Mapping heterogeneity in retirement transition pathways: integrating qualitative longitudinal and text-mining analysis

Retirement transition (RT) is a dynamic process with individual heterogeneity. While the Resource-Based Dynamic Perspective (RBDP) highlights fluctuations in key resources, studies remain largely quantitative and cross-sectional. This study aims to identify distinct RT patterns and examine shifts in resource-related discourse over time.

A qualitative longitudinal design with computational augmentation tracked 12 Taiwanese near-retirees across 5 waves spanning 6 months pre- to 6 months post-retirement. 60 qualitative transcripts were analyzed using longitudinal case methods, complemented by computational text mining (Sentence-BERT) to trace trajectories across 6 resource domains, including Health Issues, Life Control, Financial Status, Social Support, Emotional Response, and Lifestyle.

Within this sample, 3 analytically derived RT patterns were observed: a transition-centered pattern in Strategic Goal-Driven (Type I, n = 4), a U-shaped pattern in Structure and Social Role Reshaping (Type II, n = 4), and a crisis-driven peak in External Shock Response (Type III, n = 4). Text-mining results broadly converged with qualitative findings, particularly in the Life Control domain. At the individual level, convergence was observed for financial status and lifestyle, whereas divergence emerged when similar quantitative frequencies corresponded to different qualitative meanings.

Findings extend RBDP by introducing resource diversion, in which family or contextual demands redirect personal resources, and a resource hierarchy highlighting the foundational role of physical health. Findings underscore the need for pathway-specific retirement support that accounts for culturally embedded caregiving obligations beyond conventional financial-planning approaches.