Complexity to inform issues of methodology, taxation and politics in a context of crisis

Soon after I started working at Durham University in 1970 Bill Williamson and I set up a course (as we called them then) dealing with both the Sociology and Social Policy of Education. I had taken the Education option in my LSE MSc – taught by Tessa Blackstone who was just five years older than me – now Baroness Blackstone after a career straddling education and politics.  Bill was more mainstream Sociology than me and had previously taught at Newcastle Poly.  I had been influenced by my Tutor at the LSE, Bleddyn Davies (lovely man) to think about territorial injustice – the subject of his book Social Needs and Resources in Local Services (1968) and there was a developing literature in educational spatial inequality which focused on the administrative deliverers of education in England, Local Education Authorities (LEAs).  Bill and I, both with working class origins in the NE of England but with parents whose educational attainment had been enabled by war service (although my mother had risen from a 17 year old Tax Officer in the Revenue to be about to begin training as an Inspector of Taxes until the marriage bar knocked her out of her permanent and eligible for promotion grade in the civil service and my birth forced her out altogether into primary teaching).  

We were both influenced by the left Weberian take on class (although I remained a more or less committed Marxist in the Rosa Luxemburg tradition) and in particular by our boss John Rex with his discussion of housing class and David Harvey and Ray Pahl’s social geographies of inequality. We decided to use data about LEAs to explore the relationships among educational attainment, defined as people remaining at school to 16 (the leaving age was 15 and staying to 16 enabled people to sit meaningful GCE O levels), 17 which meant going for A levels, and entry into higher education on the one hand and the class backgrounds, educational resources and environmental character of the LEAs in which they lived. Our focus was not on individuals but on cohorts of children and young people, in explicit contrast to studies which focused on individuals and their personal and familial characteristics.  The result of this research for which we got a Social Science Research Council Grant and employed a research assistant, Barbara Fletcher was Byrne, Williamson and Fletcher The Poverty of Education – a study in the politics of opportunity (1975) along with a Durham University Working Paper and a couple of journal articles.

We deployed two quantitative approaches in this work, alongside a qualitative element with interviews with Directors of Education and review of local policy documents and histories.  The first was essentially the partialling out of the correlations among class background, educational resources, environment and resources with our cases being LEAs. We tested a model exploring the relationships among class background of LEAs, resources devoted to education, social environment and educational attainment.

I am a great believer in the value of correlation coefficients rather than regression coefficients and in focusing on the strenght of relationships rather than their form. Whenever you look at a regression models always pay attention to the R squared / pseudo R squared. Partialling out, calculating partial correlation coeffiecients, gives us an insight into how an ultimate cause like class works through forms like differential access to resources.

Above is a failed effort to upload a Ven Diagram – I will work this out later. Our finding was always that when resources were taken into account the correlations between class background and attainment were very substantially reduced and adding environment in reduced them even further.  We first used a statistical package written locally at Durham in Fortran but moved onto using an early version of SPSS. This required  us to write scripts which as I have asserted in a previous post is a good discipline.

We had to enter all our data by hand copying from printed publications from the Census, Educational Statistics, and for resources from the published data by LEA from the Institute of Municipal Treasurers and Accountants.  These were entered as rows onto data sheets which were punched up by the punch operators onto cards and the card set was our data base – not dropping a card set which would require resorting was a must.

Our other quantitative method was cluster analysis to which we were introduced by Eric Tannenbaum, a programming advisor at Durham with a background in biological numerical taxonomy. We used the CLUSTAN package combining Ward’s method of hierarchical fusion (still my go to method) with relocation to tighten up clusters.  The clusters were used both to describe geographical differentiation among LEAs – a much improved approach compared with earlier use of Factor weightings – and as a basis for selection of LEAs to visit to interview significant policy actors – Directors of Education – and explore some policy history.  The descriptive and representative selection of cases is a powerful set of ways for deploying any kind of numerical taxonomic method.  

This study preceded the generalization of comprehensive secondary education and the whole structure of provision has been modified by the demented assault on Local Government management of school systems initiated by Thatcher’s crew and continued in the Blair / Brown years. It is a historical account of how things were and we would need a finer grained approach at the school level to replicate our exploration of the way resources and environment influence cohort attainment.  I have attempted work on this basis more recently and will comment on that in a subsequent post  but next post will address my involvement with neighbourhood housing differentiation and its history.