The Lancaster-Warwick-Stirling node: Developing statistical modelling in the social sciences Phase 2

Lead Research Organisation: Lancaster University
Department Name: Mathematics and Statistics

Abstract

Abstracts are not currently available in GtR for all funded research. This is normally because the abstract was not required at the time of proposal submission, but may be because it included sensitive information such as personal details.
 
Description a) Advancing methodological understanding and practice.

We provided this primarily through our short courses and workshops, but also through careful targeting of conference presentations to pitch methodological material at the correct level In general, our short course programme targeting a wide range of different levels of activity, while avoiding clashes with other nodes.
For each year of the node, we provided at least eight short courses: Generalised Linear Models, Introduction to R, Data mining techniques, Getting started in SPSS, Intermediate SPSS, Structural equations modelling, Missing data for social research, Introduction to STATA and Longitudinal data analysis. For the last three years we repeated the structural equations course as it was so popular.
The masterclasses, workshops and conference activities are all listed in Annex A. In general, masterclasses were aimed at more advanced users, whereas workshops were aimed at bringing practitioners and academics together to discuss new methodological advances.
A further route to methodological understanding was through the publication of books . We highlight two books published as a result of our node's work Understanding Criminal Careers (Soothill, Fitzpatrick and Francis, 2009) and Multivariate Generalized Linear Mixed Models Using R (Berridge and Crouchley,2011)
The first book presents the ideas of criminal career research to a criminological audience, targeting primarily British criminologists who need to understand this quantitative approach to criminal histories. It has been adopted as a course text for undergraduate courses in Criminal Careers both in the UK (eg Northumbria University) and in the US.
The second book presents the theory and practice of using R to fit multivariate generalized linear mixed models (MGLMMs). The book applies the principles of modelling to longitudinal data taken from panel studies using the SABRE software within R. The book presents robust and methodologically sophisticated models for analyzing large and complex datasets. It highlights the application of the trivariate ordered response model with correlated random effects to BHPS data on attitudes to gender roles. The book has been very well received with sales of almost 500 copies to date, almost a half of which have been sold in North America

b) Enhancing the UK international profile in methodological excellence and ensuring that the UK is at the forefront of international developments in social research methodology.

A strong feature of our node was that we recognised the need to enhance the international profile of our work and to disseminate our methodological research in social science internationally. We collaborated with both US (Alex Piquero, Scott Zeger, Danielle Harris) and European (Paul Nieuwbeerta, Reinhold Hatzinger, Regina Dittrich, Arjan Blokland, Francesco Bartolucci, Mogens Christoffersen) colleagues, building up methodological networks of researchers in particular substantive fields. Dissemination of our work to international conferences was an important aspect of our work for this aim, as was targeting the most important international journals.
We attended and presented at both international statistical meetings (Statistical Modelling workshop, CSDA European meetings, Italian meetings on latent variables) and also substantive international meetings on discipline-related methodology (Criminology: American Society of Criminology, European Society of Criminology, International Society of Criminology, :Sociology: RC33 meetings in Naples and Sydney; Psychology: European Society for Developmental Psychology; Society for Research in Child Development; Psychometrics Society meeting in Cambridge) We have produced a large number of papers in international journals such as the Annals of Appiled Statistics, Journal of the Royal Statistical Society A.

c) Enable members of the UK social science community to learn about recent developments in research methods

Both conference presentations and workshops to target busy researchers played an important role here. We believed that it was important to embed methodological development into substantive talks, so that different audiences of researchers are exposed to new methodology. Conference presentations were given to the British Society of Criminology, the British Psychological Society, the British Sociological Association and the Royal Statistical Society. We also organised two focused workshops to disseminate our work, both held in London at the Royal Statistical Society, and invited other leading researchers to give masterclasses in recent quantitative developments (see Annex A)

d) Undertake research and training activities relevant to research users and of benefit to the economy and society.

