PhD Candidate in Population-Scale Network Science (1.0 FTE, 4 years)

PhD Candidate in Population-Scale Network Science (1.0 FTE, 4 years)

Executive Verdict

How can network measures explain structural similarities and differences between population groups when the data cover an entire country and change over time? This employed PhD at Leiden combines scalable algorithms, network science and computational social science to address that question. It is a strong match for applicants interested in interpretable measurement and careful use of sensitive administrative data. The vacancy identifies job 16940 in Leiden: 38 hours per week, a deadline of 12 November 2026 and a four-year funded plan consisting of one year followed by three more after positive evaluation. Current vacancy status is unconfirmed: Leiden official vacancy.

The project signal is unusually coherent. The vacancy names Prof. Dr. Frank Takes as contact and places the candidate in LIACS, PLANET-NL, and the Leiden Network Science group. Frank Takes's official profile states that his research focuses on algorithms and methods for large-scale real-world social network data, with domains including computational social science, financial systems, economics, infrastructure networks and science studies: Frank Takes profile. PLANET-NL and the Leiden Network Science group both match the vacancy's focus on population-scale, longitudinal, administrative-register-based social network analysis: PLANET-NL, Leiden Network Science.

The main caution is that the project is not just “data science with social data.” It is about developing interpretable, scalable network-science measures for longitudinal population-scale data, including structural equivalence or inequivalence across population subgroups and potentially cross-country comparisons. The best-fitting applicant must show algorithmic maturity, network-science intuition, careful handling of sensitive population data, strong academic writing, and the ability to collaborate across computer science and social science. The official application rule about AI is also narrow but important: application letters written solely with AI, reflected in hollow generic language, will not be considered. The application package needs a short, self-written, position-specific cover letter of at most one page.

School, Programme, and Location Signal

Leiden University offers a solid institutional signal. Its official PhD page describes a PhD track as original scientific research under professor supervision, taking on average four years, and notes that many PhD candidates are employed by the university: Leiden PhD programmes. For this position, the more relevant unit is LIACS. Leiden's official Computer Science and AI page describes LIACS as a centre of excellence for multidisciplinary computer science and AI research and education, with themes across algorithms, AI, data science and applications: LIACS / Computer Science and AI.

The doctorate is article-based. The researcher will collaborate within both PLANET-NL and Leiden Network Science, co-supervise BSc/MSc students, undertake small teaching-assistance duties and potentially minor scientific-organisation tasks. These are part of the advertised work, not a fully teaching-free contract. Formal course requirements, progression milestones and individual supervision arrangements are not specified by the general programme description.

Ranking evidence is positive but should not be over-weighted. QS lists Leiden University at =119 in the 2027 World University Rankings, while Times Higher Education places Leiden among high-performing Dutch universities in its 2026 Netherlands table: QS Leiden profile, THE Netherlands ranking context. The stronger project-level signal comes from the research group, supervisor and publication record rather than generic university rank.

The location signal is practical. Leiden is close to The Hague, Amsterdam, Delft and other Dutch research/technology clusters, while the project is anchored in Leiden's Bio Science Park setting through the Faculty of Science. Living costs may be lower than in central Amsterdam, but this is not a verified personal cost comparison and housing remains material. National guidance estimates Dutch student expenses at roughly €1,000–€1,500 per month, with rent often the largest item; Leiden's international staff housing system explicitly includes PhD candidates for long-stay housing support, but housing should still be treated as a planning risk rather than a guarantee: Study in NL cost guidance, Leiden international staff housing.

Supervisor and Lab Assessment

Supervisor Research Directions

Frank Takes is a strong match for the vacancy. His official profile verifies his professorial role, email, director-of-education role in computer science, and research focus on algorithms and methods for large-scale social network data. It also identifies him as head of the Leiden Network Science research group, academic co-director of PLANET-NL, co-organizer of LCN2, and board member/former chair of the Dutch Network Science Society: Frank Takes profile.

The group context is also strong. The Leiden Network Science group says it works on methods and algorithms for real-world social networks, aiming to understand dynamic complex networks across domains including online social networks, population-scale networks, collaboration networks, infrastructure networks and economic networks: Leiden Network Science. PLANET-NL adds infrastructure language: tools, software, data, methods and tailored computational infrastructure for population-level social-network analysis, with explicit policy and community goals: PLANET-NL.

