Writings on Educational Inequalities

// Agent-based Computational Modeling of Educational Inequalities (after The Spiral of Inequality)


In 2013, a special issue on "Class and Stratification Analysis" was edited by Gunn Elizabeth Birkelund in Comparative Social Research (table of contents) (guest editor's introduction to the special issue).


My contribution extends in several ways the analysis of educational stratification presented in my 2009 book, The Spiral of Inequality (see below). In particular, the formal model on which previous work was based is reformulated on a theoretical level, and  it is implemented in a more parsimonious way. The model is also studied by using a more transparent strategy. In addition, the model's implications are tested against new data concerning the statistical association between individuals’ education and the education of their parents, rather than the statistical association between individuals’ education and the social class of their parents


Abstract excerpt: (...) this paper introduces a micro-founded formal model of the macro-level structure of educational inequality, which frames educational choices as the result of both subjective ability/benefit evaluations and peer-group pressures. (...) To assess the relative explanatory contribution of the micro-level and network-based mechanisms hypothesised, the paper opts for agent-based computational simulations. (...) The paper’s main result is that ability and subjective perceptions of education benefits, no matter how intensely differentiated across agent groups, are not sufficient on their own to generate the actual stratification of educational choices across educational backgrounds existing in France at the beginning of the twenty-first century. (...)


Read the paper -- Manzo, G. (2013). “Educational Choices and Social Interactions: A Formal Model and A Computational Test”, Comparative Social Research, 30, 47-100.


This paper won the 2014 Mathematical Sociology Outstanding Article Award (Mathematical Sociology Section, American Sociology Association). 


// Agent-based Computational Modeling of Educational Inequalities (The Spiral of Inequality)


The Spiral of Inequality - Educational Choices in France and in Italy during the 20th Century, published in French in 2009, presents my doctoral dissertation in a condensed and refined form. In this work, I attempted to combine the survey-based, statistically-oriented tradition of the studies of social mobility with the emerging theoretical perspective seeking fine-grained, mechanism-based explanations. In particular, I focused on both the absolute and the relative dimension of educational inequalities in France and Italy in the 1990s as well as during the twentieth century. First, I described both aspects of these inequalities relying on Gini-based analysis and log-linear statistical modeling. Then, I formulated a theoretical model to explain the statistical patterns. According to this model, sociocultural constraints, the social return of education, cost-benefit considerations, and social interactions are responsible for the emergence of educational inequalities. Finally, in order to study the implications of this generative model, I translated it into an agent-based computational model and systematically studied it by numerical simulations. The analysis of this artificial society allowed to show the extent to which segregated social networks tends to amplify individual-level differences in terms of resources and preferences, thus finally leading to systematic asymmetry in the distribution of education and educational opportunity across social groups .


Since its publications, the book was widely commented in sociology --European Journal of Sociology (2011), Revue Française de Sociologie (2011), Sociologie du Travail (2012), Mathematics and Social Sciences (2009), Bulletin de Méthodologie Sociologique (2010), Rassegna Italiana di Sociologia (2010), Sociologica (2010)--, in political sciences --Revue Française de Science Politique (2010)--, in geography --Cybergeo: European Journal of Geography (2013)--, in Philosophy --Revue de Métaphasique et de Morale (2009), as well as in Popular Social Sciences magazines --Sciences Humaines (2009)-- or website --Liens Socio (2010).


Look at the table of content (French) (English) (Italian) ///// Book webpages (Presses de l'Université Paris Sorbonne) (Amazon)


Manzo, G. (2009) La spirale des inégalités. Choix scolaires en France et en Italie au XXe siècle, Paris, Presses de l'Université Paris-Sorbonne (PUPS), coll. L'intelligence du sociale, 336 p. (ISBN 978-2-84050-6444-7). 