All of our research strands are on highly relevant and topical themes. The relevance of our work to the economy and society is evidenced by the impact of our work and the future impact new projects arising out of our node's work. One method that we used to target researchers was to give targeted mini-meetings on methodological topics before conferences. Thus, developmental psychologists were targeted in half day meetings before the British Psychological Society Developmental Psychology Section. We specifically targeted practitioner meetings to present our research work for example, for the crime strand, talks were given to the ESDS Government Research Conference and the Government Statistical Service methodology conference; in education research , to the Harrow Children & Young Peoples' Service and Hertfordshire County Council Children, Schools and Families Service Meeting, and in Sociology, to the European Survey Research Association.

e) Play a strategic role in the promotion of high quality research methodology.

To assist with this strategic aim, we sought to join committees relevant to the promotion of quantitative methodology for the social sciences. Thus Francis joined the ESRC Research Grants Board in 2005, serving for five years, and also joined the ESRC working group on quantitative methods. He also chaired a commissioning panel for a quantitative research initiative HUMVIB for the European Science Foundation, accepted an invitation to join the Home Office advisory subcommittee on Survey, Design and Statistics(2006-2011), and joined the editorial board of the Journal of Quantitative Criminology. Similarly, Diggle played a similar role at the Medical Research Council, being a Member of Public Health and Health Services Research Board from 2007-8, then a member of Population and Systems Medicine Research Board, 2008-2011 and is now MRC Chair of Strategic Skills Fellowships Schemes Panel.
Exploitation Route There are at least two potential sources of future impact.
a) The work on the statistical modelling of criminal career data has led to the Lancaster group being commissioned to carry out research on the Criminal Careers of Organised Crime by the Home Office. This work is still ongoing, but is likely to led to impact once peer reviewed and published.

b)The work on Latent Class Analysis, combined with the work on generalised nonlinear models, both carried out as part of these two grants are, in combination, being used as part of a new grant from the Australian Research Council as part of the Project "Statistical Modelling of Social Networks" with Murray Aitkin, at Melbourne University. This will lead to a book and new methods of modelling social network data
Sectors Communities and Social Services/Policy,Education,Healthcare,Government, Democracy and Justice

URL http://www.socialstatistics.org/
 
Description There are three specific area of impact which can be reported. a) On the criminology strand, in 2008 comparative work was undertaken on the long term recidivism risk of offenders compared to the risk of offending for those with no offending history. We analysed data from the Home Office's Offenders Index which looked at the yearly risk of reconviction given a single non-violent conviction Two birth cohorts of offenders were examined - 1953 and 1958. This risk was compared with the risk of first conviction for a non-offending group obtained by combining the Offenders Index data with population data and using a discrete time hazard rate analysis. We found that the risk profiles had nearly converged by age 35 in both cohorts. This provided evidence that the retention (or disclosure) of criminal convictions long in the past was of no operational value. This research was used by the Information Commissions in an Information Tribunal appeal case where five Police Authorities were appealing against a decision by the Information Commissioners that the Police authorities should delete old criminal convictions from the Police National Computer. Soothill and Francis appeared as expert witnesses at the tribunal. The case continued and the issues raised led to the appointment of an Independent Advisor for Criminality Information Management (Sunita Mason). The following year, a report by the Jill Dando Institute was produced, providing evidence to the then Labour government on the retention period for DNA profile data for those arrested but not found guilty. Soothill and Francis criticised the methodology of the report pointing out that the correct method to use was that used in the Information Tribunal work. This was referenced in The Guardian. The researchers were then involved with the Home Office in advising on more appropriate methodology, attending a day meeting and peer reviewing the research. This led to a revised research report produced by Home Office researchers and , under the new government, has eventually led to legislation under the Protection of Freedoms Act 2010-12 which now suggests a five year retention period for those arrested but not found guilty. The full description of the impact of this work is an ESRC Impact case study. http://www.esrc.ac.uk/impacts-and-findings/features-casestudies/case-studies/7724/reviewing-the-dna-database.aspx b) Work with Peter Diggle and Ivonne Solis-Trapala worked with Lancashire County council Multi-Agency Data Exchange (MADE), which is a data warehouse tool for datasets that are relevant to crime and disorder in Lancashire. The MADE project was established in 2001 to help people living in Lancashire to make better-informed decisions about community safety issues in their neighbourhood and to create a common collation and dissemination facility which will improve the speed and reliability of multi-agency information exchange for crime and disorder strategies and other multi-agency policies throughout Lancashire. The team modelled the spatio-temporal distribution of crime in Lancashire, developing appropriate statistical methodology. The work was used as a case study on the reporting of spatial crime data by the Home Office advisory committee on surveys and statistics. c) Work with Nadja Reissland on the development of fetal expressions (and the measurement and assessment of such development using longitudinal ordinal models) led to a publication in PLOS-ONE and subsequent substantial media interest in the development of smile faces and cry faces in fetuses. See the comments tab on http://www.plosone.org/article/info:doi/10.1371/journal.pone.0024081 d) Impact of the training materials: The SRME site went live on Restore in December 2011. The SRME-Restore site has 8,996 page views between 01/01/12 and 17/04/12 (data supplied by Restore). A very high proportion are from US (75%), with 12% from UK and remaining from across the globe. The tutorial videos created for the site have been available since 2010 and are continuing to prove popular on Youtube. The eighteen videos produced have been viewed over 165,500 times at time of writing (see http://www.youtube.com/user/StatRegMethEdu). The chi-square analysis video has proved particularly popular, having been viewed over 61,800 times, while the three simple linear regression have been viewed over 23,000 times, logistic regression 18,000 views and Multiple linear regression with nearly 10,000 views. Comments on these videos from students have been extremely positive, indicating that the resources are meeting a real need.
First Year Of Impact 2011
Sector Education,Government, Democracy and Justice
Impact Types Societal

