Education has continued to grow in stature and significance as an academic discipline. In addition to world renowned research studies the growth of education has been seen in the methodology and methods underpinning its research. The BERA/SAGE Handbook of Educational Research provides a cutting edge account of the research and methodology that is creating new understandings for education research, policy and practice. Over two volumes, the handbook addresses educational research in six essential components: Section 1: Understanding Research Section 2: Planning Research Section 3: Approaches to Research Section 4: Acquiring Data Section 5: Analysing Data Section 6: Reporting, Disseminating and Evaluating Research Featuring contributions from more than 50 of the biggest names in the international field, The BERA/SAGE Handbook of Educational Research represents a very significant contribution to the development of education.
Chapter 44: Multilevel Modelling for Educational Data
Multilevel Modelling for Educational Data
Introduction: Hierarchically structured data
Interesting real life data rarely conform to classical text book assumptions about data structures. Traditionally these assumptions are about data that can be modelled with independently, and typically identically, distributed ‘error’ terms. More often than not, however, the populations that generate data samples have complex structures where measurements on data units are not mutually independent, but depend on each other through complex structural relationships.
For example, a household survey of voting preferences will typically show variation among households and voting constituencies (constituencies differ and households differ, on average, in their political preferences). This implies that the replies from individual respondents within a household (or within a constituency) ...