Projects

Explore further details on our ongoing projects for a deeper understanding of our current initiatives.

Bayesian Adaptive Survey Design Network (BADEN)

BADEN funded by The Leverhulme Trust gathers researchers from academia and national statistical offices and gives a strong impetus to theory development and practical implementation of adaptive survey designs.

Adaptive survey designs differentiate features for different population subgroups based on auxiliary data about the sample obtained from frame data, registry data, or paradata.

The development of a Bayesian framework will allow the learning and constant updating of key input parameters to these designs. More specifically, our objectives are as follows.

  • To bring together researchers periodically and to speed up theoretical development and practical implementations of adaptive survey designs.
  • To establish a cross-institute research agenda for the main research aim.
  • To design and implement joint simulation studies.
  • To support discussion on theory and the exchange of empirical results.
  • To disseminate work to a larger public to advocate ideas and to get feedback on feasibility and utility.
  • To lay the groundwork for joint papers and other forms of collaboration.
  • To assist implementation of adaptive survey designs.

People

  • Prof Natalie Shlomo - Principal Investigator (PI)
  • Dr Stephanie Coffey - US Census Bureau
  • Dr Gabriele Durrant - University of Southampton
  • Dr Peter Lundquist - Statistics Sweden
  • Mr Daniel Pratt - RTI International, North Carolina
  • Dr Barry Schouten - Statistics Netherlands
  • Dr James Wagner - University of Michigan
  • Ms Rebecca Moore - Network Facilitator

Centre on Dynamics of Ethnicity (CoDE)

CoDE is an interdisciplinary programme of research concerned with understanding changing ethnic inequalities and identities.

Our team has more than 20 academics, many affiliate members, and PhD students working from Glasgow, Oxford, and Manchester. We also have many valued partners.

CoDE utilises a variety of research techniques and tools to ensure that the potential economic and social benefits of our research are realised.

Our focus is on the changes within ethnic groups (their internal structures and formulations of identities) and their external relationships and position in British society. Bringing together sociologists, demographers, historians, geographers, and political scientists, we are researching:

  • how class, gender, generation, age, and place produce different experiences and visions of ethnicity across the UK;
  • how changes in ethnic identities over time were expressed through the emergence of new or mixed identities, as well as the shifting significance of language and religion as a marker of ethnicity;
  • the significance of the context of emigration and arrival in shaping ethnic identities and the long-term trajectories of migrants in British society;
  • how major social changes in Britain’s economic and political structures have impacted the ethnic inequalities experienced in employment and politics today.

Family Capital and Pathways to Inclusion

Using a longitudinal outlook, the project investigates the outcomes and trajectories of ethnic minority inclusion within schools; the labour market; and civic and political life in Britain and Canada, as well as the role that 'family capital' plays in determining these inclusion patterns.

Social cohesion is perceived as an important social goal within academic and policy circles.

A crucial precondition to this goal is social inclusion, i.e. ensuring that individuals become full members of society by accessing societal resources and institutions.

In recent decades, fast-growing ethnic diversity has led to intensifying policy and academic debates on how this is affecting processes of social inclusion and cohesion.

Examining the factors influencing ethnic inclusion has thus increased in importance.

The project contributes to this debate by exploring the dynamics of ethnic inclusion into the economic and civic-political spheres and the interplay of these two separate but closely connected spheres.

It also investigates how these lifelong processes are shaped by family capital, i.e. a broad and complex range of cultural, social, economic, and political influences that are created within the family context. More specifically, the research addresses four core questions.

  1. What are the pathways to socio-economic and civic-political inclusion of ethnic minorities into British and Canadian institutions?
  2. What role does family capital (financial, human, social, and cultural) play?
  3. What are the best ways to address the differentials in family influences?
  4. Do different policy environments generate different types of outcomes?

To answer these questions, the study draws on available secondary data sources (including Understanding Society and the Millennium Cohort Study in Britain and the National Longitudinal Survey of Children and Youth and the Survey of Labour and Income Dynamics in Canada) and uses advanced longitudinal and multilevel methods of data analysis.

National Centre for Research Methods (NCRM)

The National Centre for Research Methods (NCRM) was set up at the University of Southampton in 2004. NCRM is tasked with:

  • increasing the quality and range of methodological approaches used by UK social scientists through a programme of training and capacity building;
  • driving forward methodological development and innovation through its research programme.

In 2014, NCRM was recommissioned and Southampton formed a consortium with two further institutions with international reputations in methodological research and training in the social sciences: the universities of Manchester and Edinburgh.

The University of Manchester has longstanding experience in delivering high-quality methodological research and training through the Cathie Marsh Institute for Social Research (CMI); methods@manchester has hosted two NCRM nodes and the Research Methods Programme.