 
Description A likelihood based method for analysing overdispersed correlated count data with subject specific covariates 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Seminar to the Department of Statistics at Warwick

none
Year(s) Of Engagement Activity 2012
 
Description Criminal career specialisation for female offenders : a latent Markov approach 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Seminar presentation at the Universidade Federal Fluminense in Rio de Janeiro

video produced of talk
Year(s) Of Engagement Activity 2011
 
Description Empirical modelling of spatio-temporal variation in meningitis incidence 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Presented at 3rd Meningitis Environmental Risk Information Technologies technical meeting and national workshop

none
Year(s) Of Engagement Activity 2010
 
Description Executive function as social interaction 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Presented at workshop titled 'Lifespan Development of Executive Control', May 7-9, 2009 - Saarland University, Saarbrücken

none
Year(s) Of Engagement Activity 2009
 
Description Graphical approaches in the modelling of cognitive development 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Presented at ESRC seminar "Using the Microgenetic Method to Investigate Cognitive Development",

none
Year(s) Of Engagement Activity 2009
 
Description How useful are microclasses? : an analysis of detailed parental occupational differences and their effects on filial school attainment in Britain 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Presented at Social Stratification Research Seminar 2011: Modelling Patterns of Social Stratification

none
Year(s) Of Engagement Activity 2011
 
Description HumVIB (cross-national and multi-level analysis of human values, institutions and behaviours) EUROCORES programme panel 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Presented at 1st Workshop of the ESF MOForum on Peer Review I

nfluence on peere review process within ESF.
Year(s) Of Engagement Activity 2010
 
Description Innovative methods in visual sociology 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Presented at Qualitative Innovations in CAQDAS [QUIC]- Geo-Referencing Seminar

none
Year(s) Of Engagement Activity 2011
 
Description Is Domestic violence the exception to declining crime rates? conference presentation to American society of criminology, November 2014 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Talk generated a lot of questions and discussion

plans to write a research paper
Year(s) Of Engagement Activity 2014
 
Description Joint modelling of longitudinal and survival data : a dimple transformed Gaussian approach 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact 4th Armitage Lecture, U.K. Medical Research Council (MRC) Biostatistics Unit, Cambridge

a prestigious lecture
Year(s) Of Engagement Activity 2012
 
Description Latent transition analysis and offending lifestyle specialisation in England and Wales 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact Part of Statistics Norway (Statistisk sentralbyrå) invited seminar series

