Premium
Combining multilevel analysis with national value‐added data sets—a case study to explore the effects of school diversity
Author(s) -
Schagen Ian,
Schagen Sandie
Publication year - 2005
Publication title -
british educational research journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.171
H-Index - 89
eISSN - 1469-3518
pISSN - 0141-1926
DOI - 10.1080/01411920500082144
Subject(s) - multilevel modelling , certificate , multilevel model , set (abstract data type) , educational research , national curriculum , diversity (politics) , educational attainment , psychology , mathematics education , pedagogy , computer science , sociology , curriculum , political science , algorithm , machine learning , anthropology , law , programming language
The advent of large‐scale matched data sets, linking pupils' attainment across key stages, gives new opportunities to explore the effects of school organisational factors on pupil performance. Combined with currently available sophisticated and efficient software for multilevel analysis, it offers educational researchers the chance to develop objective evidence about issues both old and new. Previously reported research, based on separate data sets from Key Stage 2 1997 to Key Stage 3 2000 and Key Stage 3 1998 to General Certificate of Secondary Education (GCSE) 2000, focused on the impact of selective, specialist and faith schools. We now have access to a national data set with 380,000 pupils' Key Stage 2 levels in 1996 matched to their GCSE performance in 2001, and this has enabled us to rework these analyses based on value‐added results across all five years of secondary education. Results of the analysis largely confirmed the findings of previous work, in terms of the apparent impacts of different school types on pupils' progress. The article emphasises the important contribution that the marriage of national matched data sets and multilevel modelling will continue to make to educational research.