Research environment
The department has a rich and varied research environment and the Social Statistics group has close links to other departments, including Methodology and Social Policy. There are regular social statistics meetings and seminars, as well as workshops and conferences.
Research areas for PhD research
Latent variable models and structural equation models for categorical data
Areas for methodological research include treatment of missing values and drop out; modern estimation methods and testing under complex sample designs; longitudinal data; multi-group data analysis; measurement equivalence; survival analysis; distributional assumptions of latent and observed variables; sensitivity of the model to other model assumptions.
Potential areas of applications
-
Analysis of attitudinal and behavioural data from social surveys such as the British social attitudes, European social survey, Eurobarometer, etc. For examples of applications and methods in the analysis of such data, please see the pages of the LCAT research project hosted by the department
-
Educational testing, behavioural sciences including psychometrics
-
Epidemiology and public health
Further reading
Vasdekis, V., Cagnone, S. and Moustaki, I. (2012). A Composite likelihood inference in latent variable models for ordinal longitudinal responses. Psychometrika, 77 (3), pp. 425-441. ISSN 0033-3123
Katsikatsou, M., Moustaki, I., Yang-Wallentin, F. and Jöreskog, K. (2012). Pairwise Likelihood Estimation for factor analysis models with ordinal data. Computational Statistics and Data Analysis. 56 (12), pp. 4243–4258. ISSN 0167-9473
Modelling response strategies and detection of outliers in educational and behavioural sciences
Areas for methodological research include aberrant response patterns; detection of outliers; modelling the response strategies being employed by respondents; switching response strategies due to time or other constraints.
Potential areas of applications
-
Educational and behavioural sciences
Further reading
Mavridis, D. and Moustaki, I. (2009). The forward search algorithm for detecting aberrant response patterns in factor analysis for binary data. Journal of Computational and Graphical Statistics. 18 (4), pp. 1016-1034. ISSN 1061-8600
Moustaki, I. and Knott, M. (2013). Latent variable models that account for atypical responses. Journal of the Royal Statistical Society, series C (applied statistics), online. ISSN 0035-9254 (In Press)
Latent transition and latent class models for modelling diagnostic tests
Areas for methodological and applied research include evaluating tests and diagnostics; detecting a golden standard; measuring health conditions and health states; sensitivity analysis of measurement non-equivalence; combine structural equation models and multilevel models to account for nested multivariate data.
Potential areas of applications
-
Epidemiology and public health
Further reading
Koukounari, A., Moustaki, I., Grassly, N.C., Blake, I. M., Basáñez, M-G., Gambhir, M., Mabey, D. C. W., Bailey, R. L., Burton, M. J., Solomon, A. W. and Donnelly, C. A. (2013). Using a Nonparametric Multilevel Latent Markov Model to Evaluate Diagnostics for Trachoma. American Journal of Epidemiology, 177 (9). pp. 913–922. ISSN 0002-9262
Estimation from survey data (and related data), taking account of nonresponse and using auxiliary information
Areas for research include estimation under complex sampling schemes, including alternative forms of auxiliary information and treatment of non-response; non-random samples; surveys conducted over time; associated survey design issues; censuses and surveys in the presence of coverage errors; combination of survey and administrative sources; treatment of linkage errors; treatment of measurement errors.
Potential areas of applications
-
Data from various kinds of actual surveys, including possible cooperation with survey agencies or statistics groups in government.
Further reading
Skinner, C.J. and D’Arrigo, J. (2011) Inverse probability weighting for clustered nonresponse. Biometrika, 98 (4), pp. 953-966. ISSN 0006-3444
Micklewright, J., Schnepf, S.V. and Skinner, C. J. (2012) Non-response biases in surveys of school children: the case of the English ‘Programme for International Student Assessment’ samples. Journal of the Royal Statistical Society, series A (statistics in society), 175 (4), pp. 915-938. ISSN 0964-1998
Analysis of complex survey data
Areas for research include methods of weighting (and other point estimation), variance estimation and testing under complex sampling schemes for various methods of data analysis, including regression, multilevel modelling and latent variable modelling.
Potential areas of applications
-
Various methods of analysis for various actual surveys, with possible social science applications.