Our contribution to NCRM is to run advanced short courses and the bursary scheme, and to provide leadership for the autumn school programme and the programme of international visits.

We also have two three-year research projects funded by the centre – one on survey methods and biosocial data and the other on disclosure risk.

Q-Step

Q-Step is a £19.5 million programme designed to promote a step-change in quantitative social science training.

Q-Step was developed as a strategic response to the shortage of quantitatively skilled social science graduates.

It is funded by the Nuffield Foundation, the Economic and Social Research Council (ESRC), and the Higher Education Funding Council for England (HEFCE).

Representative Indicators for Survey Quality (RISQ)

Missing data due to non-response imposes a serious threat to the quality of surveys and register-based statistics.

Non-response is often not a random phenomenon. It usually depends on demographic and socio-economic characteristics of individuals or enterprises. Also, the data collection process may have a substantial influence.

The response rate is often used as an indicator of survey quality. It has the advantage that it can be easily computed. However, low response rates will not necessarily cause estimates to be biased. There are ample examples in the literature where increased data collection efforts have led to a higher response rate but also to a larger non-response bias.

To assess the effects of non-response on the quality of statistics, other quality indicators are needed. These indicators should measure the degree to which respondents and non-respondents differ from each other. In other words, such indicators should measure the degree to which the group of respondents in a survey or register resembles the population. The indicators are called Representativity Indicators or, for short, R-indicators.

It is the objective of RISQ to develop R-indicators, to explore their characteristics and to show how to implement and use them in a practical data collection environment.

The project will demonstrate that R-indicators are not only used in the analysis of survey data but also during fieldwork. They can be used to monitor data collection processes, and therefore facilitate efficient allocation of interviewing resources.

The 7th Framework Programme 

RISQ is financed by the 7th Framework Programme (FP7) of the European Union. It is supported by the "Cooperation Programme". This programme has ten distinct themes. RISQ is part of the theme "Socio-economic sciences and the Humanities", and within this theme of Activity 8.6 ("Socio-economic and scientific indicators"), 8.6.3 ("Provision for underlying official statistics").

Publications

Deliverables

Work Package 2 / Deliverable 1: Schouten, B. (2008), Documentation of datasets

Work Package 3 / Deliverable 2.1: Shlomo, N., Skinner, C., Schouten, B. & Bethlehem, J. (2008), Statistical Properties of R-indicators, Version 2

Work Package 3 / Deliverable 2.2: Shlomo, N., Skinner, C., Schouten, B., Heij, V. de., Bethlehem, J. & Ouwehand, P (2009), Indicators for Representative Response Based on Population Totals

Work Package 4 / Deliverable 3: Schouten, B., Morren, M., Bethlehem, J., Shlomo, N. & Skinner, C. (2009), How to use of R-indicators?

Work Package 5 / Deliverable 4: Shlomo, N., Skinner, C., Schouten, B., Carolina, T. & Morren, M. (2009), Partial Indicators for Representative Response, Version 2

Work Package 6 / Deliverable 5: Loosveldt, G. & Beullens, K.(2009), RISQ - Fieldwork Monitoring

Work Package 7 / Deliverable 7: Fosen, J., Kleven, Ø, Lagerstrøm, B.O., Luiten, A. & Wetzels, W. (2010), Indicators and data collection control, Report from the pilots in Statistics Norway and Statistics Netherlands

Work Package 7 / Deliverable 8.1: Lagerstrøm, B.O., Bjørshol, E.(2010),Q - Fieldwork Monitoring. Indicators and Data Control, Work plan and Preliminary Findings - Pilot Statistics Norway

Work Package 7 / Deliverable 8.2: Luiten, A. & Wetzels, W. (2010),Q - Fieldwork Monitoring. Indicators and Data Control, Work plan and Preliminary Findings - Pilot Statistics Netherlands     

Work Package 8 / Deliverable 9: Schouten, B. & Bethlehem, J.(2009), Representativeness Indicators for Measuring and Enhancing the Composition of Survey Response

Work Package 8 / Deliverable 12.1: Heij, V. de., Schouten, B., Shlomo, N. (2010), RISQ manual, Tools in SAS and R for the computation of R-indicators and partial R-indicators

Work Package 8 / Deliverable 12.2: Bethlehem, J. (2010), R-Cockpit manual, A tool for graphical analysis of representativity

Papers

Schouten, B. & Cobben, F. (2007), R-indexes for the comparison of different fieldwork strategies and data collection modes. Discussion Paper 07002, Statistics Netherlands, Voorburg, The Netherlands        

