The LinDOR – Linking Data for Open Research project addresses the methodological and operational challenges involved in integrating administrative microdata and survey sources, with the overarching goal of developing an advanced and replicable data infrastructure to support research on the work and health biographies of an ageing population. Its central objective is the creation of an anonymised, interoperable, and enriched Micro File for Research (MFR), enabling longitudinal analyses of employment trajectories and health outcomes, with particular focus on transitions to retirement, social inequalities, sustainable employment in later life, and the evaluation of equity-oriented public policies.
LinDOR aims to design innovative methodologies for record linkage across heterogeneous datasets (e.g., WHIP-Salute, National Student Registry), establish standards for the secure dissemination of microdata, and leverage technologies such as machine learning and large language models (LLMs) to enhance and semantically enrich information. In parallel, the project seeks to conduct a pilot study on firm-level determinants (wages, productivity, contractual arrangements) and work-health trajectories, identifying risk patterns and critical factors over the life course.
By combining advanced statistical techniques, differential privacy solutions, and an open-science approach, LinDOR aspires to build a replicable model for the ethical production and sharing of integrated microdata, providing robust evidence to inform public research and policy-making for active ageing and sustainable employment.
April 2024 – March 2025
The LinDOR – Linking Data for Open Research: Methodological Issues and a Pilot on Work and Health Biographies in an Ageing Society project (Age-It, Spoke 6) aims to develop innovative methodologies for integrating administrative microdata and survey data to build a Micro File for Research (MFR) that documents work and health biographies of older individuals, supporting research on ageing, pension equity, and sustainable employment. During the first two semesters, the project completed WP1 tasks (review of existing databases and data access procedures, design of record linkage protocols, initiation of a dissemination conference) and WP2, producing the first version of an anonymised MFR (sampling, generalisation, perturbation) and enhancing it through a machine learning-based pipeline that synthetically reconstructed educational attainment, probabilistically matched with the National Student Registry (Deliverables D2.2-D2.3). Preliminary analyses in WP3 were launched to investigate firm-level determinants of productivity, wages, and inequalities in older workers' careers, while WP4 explored career-health trajectories, occupational risk factors and chronic disease incidence approaching retirement, supported by a large language model and a preliminary job-exposure matrix. A critical issue related to delays in the release of the updated WHIP-Salute dataset required reliance on the 2012 version for the MFR prototype, leading to approximately a one-month postponement of advanced analyses, without compromising final objectives. The current progress is estimated at 90% (WP1), 75% (WP2) and 30% (WP3–WP4), with a dissemination conference planned for mid-2025