The Centre for Multilevel Modelling (CMM) is a research centre based at the University of Bristol. Our researchers are drawn from the School of Education. We collaborate with a range of researchers in the School of Geographical Sciences, Population Health Sciences and Bristol Veterinary School working with multilevel models.
Multilevel Modelling is one of the basic techniques used in quantitative social science research for modelling data with complex hierarchical structures. The Multilevel Modelling research theme focuses on producing new statistical methods for tackling research questions, developing new software for implementing this methodology and disseminating these techniques to the national and international social science community.
Research
We develop new statistical methodologies and implement them in various software tools in order to address unsolved issues in quantitative modelling of social processes.
Understanding intersectionality: developing intersectional quantitative methods
This project aims to refine and disseminate a recently proposed multilevel modelling approach to studying intersectionality of individual outcomes. That is, the notion that individuals' various social and political identities result in unique combinations of discrimination and privilege. More information
The Bias In Primary Education Project (BIPE Project)
Quite recently, teachers were asked to predict students’ GCSE and A-level marks because exams could not take place due to the Covid-19 crisis. This sparked public debates about a concern that has been frequently highlighted over the last decades: teacher judgements might be biased putting students from certain groups of the population at disadvantage.
Social or ethnic teacher bias is problematic because teachers have an important influence on students’ self-esteem, academic motivation and school achievement. Moreover, in many countries they make critical decisions about ability groupings, school tracking and grade retention. If teachers underassess the performance, effort or enjoyment in school of certain students, they can harm students’ self-esteem and hinder them to achieve the education and life outcomes they could have attained through their abilities. Research shows that positive teacher-student relationships, which are characterised by openness and mutual understanding, are highly beneficial for students, especially in primary school. Teachers’ inaccurate and possibly biased perceptions of students can be detrimental for teacher-student relationships. More information
The 2020 GCSE and A-level 'exam grades fiasco'
The awarding of the 2020 GCSE and A-Level exam grades in England was widely viewed as a ‘fiasco’. When COVID-19 forced the cancellation of exams, DfE and Ofqual asked centres (schools and colleges) to submit Centre Assessment Grades (CAGs) and rankings. Namely, the grades and rank orders within their centres that teachers thought students would have achieved had they sat their exams. Ofqual, tasked with preventing grade inflation and ensuring grading consistency, viewed students’ CAGs as overly optimistic and so replaced them with calculated grades predicted via their Direct Centre-level Performance (DCP) algorithm. The result was that 40% of CAGs were downgraded by one or more grades. After public outcry, Education Secretary Gavin Williamson instructed Ofqual to revert to the original CAGs. Students are accepted into universities and employment based on their GCSE and A-level grades. Their grades directly impact their immediate future. It is therefore vitally important for society to understand the extent to which students’ grades were unfairly awarded in 2020 and 2021 with biases potentially varying across individual centres and by student and school characteristics. Our overarching aim is therefore to conduct an independent and rigorous secondary data analysis of the 2020 and 2021 GCSE and A-level exam grades to explore not just what went wrong statistically, but to identify what could be improved statistically when predicting grades in future years. More information
The role of sample characteristics in the stability of value-added estimates of school effects
This project aims to better understand multilevel models for studying the stability of school value-added effects for school accountability and improvement. Particular focus will be on studying their sensitivity to sample size requirements, student mobility, and changing student intake sociodemographic composition. More information
Gallery of Multilevel Papers
Software
The Centre for Multilevel Modelling's own software enables quantitative social science researchers to become effective multilevel modelling practitioners.
- Purchase MLwiN
- Upgrade to latest version
- Download for free if you are a UK academic
- Software help: MLwiN FAQs, MLwiN user forum.
Learning
Introduction to Multilevel Modelling Using MLwiN, R or Stata in-person and online training course
This three-day course held in January and July every year provides an introduction to multilevel modelling and includes software practicals in your choice of software: MLwiN, R or Stata. We focus on multilevel modelling for continuous and binary responses (dependent or outcome variables) when the data are clustered (nested or hierarchical). More information
Online course
Graduated modules starting from an introduction to quantitative research, progressing to multilevel modelling of continuous and binary data. More information and course modules.
Videos and voice-over presentations
Including Random Intercept Models, Residuals, Measuring Dependency, Covariance and Correlation Matrices, and Random Slope Models. More information.
Advanced Quantitative Methods in Social Sciences and Health
The Advanced Quantitative Methods (AQM) pathway of the SWDTP offers ESRC +3 and 1+3 postgraduate research training in the application of advanced quantitative methods in the social sciences and health. More information.
If you would like to be informed of future workshops, you can subscribe to the Centre for Multilevel Modelling Newsletter.
News
- MLPowSim 1.1.2 released 31 October 2024
- MLwiN 3.13 released 11 October 2024
- Introduction to Multilevel Modelling Using MLwiN, R or Stata, 7-9 January 2025 17 September 2024