Teacher attrition research faces a persistent methodological limitation: existing administrative databases record employment transitions without capturing the processes leading to departure. National databases (e.g. Central Bureau of Statistics Onderwijspersoneeltab and DUO Open Onderwijsdata) document that teachers leave but not why, when warning signs emerged, or what support has been provided before teachers’ departure. This creates a “black box” around attrition that constrains both scientific understanding of attrition dynamics and the development of timely, evidence-based retention interventions. 

While national and regional studies have explored teacher turnover and retention in the Dutch context, they typically focus on specific predictors or stated motives of attrition rather than integrated, longitudinal tracking from career entry to exit. Existing research shows that teachers’ decisions to leave can be associated with individual (e.g., attitudes, perceived norms, autonomy, and competence) and organizational factors (e.g., work pressure, leadership quality, and limited professional development and induction support) (den Brok et al., 2017; Helms-Lorenz et al., 2013; van Casteren et al., 2023). However, like administrative databases, these studies lack the infrastructure to link baseline characteristics and support provision to subsequent employment outcomes over time. The present project StartSterk addresses this gap by developing and validating an integrated data infrastructure that enables prospective tracking of beginning teachers (BTs) from career entry through potential departure.

This project establishes a regional register system across approximately 30 secondary school boards in the Northern Netherlands, spanning the provinces of Friesland, Drenthe, Flevoland, Overijssel, and Groningen, organised into two educational regions (Friesland Noordelijk Flevoland and Noorderwijzer).  The infrastructure addresses four critical limitations of conventional workforce data: 1) Unlike retrospective exit studies relying on post-hoc recall, the system implements event-triggered data collection. When HR systems record a resignation, an emergency protocol immediately initiates an Exit Survey before the teacher’s real departure; 2) Tri-annual measurement waves (October, January, June) track intake timing, mid-year mobility, and year-end patterns; 3) A linkage structure with unique teacher identifiers enables connection to a profiling instrument (Career-entry Profile Booklet “Ik in Balans”); 4) Standardised categorisation creates comparable data across diverse school administrative systems.

This study presents the infrastructure design and reports baseline measurement (T0) findings, demonstrating system functionality. The theoretical framework integrates three perspectives. Research infrastructure theory (Edwards et al., 2013) illuminates how data systems require sustained coordination among multiple stakeholders. Prospective cohort methodology (Checkoway et al., 2007) provides design principles for tracking populations while minimising attrition-related bias. Induction effectiveness research (Helms-Lorenz et al., 2016, 2020; Ingersoll & Strong, 2011) establishes the theoretical rationale for connecting entry characteristics, support provision, and retention outcomes.

We address two research questions:

RQ1: What architectural features are required for a data infrastructure to effectively capture and understand the processes of beginning teacher retention and attrition from career entry to exit? (Design)

RQ2: What does the baseline measurement reveal about beginning teacher population characteristics, and what data quality metrics indicate system readiness for longitudinal tracking? (Validation)

The study contributes a transferable infrastructure model addressing recognized limitations in teacher workforce research. While serving the Northern Netherlands, the architectural principles (i.e., event-triggered exit documentation, multi-wave tracking, linkage-ready identifiers, and cross-system standardisation) provide a template for European regions seeking to move beyond retrospective, aggregate-level administrative statistics toward prospective, process-oriented causal retention research.

This study employs design-based research methodology (McKenney & Reeves, 2019) to develop the register infrastructure, combined with descriptive analysis of baseline data to demonstrate system functionality. Infrastructure design (RQ1): The data architecture comprises three integrated components. The core register captures standardized Human Resource data using Excel templates with dropdown menus ensuring consistent categorization and to reduce extensive data cleaning practices after the data collection. The templates register demographic information, qualification pathway and status, subject specialization (DUO-standardized codes), employment conditions (contract type, FTE, appointment dates) and induction/support arrangement. The exit documentation protocol is triggered upon resignation notification: regional coordinators receive immediate alerts and initiate an exit survey capturing departure reasons, destination, and support experiences before the teacher’s final working day. The linkage architecture assigns persistent teacher identifiers enabling connection to the Career-entry Profile Booklet (currently in pilot). This co-designed instrument captures BTs’ personal and professional backgrounds, work context, job demands and resources, competencies, wellbeing, and career intentions to determine profiles. The profiles are translated to individualized development and support pathways. Data collection process: School board HR departments enter data directly into secure UniShare folders following a structured six-week protocol per measurement wave. Regional coordinators (FROSK/Friesland, Noordermarke/Drenthe, and soon Onderwijsregio Groningen) provide ongoing support and conduct quality checks. The research team performs identity matching, duplicate detection, and cross-school validation before merging into the unified regional dataset. Tri-annual measurements (October, January, June) captures seasonal intake, mid-year mobility, contract transitions, and cumulative attrition/retention (inflow, throughflow, outflow). The event-triggered exit protocol operates continuously between scheduled waves. BTs identified through the register are approached to participate in profiling using the Career-entry Profile Booklet, creating prospective data on which entry characteristics predict/are associated with the subsequent retention. Baseline validation (RQ2): The T0 dataset (October 2025) includes BTs (0-5 years experience) across 45 secondary schools in 31 cities spanning five geographical provinces and two educational regions. Descriptive statistics document population characteristics. Data quality metrics assess the completeness rates, cross-field consistency, duplicate identification, and standardization compliance. This methodology enables both formative evaluation of the infrastructure itself and substantive baseline analysis demonstrating the system’s capacity to generate actionable workforce intelligence for evidence-based induction support.

