Working paper by Gianluca Manzo, Educational Choices and Educational Traps, FMSH-WP-2013-48, GeWoP-1, octobre 2013.

The author

CNRS, University of Paris IV – Sorbonne (GEMASS) Gianluca Manzo is a permanent research fellow in sociology at the Centre National de la Recherche Scientifique (CNRS, France) and teaches quantitative and computational methods at the University of Paris-Sorbonne. He is an international research affiliate at the Institute for Futures Studies (Stockholm) and has served as visiting scholar and professor at several universities, which include Columbia University (ISERP) and the University of Oxford (Nuffield College).

The text

This is the original draft of my contribution to the special issue on « Class and Stratification Analysis » to appear in Comparative Social Research (guest editor: Gunn Birkelund). The article eventually published in Comparative Social Research (2013, 30: 47-100) as “Educational Choices and Social Interactions: A Formal Model and A Computational Test” is a new, deeply modified version of this draft. Readers are warmly invited to read and quote the published version of my article. It was first published by the GEMASS in November 2011 as GeWoP-1.

This paper is published by the Groupe d’étude des méthodes de l’Analyse sociologique de la Sorbonne (GEMASS – CNRS, Paris Sorbonne, FMSH) in Paris, under the cover of the Fondation Maison des sciences de l’homme’s collection of working papers.

To donwload this paper on HALSHS

Abstract

From a substantive point of view, the paper analyzes the statistical association between individuals’ educations and those of their parents in France and in Italy at the beginning of the twenty-first century – a phenomenon for which a comparison specifically focused on these two countries is still lacking. From a theoretical point of view, the paper develops a formal explanatory model of the macro-level structure of educational inequality – the «interdependent educational choice model» (IECM) – which frames educational choices as the result of both individual benefit/cost evaluations and peer-group pressures. It thus enriches the sociological educational rational-choice approach through the economic perspective of the so-called «membership theory of inequality». From a methodological point of view, the paper adopts agent-based simulations as a means to test the explanatory relevance of the IECM. Three groups of results are discussed. First, the paper demonstrates that simulation of the agent-based implementation of the IECM is able to generate a macro-level association between agents’ educational backgrounds and their educational outcomes whose basic structure well reproduces all the qualitative aspects of both educational outflows and opportunities observed in France and in Italy at the beginning of the twenty-first century. Second, by analyzing data collected at the agent-level during the simulation, the paper demonstrates that these macro-level patterns are produced on the basis of micro- and relational-level regularities which qualitatively mimic the real-world counterparts for which we have (at least, partial) empirical evidence. Finally, by manipulating the characteristics of the artificial network in which agents are embedded, the paper demonstrates that the IECM generates a realistic macro-level structure of educational inequalities on the basis of a cumulative process in which highly educationally-homophilic ego-centred dyadic networks fuel educational traps that progressively amplify the initial within- and between-group heterogeneity in agents’ perceptions of the benefits and costs of educational levels. Overall, the paper is a preliminary attempt to create, within the sociology of stratification and social mobility, an interface among formal theoretical modelling, the quantitative analysis of empirical data, and computational techniques.

Keywords

educational choices, social interactions, computational agent-based modelling, log-linear and regression

Choix et pièges éducatifs. Vers une intégration de la modélisation statistique et informatique au sein de la sociologie quantitative de la stratification sociale

Résumé

L’article étudie l’association statistique qui existe entre le niveau scolaire des enfants et celui de leurs parents en France et en Italie au début du vingtième siècle – un lien pour lequel il n’existe pas encore une comparaison systématique centrées sur ces deux pays. D’un point de vue théorique, l’article présente un modèle explicatif de la structure macroscopique des inégalités éducatives qui conceptualise les choix scolaires comme résultant de calculs coûts/bénéfices ainsi que d’influences relationnelles. De cette façon, l’article se propose d’enrichir l’approche du choix éducatif rationnel grâce à la perspective économique de la « théorie de l’inégalité fondée sur l’appartenance ». Sur le plan méthodologique, l’article utilise la simulation à base d’agents artificiels pour tester la pertinence explicative du modèle théorique proposé. L’article discute trois groupes de résultats. Tout d’abord, l’article démontre que tous les aspects de la structure macroscopique de l’association statistique sous étude peuvent être déduits par simulation à partir du modèle théorique proposé. Ensuite, en étudiant le modèle de simulation du point de vue des agents artificiels qui le composent, l’article montre que ces résultats macroscopiques sont produits sur la base d’hypothèses réalistes au niveau micro et relationnel. Enfin, en manipulant les caractéristiques du réseau artificiel dans lequel les agents sont encastrés, l’article démontre que le modèle théorique proposé engendre une structure macroscopique des inégalités éducatives en créant un processus cumulatif dans lequel la ségrégation des liens artificiels reliant les agents amplifie progressivement les différences initial dans leurs perceptions des bénéfices et des coûts de l’investissement éducatif. Dans son ensemble, l’article constitue une tentative préliminaire pour créer au sein de la sociologie de la mobilité et de la stratification sociale une interface entre la modélisation formelle, l’analyse quantitative de données empiriques et les méthodes de simulation  informatique.

Mots-clefs

choix éducatifs, interactions sociales, modèles de simulation à base d’agents artificiels, modèles log-linéaires et de régression

Sommaire

Introduction

Intergenerational Educational Mobility in France and in Italy

The Generative Computational Model: The “Interdependent Educational Choice Model” (IECM)

The Macro-level Patterns Generated by the IECM

Agents and networks at work within the artificial society driven by the IECM

Agents’ perceptions of education benefits and costs, agents’ failures and agents’ neighbourhoods
The relative importance of objective returns, subjective returns, costs and social influence
The determinants of an agent’s final educational outcome

Network manipulations and the macro-level patterns generated by the IECM

The quantity of dyadic links
The configuration of the inter-group links
The quantity of inter-group links

Conclusion

References

To donwload this paper on HALSHS