Work accident statistics of insured persons in Türkiye, 2016–2025 (SGK): tidy category-level panel, results and code
Description
Data and code accompanying the article "Descriptive analysis of insured persons with work accidents in Türkiye, 2016–2025, using Social Security Institution data" [article details to be added after publication]. The dataset converts the nine work accident classification tables of the Turkish Social Security Institution (SGK) Statistical Yearbooks for 2016–2025 (insured persons under Article 4/1-a of Law No. 5510) into one tidy panel: injury-causing incident, general activity, specific activity, deviation, material agent, working environment, workstation, time of accident and workplace size, by sex and work-incapacity level, with death counts where published. It also contains the SGK source workbooks, a validation of the analysis totals against SGK's own totals (90 table-years), all result tables and figures of the article (case-fatality ratios, ordinal work-incapacity score, ranking stability, sensitivity analyses, registered work accident frequency per 1,000 insured persons by workplace size, keyword theme assignments) and the Python code that reproduces them. The indicator is the number of insured persons with a work accident in a year, not a count of independent accident events. The data are aggregated at category level and are not record-level microdata; categories of different classification dimensions cannot be linked to individual accident cases. Original data: Social Security Institution (SGK).
Files
Steps to reproduce
Download all files and keep the folder structure (README.md describes every file). Install Python 3.12 and the packages listed in code/requirements.txt. From the dataset's root folder run: python code/run_all.py The scripts read the SGK source workbooks in raw_data_SGK/ (Tables 3.1.18–3.1.26 for 2016–2025 and SGK Table 1.9 for the workplace-size denominator), harmonise the nine variables across years by category code and description, exclude parent-category rows and build data/panel_long.csv. They validate the totals against the total row printed by SGK beneath each table (data/validation_totals.csv), calculate the case-fatality ratios, ordinal work-incapacity scores and registered work accident frequencies, apply the thresholds and sensitivity analyses, assign the themes, and write results/*.csv, figures_EN/ and Supplementary_Material_EN.xlsx. When reading the CSV files, read the 'code' column as text (leading zeros are meaningful).
Institutions
- Doğuş UniversityIstanbul, Istanbul