Cobben, F. & Schouten, B. (2008), An empirical validation of R-indicators. Discussion Paper 08006, Statistics Netherlands, Voorburg, The Netherlands 

Bethlehem, J., Cobben, F. & Schouten, B.(2008), Indicators for the Representativeness of Survey Response. Proceedings of the Statistics Canada Symposium 2008, Gatineau, Canada

Skinner, C., Shlomo, N., Schouten, B., Zhang, L. & Bethlehem, J. (2009), Measuring Survey Quality Through Representativeness Indicators Using Sample and Population Based Information. Paper presented at the NTTS Conference, 18-20 February 2009, Brussels, Belgium

Schouten, B., Shlomo, N. & Skinner, C. (2010), Indictators for Representative Response. Paper presented at Q2010, Helinski, Finland

Beullens, K. & Loosveldt, G.. (2010), R-indicators and Fieldwork Monitoring. Presentation at Q2010, Helinski, Finland

Luiten, A. & Wetzels, W. (2010), Differential Survey Strategies based on R-indicators. Paper presented at Q2010, Helinski, Finland

Schouten, B., Shlomo, N. & Skinner, C. (2011), Indicators for Monitoring and Improving Representativeness of Response. Journal of Official Statistics 27, pp. 1-24

Shlomo, N., Skinner, C. & Schouten, B., (2011), Estimation of an indicator of the Representativeness of Survey Response. Journal of Statistical Planning and Inference 142, pp. 201-211

Schouten, B., Bethlehem, J.G., Beullens, K., Kleven, O, Loosveldt, G., Luiten, A., Rutar, K., Shlomo, N. & Skinner, C. (2012), Evaluating, Comparing, Monitoring, and Improving Representativeness of Survey Response Through R-Indicators and Partial R-Indicators. International Statistical Review 80, pp. 382-399

Luiten, A. & Schouten, B., (2013), Tailored fieldwork design to increase representative household survey response: an experiment in the Survey of Consumer Satisfaction. Journal of the Royal Statistical Society A 176, pp. 169-189

Shlomo, N. & Schouten, B. (2013), Theoretical Properties of Partial Indicators for Representative Response. Technical Report, University of Southampton

Shlomo, N., Schouten, B. & De Heij, V., (2013), Designing Adaptive Survey Designs with R-indicators. Paper presented at the NTTS Conference in Brussels, 2013

Schouten, B. & Cobben, F., Lundquist, P. & Wagner, J. (2015), Theoretical and empirical support for adjustment of nonresponse by design. Discussion Paper 2014|15, Statistics Netherlands

Schouten, B.(2017), Statistical inference based on randomly generated auxiliary variables, Journal of the Royal Statistical Society, Series B

Presentations

Bethlehem, J, Cobben, F. & Schouten, B. (2007), Response Enhancement in Household Surveys. University of Michigan, Ann Arbor.

Bethlehem, J, Cobben, F. & Schouten, B. (2008), The history of the R-indicator. First RISQ Meeting, Statistics Netherlands.

Schouten, B. (2008), Representativity Indicators for Survey Quality - RISQ. First RISQ Meeting, Statistics Netherlands.

Bethlehem, J., Cobben, F. & Schouten, B.(2008), Indicators for the Representativeness of Survey Response. Presentation at the Statistics Canada Symposium 2008, Gatineau, Canada.

Marujo, A. (2009), Representativity Indicators for Measuring Survey Quality. ITACOSM09, Siena, Italy.

Beullens, K. & Loosveldt, G. (2009), R-indicators and Data Monitoring. ESRA Conference, Warsaw, Poland.

Schouten, B., Shlomo, N. & Skinner, C. (2009), Representativeness Indicators for Measuring and Enhancing the Composition of Survey Response. 57th Session of the ISI, Durban, South Africa.

Skinner, C., Shlomo, N., Schouten, B., Zhang, L. & Bethlehem, J. (2009), Measuring Survey Quality Through Representativeness Indicators Using Sample and Population Based Information. Presentation at the NTTS Conference, Brussels, Belgium.  

Schouten, B., Shlomo, N. & Skinner, C. (2010), Indicators for Representative Response. Presentation at Q2010, Helsinki, Finland.

Beullens, K. & Loosveldt, G. (2010), R-indicators and Fieldwork Monitoring. Presentation at Q2010, Helsinki, Finland.

Kleven, Ø., Fosen, J., Lagerstrøm, B. & Zhang, L., (2010), The Use of R-indicators in Responsive Survey Design: Some Norwegian Experiences- Presentation at Q2010, Helsinki, Finland.           

Luiten, A. & Wetzels, W. (2010), Differential Survey Strategies based on R-indicators. Presentation at Q2010, Helsinki, Finland.