Recent Important Publications

The recent publication trail is highly relevant. The 2023 Scientific Reports article “The anatomy of a population-scale social network” analyzes the Dutch population through administrative-register-derived family, household, school, work and neighbor layers. It introduces the need for measurement tools tailored to a node-aligned multilayer population-scale network and reports analyses of degree, closure and distance: Scientific Reports 2023.

A 2024 Social Networks paper, listed by PLANET-NL, studies socio-economic segregation in a population-scale social network and is close to the vacancy's aim of interpretable subgroup comparison: PLANET-NL publications, Social Networks article page. Three linked 2026 manuscripts illustrate the group’s direction; their cited versions are on arXiv, and those links do not establish later peer-reviewed publication. “Link Fraction Mixed Membership Reveals Community Diversity in Aggregated Social Networks” introduces a mixed-membership method consistent under aggregation and applies it to a Netherlands population-scale social network; this is directly relevant to aggregated and interpretable subgroup comparison: LFMM 2026. “Fragmentation of a longitudinal population-scale social network” analyzes twelve annual snapshots from 2010–2021 and finds decreasing structural social cohesion in the Netherlands: Fragmentation 2026. “MARS: A framework for modelling register-based social networks” introduces a modelling framework for register-based networks and is relevant to methodological extensions beyond descriptive analysis: MARS 2026.

The publication strength is clear: the group already has empirical data infrastructure, methodological questions, privacy and aggregation problems, and policy-relevant applications. The risk is also clear: the candidate will need to contribute something methodologically new rather than only apply existing metrics.

Lab, Resources, and Collaboration Environment

The vacancy embeds the candidate in LIACS, PLANET-NL and Leiden Network Science, with further interaction through the Dutch Network Science Society and national network-science events: Dutch Network Science Society. This matters because population-scale network science is hard to do in isolation. It requires access to sensitive data, scalable computation, privacy constraints, interpretability, and social-science framing. The vacancy says the Dutch population data comes from Statistics Netherlands. That is a strong data signal, but the exact data-access procedure, secure environment, allowed outputs, and cross-country data availability remain interview questions. Statistics Netherlands is the official national statistical office, but access to microdata or linked administrative networks is not automatically implied by a vacancy description: Statistics Netherlands.

Research Direction and Field Outlook

Field Momentum

The project sits in a timely research area: population-scale network analysis using administrative data, temporal network methods, interpretable subgroup comparison, privacy-preserving social network publication, and computational social science. The field is moving beyond sampled online traces toward register-based or institutionally curated population networks. The Scientific Reports 2023 paper explains why online or mobile communication samples can be non-representative and why official-register-derived networks provide a different opportunity for understanding social structures: Scientific Reports 2023.

The vacancy's structural equivalence or inequivalence focus is credible because recent group work already explores aggregation, social cohesion, segregation, mobility and social capital. Related privacy work also matters: large-scale social-network data is sensitive, and recent Takes-linked work on k-anonymity and network anonymization highlights that anonymity measures, attacker scenarios and utility trade-offs need careful methodological design: Anonymization problem, k-anonymity measures.

Potential Research Angles

Plausible research angles include: structural-equivalence measures for aggregated or subgroup-level population networks; scalable algorithms for temporal subgroup comparison; null models or benchmarks for register-based social networks; cross-country comparability when data definitions differ; visual and statistical methods for interpretable subgroup differences; privacy-preserving release or analysis of subgroup-level network metrics; and fairness-aware or policy-facing interpretations of social opportunity structures. The strongest angle would combine algorithmic novelty with a clear social-science question, rather than treating the population data as only a large graph.

Career and Market Outlook

Career signal is broad. The immediate academic market includes network science, computational social science, data science, privacy-preserving analytics and public-sector data infrastructure. Outside academia, skills in scalable graph analytics, Python/R/C++, longitudinal data, privacy-aware analytics and interpretable modelling fit roles in public agencies, research institutes, consultancies, tech, finance, urban analytics and policy analytics. Dutch labour-market sources show that ICT demand remains material even as the market has cooled from earlier peaks: UWV reports ICT vacancies and specialist knowledge needs, while the European Commission's 2026 Digital Decade report says ICT specialist shortages slow Dutch digital transformation: UWV ICT occupations, European Commission Digital Decade Netherlands 2026. The BLS operations-research outlook is not Netherlands-specific, but it is a useful comparator for analytical modelling careers: BLS operations research analysts.

