Social Statistics

Society, Statistics (and some sermons)

Social Statistics - Society, Statistics (and some sermons)

Five New Academic Posts in Quantitative Social Science at the University of Bristol

Professor of Quantitative Methods / Social Statistics (job number ACAD100623)
Lecturer / Senior Lecturer in Policy Studies with Quantitative Research Methods (job number ACAD100624)
Lecturer / Senior Lecturer in Politics with Quantitative Research Methods (job number ACAD100625)
Lecturer / Senior Lecturer in Sociology with Quantitative Research Methods (job number ACAD100626)
Senior Lecturer/Reader in Quantitative Methods/Social Statistics (job number ACAD100622)

An Introduction to Mapping and Spatial Modelling in R (draft version)

This is an introduction to R, to mapping and spatial modelling in R, and to using R as a simple GIS. Follow the instructions on page 3 to install R, the required libraries and the relevant data sets for the practicals. What appears below is a preview of the whole document. The full and most up-to-date version is always available here.

Are indices still useful for measuring socioeconomic segregation in UK schools?

In this commentary, Michael Watts raises some important questions about the measures of segregation typically used in educational research. His words of warning are not limited to such indices but could apply more generally to forms of measurement and modelling of systems that are fast changing and where the ‘object’ of measurement is poorly defined. His questions, in a nutshell, are (a) what actually has been measured (what does it really mean)? And (b) is there much point in making comparisons over time when that you are trying to compare keeps on changing? These are important points; especially so in the case of the UK education system that politicians seem to love to tinker with and force changes upon. However, in this short reply, my colleagues and I are a little more sanguine and suggest, contrary to Watts, that indices are still useful, despite the problems with their use.

Ethnicity and economic disadvantage in England in 2011

I have been working on a simple classification that categorises small area census neighbourhoods (the ‘Output Areas’) according to some dimensions of economic activity and inactivity. One group of neighbourhoods I have identified is those that are: (a) in the lowest 20% nationally for the percentage of the economically active population in full-time employment; (b) have an unemployment rate in the highest 10% nationally; and where (c) the percentage of the total population (neither student nor retired) that has a long-term sickness or disability is in the highest 10% nationally. In short, these area areas where we can expect the greatest economic stresses.  Continue reading

White flight, ethnic cliffs and other unhelpful hyperbole?

In an (unguarded?) conversation with a journalist, I talked about a ‘cliff-edge’ measure of segregation where neighbouring places have very different proportions of their resident population classified as White British in the 2011 Census. The words, rephrased as ‘ethnic cliffs’ was soon coupled with talk of White Flight from British cities and has appeared in a number of national newspapers and magazines, alongside like ‘self-segregation’ and ‘sundown segregation’ (The Sunday Times and the Daily Mail). In this presentation I look at changes to the ethnic composition of census zones in England from 2001 to 2011 and ask whether such phrases are unhelpful hyperbole or simply vivid but accurate descriptors of “Britain’s new problem” (Goodhart, 2013 writing in Prospect Magazine).

Teaching regression (and t-tests) using Shiny

Thanks to Chris Brunsdon at the University of Liverpool for first putting me onto Shiny which is used to create interactive webpages running R on the server side. In this example, I have created a simple ‘tutorial’ that can be used to explore some ideas behind simple bivariate regression. It allows the student to try changing the slope of the line, the noisiness of the data or the number of observations, to see the effect of doing so on a scatter plot, and then to explore what the change does to standard regression output (in R or any other stats package). The underlying pedagogic idea is learning by doing. Continue reading