Abstracts

Schouten, B. & Bethlehem, J. (2008), Special Topic Session on Quality Indicators. Q2008, Rome, Italy

Schouten, B., Shlomo, N. & Skinner, C. (2009). Representativeness indicators for measuring and enhancing the composition of survey response. Invited Paper, Session IMP53 Nonresponse Bias in Surveys. ISI 2009, Durban, South Africa

Tools

Random generation of auxiliary variables

In Schouten (2015, Discussion paper 2015-15, CBS, Den Haag, The Netherlands) and Schouten (2017, JRSS series B), R-indicators and coefficients of variation (CV) are evaluated under random generation of auxiliary variables. A framework is presented for various variable-generating distributions and the expected amount of explained variation is derived. In the 2017 JRSSB paper, a data set extracted from the CentERdata LISS-panel, see www.lissdata.nl, features as an example. The data set and R code to reproduce the example are available here. The SPSS file contains a codebook.

You can download the following files:

Computation of R-indicators - Version 2.1

An extended version of the RISQ 2 code in SAS and R was released at September 14, 2015. RISQ 2.1 estimates partial CV at the variable-level and at the category level. Furthermore, it approximates standard errors for the overall CV and partial CV. The manual is extended and discusses the new indicators.

You can download the following files:

Computation of R-indicators - Version 2.0

This is the new version of the RISQ code in SAS and R. It was released in February, 2014. RISQ 2 includes standard error approximations for all indicators, and it also has the coefficient of variation. Furthermore, bias adjustment has slightly changed with respect to version 1 of the code. The use of the functions is explained and illustrated in the manual.

You can download the following files:

Computation of R-indicators - Version 1.0

The computation of R-indicators and partial R-indicators are implemented in SAS and R. The code for both SAS and R can be downloaded. The code contains all functions needed to compute:

  • R-indicators
  • Unconditional partial R-indicators at the variable level and the category level
  • Conditional partial R-indicators at the variable level and the category level
  • The use of the functions is explained and illustrated in the manual.

You can download the following files:

The program Cockpit

Cockpit is a software tool that demonstrates the graphical possibilities of survey response analysis with R-indicators.Cockpit can generate in ann interactive way the following plots:

  • Box plots of response probabilities for each category of an auxiliary variables
  • Bar plots of unconditional R-indicators for sets of auxiliary variables
  • Bar plots of unconditional R-indicators for the categories of an auxiliary variable
  • Bar plots of conditional R-indicators for sets of auxiliary variables
  • Bar plots of conditional R-indicators for the categories of an auxiliary variable

Cockpit requires a data file and a metadata file. Such files can be generated from SPSS or Stata files with R scripts that are included in the package. Also included is a demonstration data set with data form a Statistics Netherlands survey.

It should be noted that Cockpit is just a demonstration tool and not a production tool. No bias correction is carried out in the computations of the various R-indicators.

A zip-file can be downloaded. This contains the following files:

  • cockpit.exe (the program itself, non-installation required)
  • cockpit.pdf (the cockpit manual)
  • gps.rin (sample metadata file)
  • gps.dat (sample data file)
  • export-spss.r (R-script to export data and metadata from SPSS to Cockpit)
  • export-stata.r (R-script to export data and metadata from Stata to Cockpit)

Download cockpit

The British Election Study (BES)

The British Election Study (BES) is one of the longest-running election studies worldwide and the longest-running social science survey in the UK.

It has made a major contribution to the understanding of political attitudes and behaviour over nearly sixty years.

Surveys have taken place immediately after every general election since 1964.

The first study conducted by David Butler and Donald Stokes in 1964, transformed the study of electoral behaviour in the UK.

Since then the BES has provided data to help researchers understand changing patterns of party support and election outcomes.

The Social Complexity of Immigration and Diversity (SCID)

Immigration is a major political issue, with increasing media coverage, rising anti-immigration sentiment, and the rise of anti-immigration political parties.

The issue of migration sits centrally within the wider debate about ethnic and religious diversity and its effects on social cohesion.

We are still, though, a long way from understanding these issues and their potential consequences.

They seem to rest on beliefs about national identity and ethnicity, but cannot be divorced from the effects of social class, education, economic competition, and inequality, as well as the influences of geographical and social segregation, social structures, and institutions.

This project will integrate two very different disciplines, social science, and complexity science, to gain a new understanding of these complex, social issues.

It will do this by building a series of computer simulation models of these social processes.

One could think of these as serious versions of the Sims computer games, programmes that track the social interactions between many individuals.

Such simulations allow ‘what if’ experiments to be performed so that a deeper understanding of the possible outcomes for society as a whole can be established based on the interactions of many individuals.