valuable opportunity to assess the accuracy of address data in key administrative sources, particularly where this information is drawn upon for specific operational or statistical purposes. In Northern Ireland (NI), address information from the health card registration system (HCRS) informs the estimation of internal migration, which is an important component of the sub-national population estimates produced annually by the NI Statistics and Research Agency (NISRA). Looking forward, administrative data sources are likely to have a key role as national statistical institutes (NSIs) such as NISRA adopt alternative data collection methods that are less costly and reduce respondent burden. Indeed, for the next census in 2021, the NSIs of the UK constituent countries have committed to using administrative data to support various elements of the operation such as informing the address register and improving the quality of the population estimates. It is therefore important to assess the accuracy of address information in administrative data sources such as the HCRS to ensure that the official statistics they inform are reliable.

In addition, analysis of under-coverage in the HCRS, i.e. those not registered for a health card but resident in NI according to the census, can provide an insight on uptake of primary health care by minority ethnic groups (MEGs). It is important to investigate whether structural barriers (e.g. socio-economic deprivation, language difficulties) are associated with non-registration for a health card within the aforementioned groups. Using data from the March 2011 Census of NI and spring 2011 HCRS, the broad aims of the project are to (i) assess the accuracy of address information in the HCRS and (ii) investigate under-coverage in the HCRS to determine the level of registration among MEGs to access primary health care services in NI. The research will provide a thorough quality assessment of address data accuracy in the HCRS, which will be of interest to NSIs in terms of the growing use of administrative data to inform official statistics. Furthermore, the findings will provide evidence on the extent of engagement by MEGs with primary health care services, which is of academic, policy and practitioner interest.

 

Project team:

Ian Shuttleworth (Administrative Data Research Centre Northern Ireland, Queen’s University Belfast)

Dermot O’Reilly (Administrative Data Research Centre Northern Ireland, Queen's University Belfast)

Brian Foley (Administrative Data Research Centre Northern Ireland, Queen’s University Belfast)

Dave Martin (Administrative Data Research Centre England, University of Southampton)

Chris Gale (Administrative Data Research Centre England, University of Southampton)


ADRC-NI Affiliated Projects 

Research projects undertaken through the ADRCs are normally approved by the ADRN Approvals Panel, which considers applications from across the UK for research using the ADRN. Some projects, which further the objectives of the Network, fall outside of the standard project definition since they have no need to use individual-level data (whether single data sets or linked). Projects that fall into this category within the ADRC-NI go through an approvals process at ADRC-NI level.

Public attitudes to data sharing in Northern Ireland

The overall aim of this project is to gather baseline information on public attitudes to data sharing across Northern Ireland.  We will include a module of questions within the long established Northern Ireland Life and Times survey 2015 and include questions on: trust in organisations; attitudes to data sharing for operational reasons/service delivery; attitudes to data sharing for research; attitudes to data sharing for commercial gain; and the role of the ADRN. The data will be used to inform the public engagement strategy for ADRC-NI and will feed into wider academic and public debate about public attitudes on this issue.

Lead researchers: Professor Gillian Robinson, Professor Helen Dolk and Elizabeth Nelson Gorman


Methodology projects

As part of our core ESRC investment ADRC-NI has within its research portfolio three record linkage methodology projects.   Based within Queens University Belfast’s School of Computer Science, led by Dr Jun Hong and Professor Weiru Liu, the aim of these projects is to contribute to the development of new applications in record linkage. ‘Record linkage’ is the term used by statisticians and epidemiologists, among others, to describe the process of joining records from one data source with another; it is the process of identifying records that refer to the same real-world entity.

(1) Self-learning classifier ensemble for unsupervised record linkage. PDRA: Dr Anna Jurek

In this project we propose a new approach to unsupervised record linkage based on a combination of ensemble learning and automatic self learning. The proposed approach improves automatic self learning by incorporating feature selection and ensemble learning into the learning process. Feature selection helps improve the automatic selection of seeds in the initial training dataset while ensemble learning allows a combination of different similarity metrics to be used, which alleviates the challenge of having to select the most suitable one. Our experimental results show that the proposed approach can achieve a level of accuracy comparable to that of the supervised approaches.

(2) Privacy-preserving record linkage in the presence of missing values. PDRA: Dr Ethan Chi

In this project, we propose a novel approach to privacy preserving record linkage when records contain missing values. The problem of missing values in records is handled by employing the k-NN clustering, in which the missing value of a field in a record is estimated by the values of the corresponding fields of its k-NNs, along with a weight vector in proportion to the distances between the record with the missing value and its k-NNs. To preserve privacy in the record linkage process, an adapted Bloom filter approach is used, which works with a list of estimated values and the associated weight vector. We have carried out the experimental evaluation of our proposed approach using three large real-world datasets. Our experimental results show the effectiveness and efficiency of our approach.

(3) Development of automatic blocking techniques for large-scale record linkage. PhD student: Kevin O'Hare

Record Linkage is a process in which records that refer to the same real-word entity are matched. In the case of both databases containing the same type of unique identifiers i.e. social security number, this is a very straightforward and quick process. However it is often the case that both databases don’t share a common unique identifier or indeed many of the same attributes at all, and instead need to approximate similarity from what attributes they do have in common (i.e. first name, surname, post code, date of birth) and decide if two records are matches or not based upon said similarity. This comparison of records requires an incredible amount of computational time, effort and cost to the interested party, two relatively small databases of 5,000 records each would require 25,000,000 record comparisons if being matched naively.  Therefore any clever innovations that would reduce this workload or speed up each step are incredibly valuable to the Record Linkage process. A pre-comparison step that dramatically increases the efficiency of Record Linkage is that of Blocking.  Put simply Blocking only allows for comparison of records with a shared commonality by placing them into blocks to be later compared rather than always checking all against all as described in the naïve case above. A very simple example of Blocking would be that of a record for a male only being compared to those records of males from the other database. This step alone reduces the workload for this comparison to approximately half of what it would have been. With large datasets more intuitive blocking constraints are needed as a simple blocking such as by gender may still leave far too many in the blocks for the comparison stage to be completed efficiently. Applying more intuitive blocking criteria though usually requires a user with domain knowledge to come up with and apply such criteria in an ad hoc fashion. This may work for a case by case example but the time, effort and availability of a person with such knowledge may not always be possible. This research aims to create a system that can automatically create optimum blocking criteria.


Page last updated: 19/06/2017