School effects on pupil outcomes: quantitative methods and applications
Educational organisations and social structures play a fundamental role in child development. The recent availability of a range of large-scale longitudinal data sets on the education achievement of pupils in England enables these influences to be examined. However, there are important challenges for quantitative researchers as to how to appropriately analyse the complex hierarchical and non-hierarchical data structures found within them.
This session presents methods for, and applications to, examining the effect of schools on pupil achievement. The papers showcase research by the LEMMA and ADMIN nodes of the ESRC National Centre for Research Methods (NCRM), a network of research groups, each conducting research and training in an area of social science research methods. For LEMMA this is multilevel modelling methods while for ADMIN this is methods for linking and jointly analysing administrative and survey data sets.
The session begins with a paper by Clarke, Crawford, Steele and Vignoles who compare the strengths and weaknesses of the popular fixed and random effects (multilevel) regression modelling approaches that are commonly used in educational research to analyse hierarchical data structures. They demonstrate that that the choice of approach is not straightforward and is very much specific to the empirical application.
The second paper by Leckie and Goldstein examines the usefulness of the Government
(Click on an abstract title to view it in PDF format.)