Rebecca's projects have to date all been in the substantive area of Education. Her first project, which she worked on with Jon Rasbash, George Leckie and Jennifer Jenkins, partitioned variation in English children's academic progress between the ages of 11 and 16 into area effects, school effects, family effects and individual effects. This work brought together the previous literature on school effectiveness (which has ignored family effects) and behavioural genetics (which has ignored school and usually area effects), although it was not able to partition the family effects into genetic and environmental effects due to limitations of the data. Further work is planned using Danish data which should be able to identify the genetic and family environmental effects and also pre-school effects, in addition to school, area and individual-specific environmental effects. This work was published in the Journal of the Royal Statistical Society Series A (see Publications) and was selected as the best paper of the year from that journal, to be presented at the 2010 Royal Statistical Society conference. The authors were also invited to adapt the paper to appear in the Royal Statistical Society's Significance magazine (see Publications).
Rebecca has also worked on school segregation with Harvey Goldstein, George Leckie and Kelvyn Jones. This work has built on Goldstein and Noden (2003)'s idea that multilevel modelling provides a better way to measure segregation than the traditionally used indices, most importantly because it separates out chance variation from systematic sorting, but also because it allows partitioning of segregation into different levels e.g. school and Local Education Authority. Multilevel modelling can also be extended very easily to look at segregation according to some characteristic with more than two categories, something which is much more difficult when using indices, and in 2009 at a conference/ workshop in Bristol entitled Measuring segregation: methods, tools and data Rebecca presented results from an analysis looking at ethnic segregation in English state schools between 2002 and 2008, and investigating to what extent this could be explained by economic segregation (see Presentations). Work has also been done on the properties of the variances estimated by multilevel models as segregation indices. A paper extending Goldstein and Noden (2003)'s comparison of the multilevel approach to the index-based approach and including some simulation work and a substantive application to economic segregation in London state schools has been published in the Journal of Educational and Behavioral Statistics.
Rebecca has done some descriptive work with the user data of the Centre for Multilevel Modelling's online training materials. This forms part of the evaluation of the Centre for Multilevel Modelling's training and capacity building activities, which aims to produce findings that can be used to improve the materials as well as to investigate what people find difficult about learning multilevel modelling. This work is with Jo-Anne Baird and Fiona Steele.
Rebecca is looking at intelligence tests and achievement for her PhD project. Intelligence tests are often used in school entrance exams in the UK, so it is important to understand their performance as predictors of later achievement. She will examine whether intelligence tests are equally good predictors of achievement across the range of socio-economic status (SES). She will then partition covariation between intelligence tests and achievement into genetic, shared environmental and non-shared environmental components and allow these to differ across SES, to examine whether the differences between individuals are due to genetics and to the environment to the same extent across SES. She has already done some preliminary methodological work evaluating methods for investigating differential heritability, which she presented at the 2010 Research Methods Festival (see Presentations); she plans to publish this as well as including it in her PhD thesis.
Note: some of the documents on this page are in PDF format. In order to view a PDF you will need Adobe Acrobat Reader