none
Year(s) Of Engagement Activity 2012
 
Description Longitudinal models for assessing cognitive development in early childhood 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Presented at Seminar to the Department of Economics, Finance and Statistics, University of Perugia Perugia, Italy, December, 2007

none
Year(s) Of Engagement Activity 2007
 
Description Model-based geostatistics with environmental and epidemiological applications 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Postgraduate students
Results and Impact Pre-conference workshop of the 8th International Conference on Geostatistics for Environmental Applications

none
Year(s) Of Engagement Activity 2008
 
Description Modelling discrete data in longitudinal studies 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Postgraduate students
Results and Impact Analysing Developmental Change Workshop

plans made for academic paper
Year(s) Of Engagement Activity 2011
 
Description Modelling gender inequalities : a longitudinal analysis of repeated ordinal data 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Social Stratification Research Seminar - Modelling Patterns of Social Stratification

ddiscussions after the talk
Year(s) Of Engagement Activity 2012
 
Description Modelling individual and group trajectories in longitudinal data analysis 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Presented at Analysing Developmental Change Workshop

short course attached to conference generated lots of questions
Year(s) Of Engagement Activity 2011
 
Description Mr bun the bake : what is the relationship between parental occupations and filial educational attainment? 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Social Surveys and Social Stratification Research Seminar

none
Year(s) Of Engagement Activity 2011
 
Description Quasi-variances and extensions 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Presented at international workshop on statistical modelling 2011

publication in proceedings volume.
Year(s) Of Engagement Activity 2011
 
Description School effects and equity gaps in educational attainment and progress at age 11 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Participants in your research or patient groups
Results and Impact Presented to Department for Education PLASC User Group (PLUG)

Section not completed
Year(s) Of Engagement Activity 2011
URL http://eprints.ncrm.ac.uk/2211/
 
Description Statistical analysis of microdevelopmental data : using latent transition analysis to explore developmental sequences 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact A key issue in the analysis of development concerns how we conceptualize and measure individual trajectories of change. In this talk I will review some recent approaches to this issue, which are not commonly used in developmental psychology. One of these approaches is latent transition analysis. I will describe how this technique can be used to extract different growth trajectories within longitudinal studies and allow other factors to be regressed onto these trajectories. I will discuss its underlying statistical assumptions and describe the main routines for conducting such analyses on commonly used statistical packages.

change in research thinking on how to analyse psychological data.
Year(s) Of Engagement Activity 2011
 
Description Statistical modelling of development of executive function in early childhood 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Presented at International Workshop on Statistical Modelling 2007

none
Year(s) Of Engagement Activity 2007
 
Description Statistical models of executive function test performance in children 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Postgraduate students
Results and Impact Paper at International Seminar for Young Psychologists on Cognitive and Developmental Sciences

none
Year(s) Of Engagement Activity 2011
 
Description What is latent class analysis? 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Participants in your research or patient groups
Results and Impact Presentation to the Social Research Association http://eprints.ncrm.ac.uk/1770/

http://eprints.ncrm.ac.uk/1770/
Year(s) Of Engagement Activity 2011
URL http://eprints.ncrm.ac.uk/1679/1/Belfast10March2010McDonald.pdf
 
Description What is the relationship between parental occupations and filial educational attainment? 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact Given at The Gordon Square Seminar

none
Year(s) Of Engagement Activity 2009
 
Description What makes us human? 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Undergraduate students
Results and Impact Paper at The 5th Edniburgh Lectures Psychology 4 Students Event

audience motivated
Year(s) Of Engagement Activity 2012
 
Description Why are longitudinal studies crucial for developmental psychology? 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact seminar on developmental psychology

none
Year(s) Of Engagement Activity 2009
 
Description Why are longitudinal studies crucial for developmental psychology? 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Presented at Analysing Developmental Change Workshop

none
Year(s) Of Engagement Activity 2011
 
Description Youth transitions in the 21st century - future directions for data, methods and theory 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Paper at Centre for Population Change Seminar

none
Year(s) Of Engagement Activity 2010