Tip:
Highlight text to annotate it
X
My name is Fiona Steele,
I am Professor of Social Statistics.
I also co-direct the Centre for Multilevel Modelling,
which is based within the Graduate School of Education.
My name is George Leckie and I'm a lecturer in Social Statistics
within the Centre for Multilevel Modelling.
FIONA: We work across the Faculty of Social Sciences
with people from Geography in particular,
and we also have a Co-director who is in the Veterinary Sciences.
We work primarily to develop methods for analysing data
with a particular kind of structure.
We also develop a software package so people can implement these methods
in their own research.
We also do a lot of training.
Multilevel modelling is a set of statistical techniques
for allowing for dependencies
between individuals who are in some kind of social group.
So, for example, children within schools,
individuals within households.
And the social grouping is often of interest in itself.
For example, if you're looking at children's educational progess,
you'd be interested in how much of that
is due to factors to do with the family,
how much to do with the school they attend or the area they're living in.
We have a range of academic users,
but also we have users in Government and in other organisations.
GEORGE: The process starts with our Centre making multilevel-modelling software,
developing the methodology and so on.
The analysis which the Government runs
on their datasets
of students' exam performances
is summarised by the multilevel model,
the results from the multilevel model
are then presented on the Government's webpages
and then this information is provided
to parents to help them look at the performances of schools,
to help them hold school accountable,
to help them choose where to send their child for secondary education.
So we have a nice chain from the development of the software
and the Centre's promotion of multilevel-modelling methods
all via researchers to end users - the general public.