Replication code for the analysis of graduate tracking indicators and mandatory postgraduate training in the health professions

This deposit contains the code that reproduces the analysis reported in an associated manuscript The study examines whether graduate tracking systems built on the linkage of education registers to social security records describe adequately the transition to professional practice in health degrees whose licensing requires a prior period of mandatory postgraduate training. It uses the Polish graduate tracking system (Ekonomiczne Losy Absolwentów, ELA), which publishes sixty monthly observations per programme for each graduating cohort, covering eleven cohorts (2014–2024) and nine degrees. What is included ela_16_cohortes_9g.R — rebuilds the cohort-by-month series for the nine degree groups from the source file, taking the classification from the programme profile so that it is identical to that used in the tables. Produces the cohort series, the maximum of each cohort with a censoring flag, and the summary reported in Appendix Table A2. ela_17_figura2_9g.R — produces Figure 2, the cohort verification for medicine. paper_pipeline.R — assembles the display items reported in the paper: the two main-text tables, the appendix tables, the compliance checks against the journal's limits, and a checksum manifest of inputs and outputs. sessionInfo.txt and manifest.csv — the R session under which the results were produced, and MD5 checksums of every input and output file, so that any reconstruction can be verified against the version used here. What is not included, and how to obtain it The source data are not redistributed here. The file graduates-major-data.csv is published by the Polish National Information Processing Institute (OPI PIB) and is available through the researcher access area of https://ela.nauka.gov.pl/. The data are aggregated by programme and cohort; the system does not release individual microdata. Once that file is placed in the input folder, the scripts reproduce every series, table and figure reported in the paper. The derived series are likewise not deposited. They are fully reconstructed by the scripts from the public source, and depositing them would substitute a processed artefact for the provenance that makes the result verifiable. Derived files are available from the authors on reasonable request. Requirements R 4.5.3 or later, with the packages data.table and ggplot2. Paths are set at the top of each script.