Understanding contemporary career mobility: career pattern mining, gender disparities, and skill transferability-based reemployment

Abstract

This thesis investigates the dynamics of contemporary career mobility through a boundary-centered framework, emphasizing how boundaries punctuate, constrain, and enable career trajectories in the context of technological transformation, organizational restructuring, and persistent inequality. Drawing on large-scale career history data, occupational databases (O*NET), and longitudinal survey data (NLSY79), it develops advanced empirical methods to analyze career patterns, gendered disparities, and skill transferability-based reemployment.
The first study introduces DECPPM (Density Equilibrium Career Path Pattern Mining), a novel method for analyzing large-scale, real-world career path data. DECPPM comprises two steps: job title clustering and career pattern mining. In the first step, existing job title clustering methods rely heavily on lexical or semantic job title similarity, which tend to produce inaccurate or overly coarse clusters. DECPPM addresses these shortcomings by integrating job transition information directly into job representation through a novel sequence-enriched embedding and then applying a hierarchical and iterative clustering algorithm to distinguish between job functions and levels. This clustering structure solves the problem of occurrence imbalance among jobs in real-world career paths. In the second step, DECPPM integrates the frequent sequential mining (FSM) algorithm into career pattern analysis, providing a unique perspective compared to past sequence-clustering-based CPM methods. We evaluated our approach on an online professional dataset of 1.55 million career path sequences from IT professionals, and it outperformed other baseline methods in both the job title clustering and career pattern mining steps. This study improves career path analysis by uncovering nuanced transition pathways and thereby offers a more granular understanding of career progressions. By operationalizing the multi-dimensional nature of job mobility, this study refines boundaryless career theory with empirical evidence on how individuals navigate complex mobility structures in real-world labor markets. At the same time, it demonstrates how these transitions function as boundary punctuation events—not isolated or episodic, but cumulative and patterned markers that segment careers into recognizable stages of progression, plateau, or redirection.
The second study focuses on gender disparities in career mobility within the IT industry. Drawing from the same large-scale dataset, the study uses the job clusters and career patterns generated in the first study to explore gender-based differences in mobility patterns. The findings reveal that while men and women advance at similar rates in terms of job level progression, women are significantly less likely to receive internal promotions. Instead, they rely more heavily on external job switching to achieve upward mobility. At the same time, while mentored women show lower levels of external mobility compared to men, sponsored women demonstrate similar rates of external job transitions as their male counterparts.
From a boundary-centered perspective, these results show how career boundaries operate as constraints: internal organizational boundaries, reinforced by male-dominated networks and limited sponsorship, systematically restrict women’s upward progression. As a result, women are forced to cross external boundaries more often to pursue advancement, bearing higher costs and greater risks. The second study makes important theoretical contributions by refining our understanding of gendered career mobility through a longitudinal lens. It demonstrates how boundary constraints accumulate over time and generate enduring disparities between groups. This study advances four key theoretical perspectives. First, it refines boundaryless career theory by showing that women’s mobility is constrained by structural barriers, despite assumptions of open movement. Second, it contributes to social capital theory by identifying a closure penalty, where dense male networks limit women's access to advancement. Third, it supports the invisibility hypothesis by demonstrating that women must attain greater visibility for equal recognition. Finally, it also offers a gendered extension of sponsored mobility theory by highlighting women’s reliance on external pathways due to limited access to internal sponsorship. Practically, the second study highlights the need for organizations to reassess internal promotion and sponsorship practices. It shows that informal sponsorship mechanisms systematically disadvantage women, suggesting the importance of formalized mentorship programs, transparent evaluation criteria, and inclusive leadership pipelines. The findings also underscore the value of external networks for women’s career advancement in male-dominated industries, offering actionable strategies to improve gender equity in organizational mobility systems.
The third study shifts the focus from career sequence structure to the transferability of skills across occupations, especially in the context of job displacement. It introduces a novel deep learning model, JR-CGN (Job Relatedness-aware Composition Graph Network), which estimates skill-level transferability between occupations by learning the underlying relationships from occupational data in O*NET. Based on the transferability estimates produced by JR-CGN, the study develops the occupational centrality (OC), a novel metric that quantifies the extent to which a given occupation's skill set is transferable to other jobs. Empirical analysis with NLSY79 survey data on displaced workers shows that individuals from occupations with higher OC scores are more likely to switch occupations after displacement and to find reemployment more quickly. Theoretically, the third study advances task-specific human capital theory by operationalizing the concept of skill-based ladders through the development of OC, a structural measure of occupational skill transferability. From a boundary centered perspective, this demonstrates the enabling function of boundaries: occupational structures with higher centrality provide identifiable pathways that facilitate reemployment and support career resilience. By empirically linking higher OC scores to faster reemployment and greater occupational switching, the study demonstrates that inter-occupational mobility is shaped not only by individual qualifications but also by the degree of skill transferability embedded in occupational structures. Practically, the third study provides a data-driven framework for employers, policymakers, career counsellors, and individual workers. By combining skill transferability measures with OC measure, it enables more precise identification of transferable competencies and structurally connected occupations. This supports targeted reskilling, competency-based hiring, and informed reemployment decisions. For individuals, especially those in vulnerable roles, it promotes more confident and strategic career transitions through a proactive, evidence-based understanding of occupational change.
Overall, this thesis offers a comprehensive, data-driven investigation into the mechanisms of career mobility in contemporary labor markets. By integrating large-scale career data, advanced modeling techniques, and critical theoretical perspectives, it contributes both methodological innovations and empirical insights to the study of career dynamics. Through its focus on multi-dimensional job transitions, gendered disparities on career promotion patterns, and reemployment pathways shaped by skill transferability, the thesis demonstrates that careers are best understood as sequences of boundary encounters in which punctuation, constraint, and enablement interact to shape trajectories. In doing so, it advances a more nuanced and theoretically integrated understanding of how individuals navigate careers in an era of rapid technological transformation, organizational restructuring, and enduring structural inequalities.

Date of Award	15 Oct 2025
Original language	English
Awarding Institution	University of Nottingham
Supervisor	Qingxin Meng (Supervisor) & Alain Chong (Supervisor)