Funding, Application, and Living Cost

The official vacancy verifies a fully funded four-year PhD employment model. The contract starts with one year; after a positive evaluation of thesis progress, personal capabilities and compatibility, it extends by three years. Salary is €3,204–€4,051 gross/month for a full-time 38-hour appointment in pay scale P, plus 8% holiday allowance, 8.3% year-end bonus, ABP pension, public-transport commuting reimbursement, at least 29 leave days, possible flexible working patterns, hybrid work within the Netherlands where compatible, home-working allowance and laptop. The advert gives a 40-hour option that accrues an extra 96 leave hours, or the option to exchange 96 leave hours for a 36-hour week; these are alternatives subject to the employment arrangements, not benefits to add together: Leiden official vacancy.

The stated deadline is 12 November 2026 at 22:59 UTC. No corresponding Dutch local closing time is confirmed, and current application status has not been verified. Applicants should not wait for the final hour.

Application materials are unusually explicit and should not be misread. The vacancy requires: a short cover letter of at most one page detailing motivation; an academic CV; a recent writing example, typically the MSc thesis; and names plus contact details of at least two people who can be contacted for reference. It explicitly says those people should have agreed to be contacted and that there is no need to send letters. The language requirement is good English demonstrated through a near-fluent in-person conversation, not an unspecified waiver based on working in an international team. The AI-only rule is also specific: application letters written solely using AI, reflected in hollow sentences, will not be considered. The practical implication is to write the cover letter personally, connect it to the research problem and supervisor/group evidence, and avoid generic AI-styled prose.

For Chinese and other non-EU applicants, the vacancy does not state a nationality exclusion. It says candidates who require a visa may have a later start date depending on the procedure for their country of origin. Dutch IND guidance has a researcher residence-permit route covering doctoral candidates, and Leiden's knowledge-security framework shows that international staff appointments can be assessed under objective and proportional governance. This is a boundary to prepare for, not a blanket prohibition: IND researcher residence permit, Leiden knowledge-security framework.

Applicant Fit and Preparation

Best suited for

The project is best suited for applicants with an MSc in computer science, mathematics, physics, computational social science, data science or a closely related field, with demonstrated affinity for network science, complex networks or social network analysis. Strong candidates should show programming ability in Python, or otherwise R/C++, plus academic writing through a thesis or paper. The vacancy also welcomes excellent candidates who are not yet past the MSc phase if graduation in early spring 2027 is clearly in sight.

May need caution if

Applicants should be cautious if they are interested in generic machine learning but have little graph/network background; if they cannot explain social-network concepts such as structural equivalence, community structure, aggregation, longitudinal change or multilayer networks; if they have no evidence of academic writing; or if they are uncomfortable with privacy, governance and secure-data constraints. The project likely requires less “black-box prediction” and more interpretable, mathematically careful network measurement.

Skills and evidence to strengthen

Useful evidence includes: graph algorithms or network-science coursework; a thesis involving networks, large-scale data, statistical modelling or computational social science; reproducible Python/R/C++ code; familiarity with sparse matrices or scalable graph processing; a writing sample with clear argumentation; and a cover letter that explains why subgroup comparison and structural equivalence in population-scale networks matter. Applicants should also prepare a concise explanation of how they would handle sensitive population data and avoid overclaiming from administrative-network measures.

Risks and Questions to Verify

High-priority questions

1. What exact population-network data will the PhD candidate be allowed to access, and through what secure environment?
2. What is the expected primary methodological contribution: structural-equivalence metrics, scalable algorithms, longitudinal modelling, cross-country comparison, privacy methods, or a combination?
3. Which cross-country datasets or collaborators are realistically available for the comparative part?
4. How will the one-year evaluation judge thesis progress, compatibility and personal capabilities?
5. How will Leiden's knowledge-security and privacy governance apply to an international candidate working with sensitive population-scale data?

Medium-priority questions

1. What computation stack is available for large longitudinal networks?
2. How much teaching assistance and BSc/MSc co-supervision is expected each year?
3. Who besides Frank Takes will be involved in day-to-day supervision within PLANET-NL or Leiden Network Science?
4. What outputs are expected: theory/method papers, empirical computational social-science papers, software, or policy-facing analyses?
5. What housing support is realistic for a January 2027 start, especially if visa processing shifts the start date?

Evidence and Sources

The cited 2026 arXiv papers are preprint versions; no claim is made here about a later peer-reviewed version.

Key sources

• Leiden official vacancy
• Additional AcademicTransfer listing
• Frank Takes profile
• LIACS
• PLANET-NL publications
• Scientific Reports 2023
• LFMM 2026
• Fragmentation 2026
• MARS 2026
• Study in NL cost guidance
• Leiden knowledge-security framework