Modelling longitudinal sequences by optimal scaling techniques

The development of monitoring systems for the measurement of the quality of Public and Private Institutions that distribute services to the labour force (as job placement, training courses, economic assistance programs and other supports to workers) has stimulated various evaluation strategies in order to help the institutional stakeholders to improve the processes, the outcomes and the service distributed. Particularly, the development of monitoring informative systems based on administrative archives has been growing in importance. To understand the labour market dynamics from such large datasets, it is necessary to be able to identify patterns, trends and relationships in the data and identify “statistical occurrences” in order to simplify their global structure to facilitate decision-making. The aim of the present article is to classify the temporal evolution of vocational experiences of workers’ careers in a specified territorial context (Province of Milan) in terms of contractual typologies, based on large administrative archives (about 2,900,000 vocational experiences related to 1,280,000 workers). The final goal is a synthetic clustering that identifies individuals in homogeneous groups as regards the succession of contractual typologies, identifying from one end the workers’ profiles that remain stable in each contractual typology and from the other end the profiles that improve or worsen contractual stability. Methodologically, after having discussed the limitations of traditional approaches, we provide a proper methodology in the framework of multidimensional scaling taking into account heterogeneity of worker characteristics.