RQ1: Four potential features operational: 1) multi-source triangulation: HR records validated against self-reports and school-based mentors’ knowledge; 2) temporal granularity: tri-annual waves capturing intake timing and mobility (in/through/out-flow) patterns; 3) event-triggered protocols: immediate exit documentation preventing recall bias; and 4) privacy-preserving linkage: enabling longitudinal tracking and connection to profiling data. RQ2: The T0 dataset includes 319 BTs across 45 schools in 31 cities. The data reflect actual BT intake across regions, as register data are expected to capture all BTs employed by participating school boards. The top 10 cities account for 72.4% of all BTs. The population displays substantial age diversity (M=31.8, range=20—58), reflecting multiple entry pathways. 57% are under 30, yet 31% are 35 or older, indicating significant lateral entry. A majority (57.9%) are not yet formally qualified, with 57.2% combining employment with ongoing training and 6.5% enrolled as lateral entry trainees. These BTs span heterogeneous career stages (<1: 31.4% and >5 years: 10.4%). Second-degree qualifications predominate (70.3%), which aligns with observed placement patterns. Most teachers are concentrated in lower secondary education (61.6%) and VMBO-oriented tracks (53.5%). However, over one-quarter of the cohort (27.4%) teach across multiple tracks (e.g., VMBO and HAVO) or across educational levels (e.g., lower and upper HAVO), and 6.0% teach at both qualification levels. Several cases of misalignment between formal qualification and teaching assignment were identified, highlighting structural boundary-crossing in early career placements. Employment conditions reflect high job insecurity: 77.4% hold temporary contracts and part-time work is normative (mean FTE=0.67). Induction documentation is incomplete. Only 47.7% have recorded mentoring start dates, with planned duration varying substantially (5–60 months), indicating considerable school-level variation in induction practices. Seven teachers show early contract termination dates, signaling potential attrition cases. These findings establish baseline metrics for prospective retention tracking while revealing structural conditions warranting policy attention.

Checkoway, H., Pearce, N., & Kriebel, D. (2007). Selecting appropriate study designs to address specific research questions in occupational epidemiology. Occupational and Environmental Medicine, 64(9), 633-638. https://doi.org/10.1136/oem.2006.029967 den Brok, P., Wubbels, T., & van Tartwijk, J. (2017). Exploring beginning teachers’ attrition in the Netherlands. Teachers and Teaching, 23(8), 881–895. https://doi.org/10.1080/13540602.2017.1360859 Helms-Lorenz, M., Slof, B., & van de Grift, W. (2013). First year effects of induction arrangements on beginning teachers’ psychological processes. European Journal of Psychology of Education, 28(4), 1265–1287. https://doi.org/10.1007/s10212-012-0165-y Helms-Lorenz, M., van de Grift, W., & Maulana, R. (2016). Longitudinal effects of induction on teaching skills and attrition rates of beginning teachers. School Effectiveness and School Improvement, 27(2), 178-204. https://doi.org/10.1080/09243453.2015.1035731 Helms-Lorenz, M., van der Pers, M., Moorer, P., Harmsen, R., & Verkade, A. (2020). Inductie in het noorden: Eindrapportage 2014–2019. University of Groningen. https://www.platformsamenopleiden.nl/wp-content/uploads/2020/09/Begeleiding-Startende-Leraren-Eindrapportage-2014-2019-3.pdf Ingersoll, R. M., & Strong, M. (2011). The impact of induction and mentoring programs for beginning teachers: A critical review of the research. Review of Educational Research, 81(2), 201-233. https://doi.org/10.3102/0034654311403323 McKenney, S., & Reeves, T. C. (2019). Conducting educational design research (2nd ed.). Routledge. https://doi.org/10.4324/9781315105642 van Casteren, W., Lodewick, J., Lommertzen, J., Luyten, E., & van Mensvoort, C. (2023). Vertrekredenen leraren en docenten in het po, vo en mbo (ResearchNed report, commissioned by the Ministry of OCW). Retrieved from https://www.aanpaklerarentekort.nl/documenten/2023/12/14/vertrekredenen-leraren-en-docenten-in-het-po-vo-en-mbo---researchned