Preliminary work for The Spiral of Inequality was also published in:


(in French) European Journal of Sociology --Manzo, G. (2007) “Le modèle du choix éducatif interdépendant. Des mécanismes théoriques aux données empiriques françaises et italiennes”, European Journal of Sociology, 48, 1, 3-53. [Read the paper]


(in French) Mathematics and Social Sciences --Manzo, G. (2006) “Actions, interactions et structure dans l’émergence de la stratification sociale des diplômes : un modèle de choix discrets avec externalités”, Mathematics and Social Sciences, 175, 3, 53-99. 


(in Italian) Quaderni di Sociologia -- Manzo, G. (2009) “Interazioni, interdipendenze e stratificazione educativa. Simulazioni multi-agente di un modello formale”, Quaderni di Sociologia, LIII, 49, 113-152. [Read the paper]



// Statistical Modeling of Educational Inequalities (before The Spiral of Inequality)

My first attempt to illustrate the claim that even very advanced forms of statistical modeling are not able to shed light on the micro-basis of macroscopic patterns of inequality is contained in an article I published in 2006 in Quality and Quantity.


Abstract Excerpt: (...) The article examines a specific variant of this approach to modeling which consists of log-linear topological models. (...) It first maintains that log-linear topological models, especially in their multi-matrix variant, are extremely useful in integrating sociological theory with empirical quantitative analysis. It then shows that the principal shortcoming of these models is that they only partially allow the accurate modeling of the generative mechanisms underlying all the empirical regularities observed in aggregate data. (...) In order not to remain at the abstract level of epistemological reflection, the article will attempt to show the well-foundedness of this thesis by constructing a multi-matrix log-linear topological model for the analysis of a contingency table which cross-classifies social origin with the educational qualification.  (...)


Read the paper --Manzo, G. (2006)  “Generative Mechanisms and Multivariate Statistical Analysis. Modeling Educational Opportunity Inequality by Multi-Matrix Log-Linear Topological Model: Contributions and Limits”, Quality & Quantity, 40, 5, 721-758.


My interest for quantitative methods that are not mainstream went back to my bachelor dissertation (Trento, 2001) in which I explored the possibility to describe the structure and the temporal evolution of educational inequalities in Italy by means of artificial neural networks (ANN). At that time, I used this method as a multidimensional classificatory technique and did not exploit its potential as a truly formal modeling technique. In particular, I used a specific kind of supervisioned ANN known as "multi-layer perceptron" (with back-propagation algorithm). My analysis suggested that this method leads to results comparable to regression-based methods for the analysis of cross-sectional patterns of inequality whereas it should not be used to study the temporal evolution of these patterns if one is interested in controlling for the temporal changes in the social and educational structures. This claim is discussed in a paper published in Sociologia e Ricerca Sociale.


Read the paper (in Italian) --Manzo, G. (2003, with C. Corposanto) “Disuguaglianze educative e loro trasformazioni nel tempo: problemi tecnici e metodologici di un’analisi compiuta mediante reti neurali artificiali”, Sociologia e Ricerca sociale, XXIV, 70, pp. 43-69.


// Early Theoretical Analysis of Educational Choices


My very first approach to the topic of educational choices and inequality was theoretical. In particular, during a class on sociology of education given by Antonio Cobalti at the University of Trento in 2000, I started to be intrigued by the opposition between action-based explanations of education choices and more cultural and stucture-oriented approaches. This was the topic of the very first article that I was able to publish, in Italian, in Studi di Sociologia.


In particular, in this paper I examined two theoretical frameworks: on the one hand, the theory of cultural capital proposed by Bourdieu and Passeron, and, on the other hand, the rational educational choice model outlined by Boudon, and, then refined by several authors among which Gambetta, Goldthorpe, Breen or Erikson and Jonsson. By systematically reviewing the empirical evidence available, I attempted to demonstrate that several aspects of empirical data are at odds with the implications one can draw from the former approach whereas the latter is able to shed light on the mix of change and stability that seems characterize macroscopic patterns of educational inequality in several western countries.


Read the paper (in Italian) --Manzo, G. (2004) “Verso una teoria delle diseguaglianze di opportunità educative”, Studi di Sociologia, 42, 1, 79-113