Further reading
Skinner, C. J. and Vallet, L.-A. (2010) Fitting log-linear models to contingency tables from surveys with complex sampling designs: an investigation of the Clogg-Eliason approach. Sociological Methods & Research, 39 (1), pp. 83-108. ISSN 0049-1241
Skinner, C.J. and Mason, B. (2012) Weighting in the regression analysis of survey data with a cross-national application. Canadian Journal of Statistics, 40 (4), pp. 697-711. ISSN 0319-5724
Kim, J.K. and Skinner, C.J. (2013) Weighting in survey analysis under informative sampling. Biometrika, 100 (2), pp. 385-398. ISSN 0006-3444
Disclosure risk assessment and statistical disclosure control
Areas for research include definition and measurement of privacy and disclosure risk under alternative threats, e.g. from record linkage; estimation of risk measures using models and resampling approaches; the impact of disclosure control methods.
Potential areas of applications
-
Different types of microdata designed for use by researchers, including those relating to both genomics and the social sciences
-
New forms of ‘open data’ released by government departments
-
Census outputs
Further reading
Skinner, C. and Shlomo, N. (2008) Assessing identification risk in survey micro-data. Journal of American Statistical Association, 103 (483), pp. 989-1001. ISSN 0162-1459
Young, C., Martin, D. and Skinner, C. J. (2009) Geographically intelligent disclosure control for flexible aggregation of census data. International Journal of Geographical Information Science, 23 (4), pp. 457-482. ISSN 1365-8816
Shlomo, N. and Skinner, C.J. (2010) Assessing the protection provided by misclassification-based disclosure limitation methods for survey microdata. Annals of Applied Statistics, 4 (3), pp. 1291-1310. ISSN 1932-6157
Longitudinal data analysis, especially event history (survival) analysis and dynamic panel models
Areas for methodological research include: non-ignorable drop-out, treatment of left-censored event history data, modelling household decisions using individual-level data, comparison of fixed and random effects estimators in dynamic panel models, handling the ‘initial conditions problem’ in random effects dynamic panel models, distributional assumptions in random effects models
Potential areas of applications
-
Housing transitions (residential mobility, choice of area, tenure change)
-
Union formation and dissolution
Further reading
Steele, F. (2008) Multilevel models for longitudinal data. Journal of the Royal Statistical Society, series A (statistics in society), 171 (1), pp. 5-19. ISSN 0964-1998
Steele, F., Clarke, P. and Washbrook, E. (2013) Modelling household decisions using longitudinal data from household panel surveys, with applications to residential mobility. Sociological Methodology, 43 (1), pp. 225-276. ISSN 0081-1750
Washbrook. E., Clarke, P.S. and Steele, F. (2013) Investigating non-ignorable drop-out in panel studies of residential mobility. Journal of the Royal Statistical Society, series C (applied statistics), online. ISSN 0035-9254 (In Press)
Multilevel simultaneous equations modelling of correlated social processes
Areas for methodological research include: identification of simultaneous equation models using covariate exclusions (instrumental variables) and repeated events, fixed effects (vector autoregression) and random effects estimators for bivariate dynamic panel models, sensitivity of random effects estimators to distributional assumptions
Potential areas of application
-
The relationship between housing transitions and other life course events such as births, employment change, and the formation and breakdown of co-residential unions
-
Reciprocal effects between family members. Examples include: (i) the effect of unemployment of either partner on their own mental health and that of their partner, (ii) the effect of decline in physical or cognitive functioning on a partner’s mental health and wellbeing.
Further reading
Steele, F., Kallis, C., Goldstein, H. and Joshi, H. (2005) The relationship between childbearing and transitions from marriage and cohabitation in Britain. Demography, 42 (4), pp. 647-673. ISSN 0070-3370
Steele, F., French, R. and Bartley, M. (2013) Adjusting for selection bias in longitudinal analyses of the relationship between employment transitions and health using simultaneous equations modelling. Epidemiology, 24 (5), pp. 703-711. ISSN 1044-3983
Steele, F., Rasbash, J. and Jenkins, J. (2013) A multilevel simultaneous equations model for within-cluster dynamic effects, with an application to reciprocal parent-child and sibling effects. Psychological Methods, 18 (1), pp. 87-100. ISSN 1082-989X
Data sources
The above research questions can be investigated using longitudinal data from panel studies and birth cohort studies. Examples of panel studies include the UK Household Longitudinal Study (Understanding Society), English Longitudinal Study of Ageing (ELSA), Survey of Health, Ageing and Retirement in Europe (SHARE), and British birth cohort studies (1970, 1958, 2000). Further details of these and other datasets can be found at the UK Data Archive.
http://www.data-archive.ac.uk/